Full-Thesis 9

MAINSTREAMING RENEWABLE ENERGY
DEVELOPMENTS INTO TRADITIONAL PLANNING
J.H.K. Kanchana Chathuranga
109204C
Degree of Master of Science in Electrical Engineering
Department of Electrical Engineering
University of Moratuwa
Sri Lanka
February 2014
MAINSTREAMING RENEWABLE ENERGY
DEVELOPMENTS INTO TRADITIONAL PLANNING
109204C
Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science in Electrical Engineering
Sri Lanka
February 2014
DECLARATION
I declare that this is my own work and this dissertation does not incorporate without
acknowledgement any material previously submitted for a Degree or Diploma in any
other University or institute of higher learning and to the best of my knowledge and
belief it does not contain any material previously published or written by another
person except where the acknowledgement is made in the text.
Also, I hereby grant to University of Moratuwa the non-exclusive right to reproduce

and distribute my thesis/dissertation, in whole or in part in print, electronic or other
medium. I retain the right to use this content in whole or part in future works (such as
articles or books).
Signature of the author Date:

J.H.K.K. Chathuranga
The above candidate has carried out research for the Masters Dissertation under my
supervision.
Signature of the supervisor Date:

Eng. W.D.A.S. Wijayapala
Senior Lecturer
Signature of the co-supervisor Date:

Prof. P.D.C. Wijayatunga
Principal Energy Specialist
Asian Development Bank
i
ABSTRACT
Electricity is one of the key driving forces of the economy of a country and
generation of electricity in an optimal way to meet the increasing demand has
become a national priority in the recent years. Due to serious concerns with regard to
energy security, global warming, rising costs and depleting reserves of fossil fuels,
many countries are now actively seeking to mainstream NCRE power generation in
to their generation portfolios as a future energy solution. Since generation planning
plays a major role in a country’s efforts to mainstream NCRE developments, the Sri
Lankan generation planning practices were examined and several methodological
changes and models were proposed to successfully integrate and evaluate NCRE
resources in the present planning approaches.
The CEB generation planning process was reviewed and associated issues
concerning NCRE planning were identified. These issues were first addressed
conceptually and the proposed solutions were subsequently applied to the Sri Lankan
system to assess their applicability. This thesis provides a new insight into the
capacity contribution of NCRE plants and also discusses the constraints to
mainstream adoption of NCRE technologies in Sri Lanka along with the present
policy and regulatory interventions relating to NCRE developments. The use of peak
period capacity factor method was suggested to calculate the capacity credit of
NCRE generation and since the associated risks are not explicitly evaluated in the
present approaches, the Mean Variance Portfolio Theory is proposed to assess the
risks of generation portfolios. Two models were developed to calculate the wind
power output from wind measurement data and to evaluate the portfolio risks of
generation mixes which can be readily used in the present practices. In addition,
methodologies were presented to model a wind power plant in WASP IV and to
evaluate the benefits of modeled NCRE plants.
Keywords:
Non-Conventional Renewable Energy
Long Term Generation Planning
Wien Automatic System Planning Package
Capacity Credit
Mean Variance Portfolio Theory
ii
ACKNOWLEDGEMENTS
First and foremost, I would like to express my sincere gratitude to my supervisors

Eng. W.D.A.S. Wijayapala and Prof. Priyantha Wijayatunga, for their continuous
guidance, constructive feedback and support extended throughout this research.
Despite their busy schedule, they always remained accessible and their guidance and
advice, expertise and insights were by all means truly invaluable.
I would also like to thank Prof. J.R. Lucas, Prof. H.Y.R. Perera, Prof. J.P.
Karunadasa, Dr. M.P. Dias and Dr. Asanka Rodrigo for their valuable feedback
during the progress reviews.
It’s with immense pleasure that I thank Mr. M.B.S. Samarasekara, Chief Engineer
(Generation Planning) and Mr. T.L.B. Attanayake, Electrical Engineer (Generation
Planning) of the Ceylon Electricity Board for their help with the WASP model, Mr.
S.C. Diddeniya, CEO of Sri Lanka Energies (Pvt) Ltd. for providing the required
wind measurement data of Mannar and Mr. Vladimir Koritarov of Argonne National
Laboratory for his guidance in modeling and evaluating NCRE plants in the WASP
model. Without their support, this research would not have been successful.
I am grateful to my parents for all the sacrifices they have made along the way and
for all the support given to me during various endeavors in my life. Finally, I would
like to sincerely thank my wife Rushani for being supportive and understanding, and
for all her love and encouragement throughout this research.

February 2014
iii
TABLE OF CONTENTS
Declaration i
Abstract ii
Acknowledgements iii
Table of Contents iv
List of Figures vii
List of Tables viii
List of Abbreviations ix
List of Appendices x
Chapter 1. Introduction 1
1.1 Objectives of the Research 4
1.2 Methodology 4
1.3 Outline of the Thesis 5
Chapter 2. Mainstreaming NCRE Developments in Sri Lanka: 7

Constraints and Strategies to Overcome the Barriers
2.1 Constraints to Mainstream Adoption of NCRE Technologies in 7
Sri Lanka
2.2 Policy and Regulatory Interventions for NCRE Developments 9
2.2.1 Policy and regulatory interventions in Sri Lanka 9
2.2.2 Policy lessons from around the world 10
Chapter 3. Review of the CEB Generation Planning Process 15

3.1 Overview of the CEB Generation Planning Methodology 16
3.2 Wien Automatic System Planning Package (WASP) 17
3.2.1 Objective function of the WASP IV package 18
3.2.2 Principal capabilities and limitations of WASP IV 19
3.3 Other Generation Planning Tools 20
3.3.1 Electric Generation Expansion Analysis System 20
3.3.2 PLEXOS for Power Systems 21
3.4 2011-2025 Generation Expansion Plan of the CEB (Base Case 22
Plan)
3.5 Identified Issues in the Present Generation Planning Process 23
3.6 Proposed Concepts to Modify the Existing Planning Process 24
Chapter 4. Capacity Credit of Renewable Generation and Wind 26

Power Output Model
4.1 Power System Planning and System Adequacy 26
iv
4.2 Methods of Calculating the Capacity Credit of NCRE 26
Generators
4.2.1 Effective Load Carrying Capability 28
4.2.2 Approximation methods 29
4.2.3 Proposed method to calculate the Capacity Credit of 31
NCRE generation in Sri Lanka
4.3 Wind Power Output Model 32
4.3.1 Output of the modeled wind park 34
4.3.2 Analysis of the wind power output model 36
Chapter 5. Modeling and Evaluating Wind Power Plants in WASP IV 38

Package
5.1 Methodology of Modeling Wind Power in WASP IV 38
5.2 Results of the WASP Simulations 40
5.3 Evaluation of the Modeled Wind Park Using WASP IV 42
5.3.1 Energy value of the wind park 44
5.3.2 Capacity value of the wind park 45
5.3.3 Emission benefits of the wind park 48
Chapter 6. Evaluation of the Sri Lankan Generation Mixes Using 50

Mean-Variance Portfolio Theory
6.1 Least Cost Vs Portfolio Based Approaches in Generation 50
Planning
6.2 Mean Variance Portfolio Theory Basics 52
6.3 Application of MVPT to Electricity Generation Assets 56
6.4 Data Required for the Application of MVPT to Generation 59
Portfolios
6.5 Evaluation of the Sri Lankan Generation Mixes of 2012 and 60
2025
6.5.1 Results and Discussion 63
6.5.1.1 Scenario one (With CO2 costs) 64
6.5.1.2 Scenario two (Without CO2 costs) 67
Chapter 7. Summary and Conclusion 70

7.1 Summary and Key Findings 70
7.2 Recommendations and Contributions of the Thesis 72
7.3 Scope for Future Work 73
7.4 Conclusion 74
Appendices 75
Appendix A Power Curve Values of the Reference Turbine 76
v
Appendix B Curve Fitting of Reference Turbine Power Curve 78
Appendix C Matlab Code of Wind Power Output Model 79
Appendix D Base Load Demand Forecast of the CEB up to Year 2030 86
Appendix E Details of the Candidate Thermal Plants used in the WASP 87
Runs
Appendix F Standard Deviation Values of Capital, Fuel, O&M and CO2 88
Costs
Appendix G Correlation Coefficients of Fuel, O&M and CO2 Costs 89
Appendix H Matlab Code of MVPT Analysis Model 90
References 104
vi
LIST OF FIGURES
Figure 1.1 NCRE Project development from 1999 to 2012 2

Figure 4.1 Probability distribution of the plant output and capacity blocks 30
Figure 4.2 Typical system load profile of Sri Lanka 31
Figure 4.3 Flowchart of the wind power output model 34
Figure 4.4 Monthly Capacity Credit of the wind park 35
Figure 4.5 Monthly energy production of the wind park 35
Figure 6.1 Expected return and risk of two-asset portfolio 55
Figure 6.2 Portfolio effect for a two asset generation portfolio 58
Figure 6.3 Costs and risks of generating technologies (with CO2 costs) 62
Figure 6.4 Costs and risks of generating technologies (without CO2 costs) 62
Figure 6.5 Efficient frontier for the CEB Generation mix - Scenario 1 64
Figure 6.6 CEB generation mixes of 2012 and 2025 against EF - Scenario 1 65
Figure 6.7 Efficient frontier for the CEB Generation mix - Scenario 2 67
Figure 6.8 CEB generation mixes of 2012 and 2025 against EF - Scenario 2 67
vii
LIST OF TABLES
Table 1.1 Present status of the NCRE sector in Sri Lanka (as at 31/12/2012) 2
Table 3.1 Principal capabilities and limitations of WASP IV 19
Table 3.2 2011-2025 Generation expansion plan of the CEB (Updated base 22
case)
Table 5.1 Estimated output of the proposed wind park 40
Table 5.2 New capacity additions of the two expansion plans 41
Table 5.3 Capacity displacements due to the addition of wind park 43
Table 6.1 Expected returns and standard deviation of assets A and B 54
Table 6.2 MVPT analysis results of two-asset portfolio 55
Table 6.3 Levelised generating costs (US$/MWh) 59
Table 6.4 Target generation mixes of 2012 and 2025 61
Table 6.5 Lower and upper bounds for the alternative technologies 61
Table 6.6 Technology shares of optimal portfolios – Scenario 1 65
Table 6.7 Technology shares of optimal portfolios – Scenario 2 68
viii
LIST OF ABBREVIATIONS
AIC Average Incremental Cost

BOI Board of Investments
CCGT Combined Cycle Gas Turbine
CEB Ceylon Electricity Board
EF Efficient Frontier
EGEAS Electric Generation Expansion Analysis System
ELCC Effective Load Carrying Capability
EMCAS Electricity Market Complex Adaptive System
ENS Energy-Not-Served
EPRI Electric Power Research Institute
EU European Union
FOR Forced Outage Rate
IEA International Energy Agency
LNG Liquefied Natural Gas
LOI Letter of Intent
LOLE Loss of Load Expectation
LOLP Loss of Load Probability
MIT Massachusetts Institute of Technology
MVPT Mean Variance Portfolio Theory
NCRE Non-Conventional Renewable Energy
NREL National Renewable Energy Laboratory
O&M Operation and Maintenance
OECD Organization for Economic Cooperation and Development
ORNL Oak Ridge National Laboratory
PUCSL Public Utilities Commission of Sri Lanka
PV Present Value
ROR Run-of-River
RPS Renewable Portfolio Standard
SLSEA Sri Lanka Sustainable Energy Authority
SPPA Standardized Power Purchase Agreements
SYSIM System SIMulation
TVA Tennessee Valley Authority
WASP Wien Automatic System Planning
ix
LIST OF APPENDICES
Appendix A Power Curve Values of the Reference Turbine 76

Appendix B Curve Fitting of Reference Turbine Power Curve 78
Appendix C Matlab Code of Wind Power Output Model 79
Appendix D Base Load Demand Forecast of CEB up to Year 2030 86
Appendix E Details of the Candidate Thermal Plants used in WASP Runs 87
Appendix F Standard Deviation Values of Capital, Fuel, O&M and CO2 88
Costs
Appendix G Correlation Coefficients of Fuel, O&M and CO2 Costs 89
Appendix H Matlab Code of MVPT Analysis Model 90
x
Chapter 1
INTRODUCTION
Governments, electricity utilities and end users of electricity are becoming
increasingly concerned about meeting the growing electricity demand. Electricity has
become a key driving force of the economy of a country and therefore, generation of
electricity in an optimal way to meet the increasing demand has become a key
challenge in recent years. Increasing environmental concerns and energy security
issues as well as diminishing fossil fuel reserves and their rising costs have made it
necessary to look towards renewable energy sources as a highly favorable solution
for meeting future energy requirements. As a result, there has been a tremendous
amount of interest in a lot of countries on Non-Conventional Renewable Energy
(NCRE) sources as a means for power generation since the past decade due to many
advantages such as their non-depletable and environmental friendly nature, local
availability and cost reductions with technology advancements.
In Sri Lanka, 54% of electricity is currently generated by burning fossil fuels and
according to the Ceylon Electricity Board’s (CEB) long term generation expansion
plan of 2010, 88% of new capacity additions will be from coal power plants and this
will increase the present thermal share of 54% to 74% by 2025. This situation may
create serious energy security concerns since global economic and political changes
could dictate the cost and availability of these energy supplies in the long term.
Therefore, mainstreaming renewable energy sources in the country is of paramount
importance so as to provide electricity at a known cost in contrast to fossil fuels.
According to the National Energy Policy of Sri Lanka, 10% of electricity is expected
to be generated through NCRE technologies by 2015. The policy emphasizes the
importance of maintaining the adequacy of the system and the continuity of supply at
the lowest possible cost while maximizing the country’s energy security by
diversifying the generation mix [1]. As shown in Table 1.1, the grid connected
installed capacity of NCRE plants was 320.6MW by the end of 2012 and these plants
had contributed 6.17% of the total electricity generation in 2012 [2].
1
Table 1.1: Present status of the NCRE sector in Sri Lanka (as at 31/12/2012)
No Description Technology Type No of Capacity
Projects (MW)
1 Commissioned Mini Hydro Power 104 234.1
Projects Biomass - Agricultural &
2 11.0
Industrial Waste Power
Biomass - Dendro Power 1 0.5
Solar Power 4 1.38
Wind Power 9 73.65
Total - Commissioned 123 320.63
2 Standardized Power Mini Hydro Power 76 171.61
Purchase Agreements
Wind Power 4 21.3
(SPPA) Signed
Biomass - Agricultural &
Projects 2 4.0
Industrial Waste Power
Biomass - Dendro Power 11 61.77
Biomass - Municipal Solid
1 10.0
Waste
Total – SPPA Signed 94 268.68
Source : CEB Website (www.ceb.lk)
350
320
300
250 227
212
200 181
161 Installed
Capacity
150
(MW)
112 119
100 88
73
50 39
23.6 31.2
8.6 11.6
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Figure 1.1: NCRE Project development from 1999 to 2012
2
Generally, long term generation planners are responsible for forecasting the future
demand and the capacity additions to meet the forecasted demand at an accepted
level of system reliability. Not meeting the demand will lead to a considerable loss to
the economy and as per the CEB’s long term generation expansion plan of 2010, this
value has been estimated as US$ 1.2 per kWh of energy not served [3]. However,
finding an optimal generation expansion plan is difficult due to the uncertainty
associated with the input data such as forecasts of electricity demand, economic and
technical characteristics of new evolving generating technologies in power
generation, frequent fluctuations of fuel costs, construction lead times, and
government regulations. This becomes an even more complex problem when NCRE
technologies are introduced into the generation planning process because the nature
of NCRE plants are regarded and characterized as intermittent and fluctuating.
Furthermore, these plants are considered to be nondispatchable, which is a significant
downside from the utilities’ point of view. The uncertainty and variability of these
technologies makes it challenging to include these plants in capacity expansion
models such as Wien Automatic System Planning Package (WASP) which is
currently being used by the CEB’s generation planning branch.
However, with the increasing levels of penetration and interest in NCRE

technologies, it is important to include these plants in the planning process to
evaluate the long term benefits, particularly the capacity benefits which are often
ignored. In the planning process, most utilities give these plants an energy credit but
not a capacity credit because these plants are not considered to be contributing to the
system reliability. Yet, many studies have concluded that NCRE plants do make a
considerable amount of contribution to system reliability and therefore, should not be
neglected [4].
The aim of traditional generation expansion planning has been to provide an

adequate supply of electricity at a minimum cost and the CEB’s generation planning
branch also uses a least-cost based approach for generation planning. But, the use of
traditional least-cost methodologies which are generally biased in favor of fossil fuel
technologies may not deliver optimal results due to many uncertainties associated
with modern power systems as mentioned above [5].
3
Therefore, the research presented in this thesis focuses on developing concepts to
mainstream NCRE technologies into the present long term generation planning
process in Sri Lanka and the application of the proposed concepts to the Sri Lankan
system. According to a number of studies conducted by the CEB and the National
Renewable Energy Laboratory (NREL) of USA, Sri Lanka has a very good wind
potential and wind is expected to play a key role in the coming years in achieving the
renewable energy targets set out in the national energy policy [6 - 8]. Mini Hydro
and Biomass are the other two NCRE resources that have a good potential in Sri
Lanka and both the technologies are adequately modeled in the present planning
approaches of the CEB [9]. Therefore, a focus on wind is chosen in the application of
proposed concepts of this research.
1.1 Objectives of the Research

The objectives of this research are to:
1. Propose a method to determine the capacity contribution of intermittent
generation and develop a model to estimate the monthly capacity credit and
energy production from a hypothetical wind power plant using wind
measurement data
2. Propose a methodology to integrate the monthly wind power generation data
in the present planning process and evaluation of energy, capacity and
emission credits of the modeled plant
3. Risk evaluation of the Sri Lankan generation mix using the Mean Variance
Portfolio Theory (MVPT)
1.2 Methodology
The research was planned in a few steps to achieve the identified research objectives
and the work started with an extensive literature survey on the approaches to
mainstream NCRE developments, constraints to mainstream adoption of NCRE
resources, policy lessons from around the world and previous efforts taken to include
NCRE resources in long term generation planning.
Secondly, the existing generation planning process in Sri Lanka was examined with
special focus on how NCRE is treated with the aim of identifying the issues that lies
4
with present approaches. The identified issues were then addressed conceptually and
ways to modify the existing planning process were examined to accommodate the
proposed concepts.
After the conceptual design, a wind power output model was developed using Matlab
to estimate the capacity contribution and energy output of a wind power plant. Two
methods were identified to incorporate the calculated wind power generation into the
present planning approaches and actual wind measurement data from Mannar was
then used to calculate the wind generation of a hypothetical wind park. The
calculated wind power output was subsequently used in WASP IV package to model
a candidate wind power plant and to optimize two generation expansion plans with
and without the modeled wind park generation to evaluate the applicability of the
proposed concepts. After optimizing the two generation expansion plans, the WASP
IV output data was used to present a methodology to evaluate the benefits of the
modeled wind park in WASP IV package.
WASP IV package uses a least cost methodology to evaluate different alternative

technologies based on their stand alone generation costs, and since the risks of
alternative technologies are not taken into account when formulating the long term
generation expansion plans, a Matlab model was developed to utilize Mean Variance
Portfolio Theory to evaluate the risks of proposed CEB generation mixes. The target
CEB mixes were then compared with efficient portfolios identified by the model.
Finally, the proposed concepts, models, methodologies and their applicability on the
present planning approaches were discussed and possible future work and
improvements were identified.
1.3 Outline of the Thesis

Chapter 2 provides the background of approaches to mainstream NCRE
developments and the barriers to mainstream adoption of NCRE developments in Sri
Lanka. Since regulatory decisions play a major role in promoting renewable energy
developments in a country, the existing regulatory interventions in Sri Lanka towards
5
mainstreaming NCRE developments were also examined along with strategies to
overcome the identified barriers.
Chapter 3 describes the long term generation planning process of Sri Lanka and
issues associated with the approaches being used at present. Identified issues were
addressed conceptually and ways to modify the existing planning process to
accommodate the proposed concepts were also discussed. All subsequent chapters
describe the application of these proposed concepts on the Sri Lankan system.
Methods of calculating the capacity credit of wind generation is discussed in Chapter

4. Several widely used methods were discussed and the method best suited to the Sri
Lankan context was proposed after taking into consideration the input data
requirement of the method and the accuracy of the prediction. Chapter 4 also
describes the computer model developed by using Matlab to calculate the monthly
capacity credit and the monthly energy production of a wind plant in the initial
planning stages.
Chapter 5 discusses the feasible methods of modeling a wind power plant in WASP
IV package, application of the preferred method its results. This chapter also details a
methodology for the evaluation of a modeled wind plant’s energy, capacity and
emission credits by using the output data of WASP IV simulations which can be
easily extended to evaluate the benefits of any NCRE plant.
Chapter 6 discusses the Mean Variance Portfolio Theory and its application in
evaluating the target CEB generation mixes of 2012 and 2025. Two scenarios were
evaluated by including and excluding the costs of CO2 in the generating costs of
alternative technologies and the results were discussed in depth.
The final chapter is of the conclusions reached in this study along with the reviews of
the objectives of this research and what it has achieved. Further, it summarizes and
discusses the results obtained and also lists possible future work that could be carried
out.
6
Chapter 2
MAINSTREAMING NCRE DEVELOPMENTS IN SRI LANKA:

CONSTRAINTS AND STRATEGIES TO OVERCOME THE
BARRIERS
Meeting the increasing demand for electric power with clean and sustainable energy
has become one of the major challenges the world is facing in the 21st century. Many
countries are actively seeking to mainstream renewable energy technologies to
address this challenge and renewable energy targets have already been set by many
countries including Sri Lanka. Therefore, proper planning is crucial for achieving the
set targets and successful implementation of policy decisions. Although certain
countries like Germany, Japan and Denmark have been successful in mainstreaming
renewable energy technologies, there are many constraints which should be
addressed so as to achieve the desired results [10]. The most common difficulties
associated with planning, developing and integrating renewable energy technologies
are discussed in the following section.
2.1 Constraints to Mainstream Adoption of NCRE Technologies in Sri

Lanka
Even though there are significant reductions in generating costs and rapid
technological advances in renewable technologies, the penetration has not been all
that successful in Sri Lanka like in many other developing countries. Therefore, it is
essential to identify the constraints to mainstream adoption of NCRE technologies
for successful penetration of these technologies. However, owing to the unique
characteristics of renewable technologies, it is difficult to apply a one-size-fits-all
approach to all the resource types. Wijayatunga et al. [11] highlights the need to have
a source specific approach when it comes to formulating renewable energy policies
and have identified source specific barriers prevailing in Sri Lanka for technologies
such as wind, biomass thermal power and Liquefied Natural Gas (LNG).
It has been found that lack of financing instruments, high initial costs and seasonal
variations in wind speeds act as barriers for the development of wind power plants.
Since most of the sites with good resource availability are small in size, the
7
advantage of economy of scale is eliminated and the project payback periods are
generally longer than other conventional renewable technologies like mini hydro.
Similar to wind power, lack of knowledge and project experience has resulted in less
financial facilities and support by commercial banks for the development of biomass
power plants. Further, the lack of feedstock assurance also hinders the biomass
power developments due to concerns regarding continuous supply of fuel wood [11].
Additionally, most NCRE technologies are highly site constrained. The most viable
NCRE sites such as wind are more often than not located far away from load centers
and existing transmission facilities. For example, the best wind resources in Sri
Lanka are found along the North-Western costal line and central hills where
transmission facilities are not well developed. Therefore, transmission facilities need
to be developed to reach the project sites or else the lack of access to the electric grid
at reasonable prices would commonly act as a barrier for all NCRE resource types.
The lack of information and public awareness has also been identified as another
constraint that weighs down the development of NCRE technologies. Generally, high
quality historical data of many years is required to conduct proper feasibility studies
before selecting a site to develop a renewable energy plant. Thus, lack of historical
data may delay such developments and would also discourage the potential investors.
The lengthy project approval process has become an added barrier for the NCRE
developers to forge through. A number of social and environmental permits have to
be obtained from several different departments and authorities in Sri Lanka such as
the Sustainable Energy Authority, Department of Forests, Department of Wildlife
Conservation, Divisional Secretary, Department of Irrigation, Department of
Archaeology, Ceylon Electricity Board, etc and hence approval process often tends
to be extremely tedious and time consuming [12].
Factors such as the lengthy approval process, lack of information and public
awareness, difficulties in financing, non-implementation of proper payment
mechanisms and non-establishment of a level playing field for NCRE technologies
can become strong barriers to the development of NCRE projects in the absence of
8
consistent and long term regulatory support. Many countries that have succeeded in
mainstreaming NCRE technologies have crafted a right mix of policies to support the
development of these technologies. Consistent policies would boost investor
confidence in these new technologies as well as public support and awareness. Since
the government policies play a major role in a country’s efforts to mainstream NCRE
technologies, the next section discusses the present status of Sri Lanka and attempts
to identify strategies to overcome the barriers mentioned above.
2.2 Policy and Regulatory Interventions for NCRE Developments

The International Energy Agency (IEA) projects world electricity demand to grow at
an annual rate of 2.5% and over 80% of this growth is expected from non OECD
(Organization for Economic Cooperation and Development) countries [13].
However, many of these countries have hardly attempted to mainstream their
indigenous renewable energy sources. As discussed previously, NCRE developments
are being challenged by various barriers and countries with high renewable energy
penetrations have shown that most of these barriers could be overcome by supportive
and consistent government policies.
2.2.1 Policy and regulatory interventions in Sri Lanka

Sri Lanka has taken several steps to promote the development of NCRE resources.
The national energy policy lays emphasis on the development of indigenous energy
resources as one of its key elements and it has identified small hydropower, wind
energy and biomass power as the three main NCRE sources to be promoted in Sri
Lanka for power generation. Due to the recent policy and regulatory interventions,
the small hydro power sector has experienced a rapid expansion and many new wind
power projects have also been connected to the national grid during the past few
years.
The establishment of Sri Lanka Sustainable Energy Authority (SLSEA) in 2007 and
the introduction of technology specific, cost based feed-in tariffs for all types of
renewable energy based technologies are identified as the two main measures that
contributed to the development of NCRE resources in Sri Lanka. The Standardized
Power Purchase Agreements (SPPA) offered by the CEB with guaranteed unlimited
9
power purchase has boosted investor confidence and have been successful in
attracting both local and foreign investors into development of NCRE projects [14].
Further, the CEB has introduced a net-metering scheme which enables the electricity
consumers to export any excess electricity generated from their renewable energy
based generating systems to the national grid. These steps along with many other
initiatives have been successful in Sri Lanka not only in terms of capacity additions
but also in diversifying NCRE developments.
2.2.2 Policy lessons from around the world

Sri Lanka has taken several constructive measures for the development of NCRE
resources and been successful to a certain extent as discussed in the previous section.
However, many obstacles still exist and aggravate the efforts of reaching national
NCRE targets. On the other hand, several countries have proven that sustained
NCRE developments can take place when a right combination of policies is in place.
The dominant roles of countries like Germany, Denmark, Japan and United States in
developing renewables have shown that the resource availability alone is not
sufficient to achieve the expected targets. Their success has been a result of
consistent and long term policy decisions, provision of access to the electric grid at
attractive prices, good tariff structures, tax and other financial incentives, technology
standards, education and improved public awareness, etc [15].
The experiences of these countries have provided successful policy elements that can
be adopted around the world. These policy elements can be categorized into several
key areas as discussed below.
• Pricing laws and grid access

Pricing laws have proven to be the most successful policy element when it
comes to promoting NCRE developments [16]. Good tariff structures not
only provide the power producers a guaranteed long term income, but also
enhance the investor confidence as well. When the market access is secure, it
is easier to obtain project financing due to low investment risks and makes it
profitable to invest in renewable technologies. As mentioned earlier, the CEB
10
also provides long term SPPAs and technology specific feed-in tariffs and
both these measures have already proven to be successful.
It is mandatory in some countries for a certain percentage of electricity to be

generated from renewable sources. These quota systems, commonly known
as Renewable Portfolio Standards (RPS) are used in many developed
countries such as Australia, Japan, Italy, United Kingdom and several states
of the United States of America for many years. Further, it has been found
that the government’s NCRE target of 10% by 2015 can be achieved with an
additional cost burden of only 1.3% of the total cost of the generation plan, as
well as that regulatory interventions such as RPS can be introduced without
significant additional costs for small power systems like in Sri Lanka [17].
• Financial Incentives
Various financial assistance schemes and other compensation mechanisms
such as low interest long term loans, tax credits, rebates and production
incentives that subsidize the investments made in development of NCRE
projects have been extensively used in Europe, Japan, India and the United
States. These financial incentives have boosted the NCRE developments in
India to such an extent that it is now one of the largest wind power producers
in the world [15].
Sri Lanka also offers tax holidays for the Board of Investments (BOI)
approved projects. However, these policies must be tailored carefully so as to
avoid any misuses. Many mini hydro projects in Sri Lanka were delayed due
to lack of specific deadlines. There had also been instances where the
developers who reserved the project sites by submitting a Letter of Intent
(LOI) to the CEB, had sold the site development rights to a third party purely
for financial gains. These unhealthy practices and situations were somewhat
brought under control by the government’s decision to introduce a time limit
for development activities to start and progress. Therefore, periodic reviewing
is very important to evaluate the effectiveness of the offered incentives.
Further, all these subsidies should be gradually reduced and phased out to
11
encourage cost reductions with the aim of passing the benefit on to the
country’s economy.
• Information dissemination, training, education and public awareness

Even if the governments offer attractive financial incentives, the NCRE
development drive will be less effective if the people are uninformed or
misinformed about the offered incentives, availability of resources,
advantages and potential of renewables, technology developments and other
relevant data resources like historical data and feasibility studies. Therefore,
it is very important to setup responsible authorities to collect data, carry out
studies and to educate the interested parties and the public in general. Further,
people should be educated and trained to work in these systems. For example,
many countries have included renewable energy related courses in schools,
universities and in many vocational training centers, renewable energy related
programs are conducted to train people to install, maintain and manufacture
renewable energy systems [18].
Without proper information, the general public and the potential investors
will be left with a perception that renewables do no work, too expensive or
either too risky to invest in. Therefore, it is vital to share the available
information across all levels to increase the awareness and for successful
policy implementation.
SLSEA is entrusted with most of the above activities in Sri Lanka. The
SLSEA is actively engaged in activities like data collection, resource
mapping and it also maintains a library which contains data of renewable
energy related studies. However, a lot of improvements are yet to be made
and the CEB is also hugely responsible in this aspect as the main electricity
utility in the country.
• Setting standards
Setting technology standards is essential to prevent substandard technologies
from entering the market, which in turn will reduce risks and boost investor
12
confidence. The availability and use of proper planning standards can even
reduce the public hesitation and opposition towards renewable projects if the
set standards address the concerns of the public such as environmental
impacts and noise. Technology standards for instance can be applied to
anything from turbines, safety systems, compatibility with the electric grid
and to performance characteristics such as harmonics generation and other
power quality issues. For example, Denmark adopted wind turbine standards
in 1979 mainly due to the popularity of wind generation and these standards
have played a major role in the development of the wind turbine
manufacturing industry in Denmark [19].
On the contrary, there is a clear lack of standards in Sri Lanka. No institution

is held responsible for formulating relevant standards and thus hardly any
standards exist with regard to the technological aspects of renewable energy.
As a result, cheap and inferior equipment such as solar panels can enter the
market and the resulting bad experiences associated with its usage will surely
damage the public perception on renewable technologies. Further, there are
no performance standards such as harmonic limits and this may create
additional problems for the CEB when it comes to maintaining power quality
of the system. Hence, establishing a proper framework for formulating the
relevant standards will surely benefit the NCRE development efforts in Sri
Lanka.
• Stakeholder involvement in NCRE developments

Public participation in regulation, project development and ownership can
also assist NCRE developments. Involvement of local communities in project
development will often ease the resistance towards new projects and ensure
smooth implementation. Local labor and other recourses could be used to a
great extent during construction and operation of the plants and by doing so,
it will implant a sense of ownership within local communities. In contrast,
when projects are forced on people without their having a fair say and a share
in it, they place very little value on them.
13
For example, in Denmark and Germany, a strong public support has been
experienced by wind energy. By 2002, 85 % of the installed capacity of wind
power providing about 15 % of total electricity consumption in Denmark had
been developed through locally driven initiatives and owned by farmers or
wind turbine co-operatives concerning about 2400MW of installed capacity.
In Germany, at least 340,000 individuals had collectively invested about 12
billion Euros in NCRE projects such as wind, solar, biomass and geothermal
projects as of 2002 [16]. Therefore, it is essential to share the ownership of
the projects with the local communities because not only would it enable to
share the benefits and risks of renewable energy but would also ease the
problems associated with financing, siting, planning and operation.
In Sri Lanka, the NCRE project developers usually respond to the needs of
the local communities by developing local infrastructures such as roads and
bridges. Several project development companies have also shared the
ownership of the projects by issuing shares to the public. This is most
certainly a very healthy situation and it has contributed to the rapid expansion
of the small hydropower sector in Sri Lanka as well.
According to the above discussion, it is evident that a careful selection of the above
mentioned policy options will play a major role in promoting NCRE developments
of a country. However, a level playing field has to be created by eliminating the
subsidies on conventional technologies and the external costs of conventional
generation shall also be taken into account. In conclusion, policies designed to
promote the development and the use of renewables can fail if they are not well
formulated, inconsistent or unsustained. Therefore, consistent and long term policy
decisions are crucial to ensure continuous market growth, to enable the development
of domestic industries and labor force, to reduce the risk of investments in NCRE
technologies and to make it easier to obtain financing for project developments.
14
Chapter 3
REVIEW OF THE CEB GENERATION PLANNING PROCESS

Modern power systems are highly complex and therefore, the system planners use
computer aided capacity expansion models to model and develop the generation and
transmission expansion plans. These expansion models require many inputs such as;
• Macroeconomic data (Electricity demand, Fuel prices, Economic growth, etc)

• Fixed costs (Capital costs, Fixed Operation and Maintenance or O&M costs)
• Variable costs (Variable O&M costs, fuel prices, heat rates)
• Financial parameters (Discount rate, Exchange rates, Interest rates)
• Technical parameters of generating units and power plants (Reliability,
Lifetime, Maintenance requirements, Emissions, Construction period)
However, finding an optimum long term generation expansion plan has become
highly complex and requires tremendous effort because most of these input data
cannot be accurately represented in an expansion planning problem due to multiple
sources of uncertainties such as;
• Technological innovations (Cost reductions, Efficiency improvements,

Storage technologies, etc)
• Climate conditions (Hydrological conditions, wind resource availability,
Solar irradiation)
• Global economic conditions and unrests
• Regulatory changes in a country
• Public support or opposition for a particular technology
Traditionally, the most commonly used criterion is to minimize the total cost to meet
a given demand and most of these computer aided planning tools have been
developed to facilitate the modeling of conventional power generation technologies
such as oil, coal and large hydro. Therefore, these traditional generation planning
tools often lack the modeling capabilities to incorporate NCRE technologies in the
capacity expansion studies. A majority of these tools fail to capture the intermittency
15
and variability associated with many NCRE technologies, and hence fails to properly
evaluate the true benefits of technologies such as wind. As a result, traditional
planning tools and methodologies are generally biased towards conventional
technologies and these shortcomings make them less efficient in planning modern
power systems.
3.1 Overview of the CEB Generation Planning Methodology

The Ceylon Electricity Board compiles the only long-term generation expansion plan
in Sri Lanka and uses computer models such as WASP and SYSIM (System
SIMulation package for hydro modeling) in their planning process. The planning
studies are conducted for 20 years and results are published for 15 years. The
generation plan is updated every year by updating the inputs of the WASP package
such as the generating costs, demand forecasts, operating parameters of the existing
plants, costs and operating parameters of candidate options and other related inputs.
All cost related inputs are based on the economic prices as of the 1st of January of the
first year of the planning period and it excludes all taxes and duties. The exchange
rate is taken as the average value during January of the same year. The fuel costs are
assumed to be constant during the planning period in the base case scenario. Other
cost inputs such as capital, operating and maintenance costs are compiled by the
CEB. An economic discount rate of 10% is used in the base case for discounted cash
flow analysis and to calculate the net present value of the alternative scenarios.
A number of sensitivity studies are carried out to examine the sensitivity of the base
case plan to the variations in key input parameters such as discount rate, fuel price
fluctuations and demand forecast. The retiring and commissioning of power plants
are assumed to be carried out at the beginning of the relevant year and a salvage
value for retired plants is also added to the cost function as a benefit. A Reserve
margin and a target Loss of Load Probability (LOLP) value could be used as
constraints in the WASP package to ensure that the planned generation capacity is
sufficient to meet the forecasted demand. However, these are not used as constraints
in the present planning process. In addition, a penalty of US$ 1.2 per kWh is
imposed on the amount of electricity that is not supplied if the plan fails to meet the
16
forecasted demand of each year, and added to the total net present value cost of the
evaluated scenario [3].
The least cost plan is generally identified as the reference case and many sensitivity
studies are carried out to evaluate the effect of many factors such as changes in
petroleum prices, low and high discount rates and demand scenarios, high NCRE
generation, etc on the base case plan.
3.2 Wien Automatic System Planning Package (WASP)

The generation planning branch of the CEB uses the WASP package to find the least
cost generation plan. The WASP package was initially developed by the Tennessee
Valley Authority (TVA) and Oak Ridge National Laboratory (ORNL) of the United
States of America as a requirement for the International Atomic Energy Authority's
(IAEA) Market Survey on Nuclear Power in Developing Countries that was
conducted in 1972–1973 [20]. This model has been used extensively since then and
many improvements have been made over the past three decades by adding various
features.
The current version, WASP-IV attempts to find the economically optimal generation
expansion plan for a power generating system under user-specified constraints. It
minimizes the discounted costs of generation over a study period, which primarily
comprise of capital investment, fuel cost, operation and maintenance cost, cost of
Energy-Not-Served (ENS) and salvage value of investments. The programme uses a
probabilistic estimation approach to calculate the generating costs, cost of ENS, and
to assess the reliability of the generation plan. In addition, a linear programming
technique is used to determine an optimal dispatch policy that satisfies user defined
constraints and a dynamic method of optimization is used to compare the costs of
alternative system expansion policies. It evaluates all feasible sets of power plants to
be added during the planning horizon while satisfying all user defined constraints
and provides the capacity to be added in the future and the cost of achieving such
capacity additions.
17
3.2.1 Objective function of the WASP IV package
The objective cost function (Bj) which is evaluated by WASP IV can be represented
by the following expression [20]:
Bj = Σ [Ij,t- Sj,t] + Fj,t+ Mj,t+ Oj,t
Where,
I = discounted capital investment costs
S = discounted salvage value of investments
F = discounted fuel costs
M = discounted operation and maintenance costs
O = discounted cost of energy not served
t = length of the study period (number of years)
j = total number of possible expansion plans
Amongst all possible expansion plans, the plan with a minimum Bj is regarded as the
optimal expansion plan.
The programme can model up to 12 types of thermal plants and two categories of
hydro plants as candidate options in the expansion plan. The hydro plants are
assumed to be 100% reliable and have no associated cost for water. The stochastic
nature of hydrology is treated by means of hydrological conditions (up to 5), each
one defined by its probability of occurrence and the corresponding available capacity
and energy of each hydro plant in the given condition. The hydrological data used to
model the energy capabilities of existing and candidate hydro plants are developed
by the SYSIM model.
The cost of energy not served represents the expected loss to a country’s economy
when a certain amount of electric energy is not supplied. Both operating costs and
investment costs for the candidate plants are considered while only operating costs
are considered for the existing plants. Costs in relation to Fuel, O&M, and ENS are
assumed to occur in the middle of the corresponding year and at the end of the
planning horizon, a salvage value is also included in the cost function, as a benefit.
18
3.2.2 Principal capabilities and limitations of WASP IV [20]
The capabilities and the limitations of some key inputs of the WASP IV package are
summarized in Table 3.1.
Table 3.1 : Principal capabilities and limitations of WASP IV

Maximum
Parameters
allowable limit
Years of study period 30
Periods per year 12
Load duration curves (one for each period and for each year) 360
Cosine terms in the Fourier representation of the load duration 100

curve of each period
Types of plants grouped by "fuel" types of which:
10 types of thermal plants; and 2 composite hydroelectric 12
plants and one
pumped storage plants
Thermal plants of multiple units. This limit corresponds to the
total number of plants in the Fixed System plus those thermal 88
plants considered for system expansion in the Variable System
Types of plants candidates for system expansion, of which:

12 types of thermal plants; 2 hydroelectric plant types, each 15
one composed of up to 30 projects; and 1 pumped storage
plant type with up to 30 composed projects.
Environmental pollutants (materials) 2
Group limitations 5
Hydrological conditions 5
System configuration s in all the study period 5,000
Source : WASP-IV with User Interface User’s Manual, IAEA, Vienna, 2006
19
3.3 Other Generation Planning Tools
There exists a wide range of generation planning tools which are diverse in terms of,
inter alia, scope of study, planning horizons they analyze, structure of the electricity
market it is applied to, technologies they consider, accessibility to tools (cost, method
of distribution, etc) and level of detail in results [21]. However, this thesis does not
intend to identify the ideal generation planning tool, but rather attempts to provide an
overview of two closely comparative and extensively used tools; namely Electric
Generation Expansion Analysis System (EGEAS) and PLEXOS in comparison with
WASP IV. A review of many other commercial and non-commercial computer tools
available for modeling of power systems and capacity expansion planning such as
EMCAS (Electricity Market Complex Adaptive System) and AURORAxmp can be
found in [22, 23].
3.3.1 Electric Generation Expansion Analysis System

EGEAS was developed by researchers at Massachusetts Institute of Technology
(MIT) in the early 1980s under the Electric Power Research Institute (EPRI) funded
project for the development of a flexible, modular, environmentally sensitive state of
the art electric generation expansion analysis system to facilitate integrated resource
planning of electricity systems [24]. Since its inception, EGEAS has been
extensively used in countries like the United States and India and its results are well
accepted by regulators [25 - 27].
New features, such as demand-side management and economic dispatch under

deregulated markets have been added over the past three decades to address the
changes in modern power systems. The model uses dynamic programming to select
capacity additions from candidate options, and it takes into account the forecasted
demand, system operation, plant retirements and demand-side management options
to decide whether new capacity additions are required or not. Reliability and
uncertainty analysis is also available, and the programme can be run in both
expansion optimization and production simulation modes. Further, emission analysis
can also be performed and if required, optimization constraints such as pollutant cap
levels can be incorporated. EGEAS supports many supply-side and demand-side
20
options and in addition to conventional power generation technologies, it also
facilitates the integration of NCRE technologies such as wind, solar, biomass and
distributed generation [28].
3.3.2 PLEXOS for Power Systems

PLEXOS is a sophisticated power system modeling tool used in more than 30
countries worldwide for modeling and planning of electricity markets. Deterministic
or stochastic programming techniques are generally used for modeling capacity
expansion problems and the objective function aims to minimize the net present
value of build and retirement costs of generating facilities, fixed operations and
maintenance costs, ENS costs and the expected production costs that follow
investment decisions [29]. The optimization can be carried out under various user
defined constraints such as the availability and operational characteristics of
generating plants, environmental constraints, reliability requirements, fuel costs and
transmission constraints. A simplified model description of PLEXOS capacity
expansion planning tool can be found in [30]. When compared with WASP IV and
EGEAS, PLEXOS has the ability to facilitate many additional functions and
applications in relation to resource planning studies such as transmission capacity
expansion planning, portfolio optimization, operational planning and optimization of
generation and transmission networks [31]. WASP IV and EGEAS dispatch the
generation system using a monthly load duration curve method in contrast to the
intra-hour dispatch capability of PLEXOS which makes it more detailed than the
other two tools and also allows co-optimizing of generation and transmission.
In addition to the above mentioned features, PLEXOS has robust renewable

integration analysis facilities which are utilized in many studies [32 - 36]. It is
capable of modeling one minute level variations in load or generation caused by
intermittent sources, and also has the capacity to evaluate its effect on system
stability. Further, it can also capture the intermittency and variability of NCRE
generation via various autoregressive models for wind speed, solar radiation and
stochastic optimization of forecast uncertainty.
21
3.4 2011-2025 Generation Expansion Plan of the CEB (Base Case Plan)
Table 3.2 summarizes the base case generation expansion plan prepared by the
generation planning branch of the CEB for the years 2011 to 2025. The proposed
candidate options according to the base case plan are shown in bold letters and other
capacity additions are from committed power plants.
Table 3.2 : 2011-2025 Generation expansion plan of the CEB (Updated base case)
Year Renewable Thermal Additions Thermal Retirements LOLP
Additions %
2011 5MW Biomass 1x315 MW Puttalam Coal - 0501
Plant power plant (Stage I)
2012 150 MW Upper - 4x5 MW Ace Power 0.419
Kotmale Matara (Furnace Oil)
2013 - - 4x5 MW ACE Power 1.473
Horana (Furnace Oil)
4x5.63 MW Lakdanavi
(Furnace Oil)
2014 5MW Biomass 2x315 MW Puttalam Coal 5x17 MW Kelanitissa 0.049
Plant power plant (Stage II) Gas Turbines (Auto
4x5 MW Northern Power Diesel)
(Furnace Oil)
4x6 MW Chunnakum
Power Extension (Furnace
Oil)
1x75 MW Gas Turbine
(Auto Diesel)
2015 - 2x35 MW Gas Turbine 14x7.11 MW Ace 0.698
(Auto Diesel) Power Embilipitiya
(Furnace Oil)
6x16.6 MW
Heladanavi Puttalam
(Furnace Oil)
4x15 MW Colombo
Power (Furnace Oil)
2016 35 MW - - 0.734
Broadlands
120 MW Uma
Oya
5MW Biomass
Plant
2017 5MW Biomass 2x250 MW Trincomalee - 0.091
Plant Coal power plant
2018 49 MW Gin 1x250 MW Trincomalee 8x6.13 MW Asia 0.079
Ganga Coal power plant Power (Residual Oil)
5MW Biomass 4x5 MW Northern
Plant Power (Furnace Oil)
22
2019 - 1x250 MW Trincomalee 4x18 MW 0.106
Coal power plant Sapugaskanda
(Residual Oil)
2020 2x5MW Biomass - - 0.450
Plant
2021 5MW Biomass 2x300 MW Coal power - 0.042
Plant plant
2022 5MW Biomass 1x300 MW Coal power - 0.035

Plant plant
2023 5MW Biomass 1x300 MW Coal power 4x9 MW 0.046
Plant plant Sapugaskanda Diesel
Ext. (Residual Oil)
2024 - - 163 MW AES 0.707
Kelanitissa (Auto
Diesel)
2025 5MW Biomass 2x300 MW Coal power 4x9 MW 0.245
Plant plant Sapugaskanda Diesel
Ext. (Residual Oil)
Total Present Value (PV) Cost up to year 2025, US$ 7,973.8 million (This value
includes US$ 389.3 Million as projected costs of committed NCRE plants)
Source : CEB Long Term Generation Expansion Plan 2010
According to the plan, 84% of the new capacity additions are from coal based power
plants and a thermal capacity of 3,654 MW needs to be added to the system by 2025,
while 900MW of thermal plants will be retired. The thermal capacity share will be
increased up to 65% by the year 2020 and will subsequently reach 74% by the year
2025 with the heavy thermal capacity additions [3].
3.5 Identified Issues in the Present Generation Planning Process

• Since, the present planning process is least-cost based, factors such as risks
related to fuel, investment as well as operation and maintenance costs are not
taken into consideration when selecting alternative capacity expansion
options.
• Only a several conventional thermal generation technologies and a few
selected large hydro plants are considered as candidate options. NCRE
options such as wind and biomass are not considered as candidate power
plants.
23
• The benefits of NCRE (Capacity Credits, Energy Credits and Emission
Credits) are not evaluated in depth and the external costs such as carbon costs
are not taken into account in the optimization process of the generation plan.
• The National Energy Policy targets to generate 10% of electricity from
NCRE sources by 2015. Even though a sensitivity study is carried out in the
CEB plan to evaluate the effect of this target, a clear plan or methodology is
not presented on achieving this targeted NCRE penetration.
• The capacity contribution from intermittent sources such as wind power is not
considered when determining the adequacy of the generation capacity.
As stated in Chapter 2, small hydropower, wind energy and biomass power have
been identified as the three leading NCRE technologies in Sri Lanka with a good
potential for development and at present, mini hydro and biomass plants are
adequately modeled in the present generation planning process. As proposed by a
previous study [37], mini hydro plants are modeled as Run-Of-River (ROR) plants.
Several plants are lumped together before including them in WASP IV so as to
reduce the number of plants and modeled as committed plants due to the limited
number of candidate options that could be included in the WASP IV model. Biomass
plants are modeled as conventional thermal plants, which is straightforward and
simple. Wind generation is also included in a limited level using a load modification
technique, where only the energy contribution is considered. The present modeling
techniques used to model wind plants in WASP IV along with the associated
shortcomings are discussed in detail in Chapter 5.
3.6 Proposed Concepts to Modify the Existing Planning Process

Even though there are limitations in WASP IV computer model as a planning tool, it
is still widely used in many countries and is considered as a powerful generation
planning tool. Therefore, it was decided to propose feasible modifications to the
existing planning process while maintaining compatibility with the present practices,
rather than proposing an entirely new planning approach. Since mini hydro and
biomass power plants are being adequately modeled at present, this thesis focuses on
modeling wind power in WASP and therefore a new methodology to model wind
power is proposed in this thesis. This will permit the generation planners to
24
incorporate the capacity contribution of wind power which in turn will enable the
modeling of wind plants as candidate options.
There are several ways to model wind power in expansion planning models such as
WASP IV. At present, the CEB’s generation planning branch uses a load
modification approach (‘Negative Load’ approach). Although this approach captures
the variability of wind power, it is more appropriate for short-term studies and could
provide inaccurate results when it comes to long-term studies [38]. In this thesis, a
supply-side approach is proposed in Chapter 5, in which a wind plant could be
modeled either as a ROR hydro plant or an unreliable thermal plant, and this
approach is more appropriate for long term studies due to its ability to capture some
variability and uncertainty of wind power [38]. Further, modeling a wind plant as a
conventional plant enables the system planners to use the WASP results to evaluate
the benefits of the plant in detail. Even though chronological hourly wind
information is not captured by this method, it does not have much of an impact on
long-term studies.
In addition to modeling a wind plant in WASP IV, a methodology to evaluate the

benefits of the modeled NCRE plants will also be proposed in Chapter 5 to calculate
the energy, capacity and emission credits of the NCRE plants by using the WASP IV
results.
Finally, application of the MVPT is proposed as a complementary planning tool to

evaluate the risks in the Sri Lankan generation mix. Energy planning also represents
an investment-decision problem and the investors often apply portfolio theory to
manage risks and to maximize portfolio outcomes under a variety of unpredictable
economic outcomes. The same theories could be used to analyze a generation mix
and decide on optimal generating portfolios that minimize costs at preferred levels of
risks. Therefore, Mean Variance Portfolio Theory (developed by Harry Markowitz in
1952) is proposed to evaluate the Sri Lankan generation mixes to assess its potential
application in the development of optimal generating portfolios that could enhance
energy security and diversification objectives of the national energy policy.
25
Chapter 4
CAPACITY CREDIT OF RENEWABLE GENERATION AND

WIND POWER OUTPUT MODEL
4.1 Power System Planning and System Adequacy
Reliability is the ultimate goal in any power system and power system reliability is
basically divided into two aspects namely, system security and system adequacy. A
power system is considered secure if it can withstand the loss of key elements such
power plants or transmission lines. A system is said to be adequate if there is
sufficient installed generation capacity to meet the demand, and transmission
capacity to deliver the generation to load centers under different conditions such as
during dry weather and maintenance of key power plants and transmission lines.
Therefore, long term generation planning is carried out in such a way that it ensures
an adequate installed capacity to meet the forecasted demands under predefined
reliability targets.
The reliability metrics such as Loss of Load Probability (LOLP) and Loss of Load
Expectation (LOLE) are often used for adequacy evaluations. The probability that
the available generation capacity will be insufficient to meet the system demand at a
given time is called the Loss of Load Probability or LOLP. The duration of time
where the demand will not be met over a defined period of time is called the Loss of
Load Expectation. The capacity contribution that any given generator makes to the
system adequacy is called the Effective Load Carrying Capability (ELCC) or in other
words, the additional load that can be served by the new generator while maintaining
the same reliability levels [39] (Often measured in LOLE).
4.2 Methods of Calculating the Capacity Credit of NCRE Generators

Even though there are concerns on how renewable generators such as wind plants
can have a capacity value when there are situations with no resource availability
(wind), it should be noted that no power generating technology is perfectly reliable.
Forced outages occur even in conventional power plants due to various reasons.
However, it is unlikely that conventional generators will experience forced outages
26
because of fuel shortages as in renewable generators. Still, there is always a non-zero
probability that it may not be available during peak load periods. Therefore, the same
methodologies which are used to calculate the ELCC of conventional generators
could be used to calculate the capacity credit of NCRE generators as well.
Several publications exist on how to calculate the capacity credit of NCRE plants and
most methods presented are based on power system reliability analysis methods.
According to Milligan and Porter (2005) [40], there are several properties that a good
capacity credit metric should possess where it should be commonly usable to assess
different types of generators, it has to be data driven, mathematically consistent and
simple. Since all generators have some probability of failure, the metric has to
identify the generators that consistently delivers during high-risk (peak) periods and
should assign a relatively high capacity value to those generators.
The most consistent and widely accepted method is using the ELCC and it is based
on well-established reliability theories and can be applied to all types of generators
including baseload, conventional and intermittent generation [41]. The ELCC of a
renewable generator could be calculated with the help of a power system reliability
model, and ELCC can differentiate generating units with differing levels of
reliability, size, and power delivery during peak and off-peak periods. It can
effectively assign high capacity values for plants that delivers consistently during
peak demand periods and when it comes to intermittent generators such as wind,
high capacity values are assigned for plants with output profiles that are highly
positively correlated with system load profiles.
Even though the ELCC method is rigorous, data-driven and consistent in

distinguishing generators that have different impacts on the system reliability, it
requires a lot of quality data such as time series of system loads, NCRE generation,
conventional generation and also a database of forced outage rates and specific
maintenance schedules of conventional generators. Hence, several simpler methods
have been developed over the years to approximate ELCC due to the unavailability
and potential difficulty in assembling appropriate databases. These approximation
methods are sometimes based on renewable generation during high system risk hours
27
(Capacity Factor methods) and in other instances, the system’s LOLP is
approximated. In addition, a Multi-State unit representation of renewable generators
in several de-rated states could also be used as an alternative to the preferred ELCC
method, which utilizes a probabilistic representation of the renewable generator.
4.2.1 Effective Load Carrying Capability

This method is widely used to find the capacity contribution of conventional thermal
generation and could also be used to find the capacity contribution of intermittent
generations as well. Conventional generators are usually modeled by their capacities
and their Forced Outage Rates (FORs). However, since intermittent generation can
have many de-rated states, NCRE sources like wind power cannot be adequately
modeled by its capacity and FOR, and thus requires a different approach than the
traditional calculation method [42]. This section describes the methodology of
calculating the ELCC of intermittent generation with the help of a reliability model.
When the capacity contribution from the intermittent generator is not considered, the
annual LOLE can be calculated as,
LOLE = P( C < L )

Where, P() denotes the probability function, Ci is the available capacity in hour i, Li
is the hourly system load and N is the number of hours in the year. Adding the
capacity from the intermittent generator will enhance the system reliability and the
new LOLE’ could be expressed as,
LOLE′ = P[( C + IG ) < L ]

Where IGi is the available output from the intermittent generator during hour i. The
ELCC of the system is the load that can be supplied at a predefined level of risk of
loss of load.

P( C < L ) = P[( C + IG ) < (L + AL )]

28
When the increased capacity due to the contribution from the intermittent generator
can supply the additional load of ALi at the same reliability level that the original
load Li could be supplied with the Capacity Ci, the value of ALi represents the
Capacity Credit or the ELCC of the intermittent generator.
Since it is not possible to analytically solve the equation, ELCC is calculated in

several steps using a reliability model. First, the system is modeled without the
renewable generator. The generation time series and the load time series is used to
find the daily LOLP and then the annual LOLE. If required, the loads are adjusted to
achieve the desired reliability level.
Secondly, the renewable generation time series is treated as a negative load and is
combined with the original load time series. The net load will be lower and the
model is re-run with the modified load to calculate the new LOLE. Due to the
reduction in the load, the LOLP will be decreased, which in turn will improve the
LOLE.
Finally, the load data is increased in small steps across all hours using an iterative
process and the LOLE is recalculated at each step until the original LOLE is reached.
The sum of incremental load steps at the original reliability level is the Capacity
Credit of the renewable generator.
4.2.2 Approximation methods

In the absence of quality data sets, it would be helpful to calculate the Capacity
Credit of a renewable generator with the minimal available data, and several methods
can be employed to calculate the capacity contribution.
I. Multi-state unit representation

NCRE generators can have a range of power output levels depending on the resource
availability. In this method, NCRE generators are modeled with partial capacity
outage states similar to conventional units with several de-rated states. Each of these
states has an associated probability and to evaluate the LOLP at a given time, the
NCRE generator is also treated as a multi-state conventional unit. ELCC of the
NCRE generator could then be calculated as described above in the previous section.
29
The probability distribution of the plant output is typically used to develop the multi-
state model with a number of capacity blocks to represent the NCRE generator in the
ELCC calculation. A major disadvantage associated with this method is the loss of
information on the correlation between demand and the timing of the NCRE plant
output. However, this concern could be addressed to a certain extent by using
different probability distributions to represent different categories of hours.
q
Pn, qn PTotal (MW)
Pn-1, qn-1
:
:
P3, q3
P2, q2
P (MW) P1, q1 0 MW
Figure 4.1: Probability distribution of the plant output and capacity blocks
II. Peak-period capacity factors

Capacity factor based methods are being used by many utilities around the world due
to its simplicity in application and reasonable accuracy [43]. Generally, the capacity
factor (defined as the ratio of the average output to the total output) of the NCRE
generator calculated over several suitable peak hours is considered as the Capacity
Credit of the NCRE generator.
This method is well suited for situations where there is a clearly notable peak period
in the daily load profile of the system and only one year of NCRE generation data is
adequate to carry out the calculations, although several years of data is always
preferred. Generally, the selected peak hours are weighted evenly, but if required,
higher weights could be placed on high load hours and low weights on low weight
hours.
30
III. Capacity factor during high LOLP hours
This method also calculates the capacity factor. However, instead of considering the
peak hours, this method calculates the capacity factor when the system LOLP is at its
highest. These hours are generally the highest load hours, and the NCRE generation
output during these hours is then used to calculate the capacity factor as an estimate
of the NCRE generator’s capacity credit.
Milligan and Parsons (1999) carried out these calculations for the top 1% to 30% of
loads using an increment of 1% to determine the usefulness of these simpler
approaches, and the results have shown that when approximately 10% or more of the
top load hours are considered, the capacity factor is within a few percentage points of
the ELCC [44].
4.2.3 Proposed method to calculate the Capacity Credit of NCRE generation

in Sri Lanka
The typical daily load profile of Sri Lanka takes the same shape throughout the year
and there is a clearly distinct peak period due to the domestic loads. According to the
system control center of the CEB, the system peak generally occurs between 19.00
and 20.00 hours with a steep increase from 18.00 to 19.00 hours followed by a
decrease in demand at a slow rate until about 22.00 hours.
Source: Study report on electricity demand and system peak reduction, PUCSL,
December 2012
Figure 4.2: Typical system load profile of Sri Lanka
31
Due to the high data requirement of the ELCC method, an approximation method is
proposed in this thesis to calculate the capacity credit of intermittent NCRE plants.
The peak period capacity factor method can provide accurate results when the system
peak period is known for the whole year, and requires less effort and data when
compared with other methods. In addition, it is suitable for long term studies and
therefore, the peak period capacity factor method was used to calculate the capacity
contribution of the NCRE plants. NCRE generation during 18.00 to 22.00 hours will
be taken into account considering the daily load profile of Sri Lanka. In the following
sections, this method is used to calculate the capacity credit of a hypothetical wind
plant using actual wind measurement data.
4.3 Wind Power Output Model

In this thesis, the capacity credit calculation of a NCRE plant was carried out using a
hypothetical wind plant proposed in Mannar which is still in its initial planning
stages. A wind power output model was developed using Matlab to calculate the
hourly wind plant output and energy production. The output from this model was
then used in capacity credit calculations and WASP optimizations. Actual hourly
wind measurement data collected in Nadukuda beach in Mannar obtained from Sri
Lanka Energies (Pvt) Ltd was used to evaluate the model.
First, the excel file containing the wind speeds and corresponding timestamps was
read into an array in Matlab and was filtered to remove any outliers by looping
through the array of wind speeds. Vestas V52 – 850kW turbine was selected as the
reference turbine in this analysis, and since the wind speed is affected by ground
surface friction, the collected wind data was then corrected for roughness to account
for the difference between the hub height of the selected turbine and the wind speed
measurement height. The collected wind measurements were available at 10, 20 and
40m and the 40m data was corrected using the commonly used Hellmann
exponential law that correlates the wind speed readings at two different heights and
is expressed by,
Hub Height 0
W = W !" # /
Initial Height
32
Where α is the friction coefficient or Hellmann exponent. For Mannar, α is assumed
to be 0.10 considering the local terrain and the measured values that were available at
10, 20 and 40m heights [45].
The power curve of the reference turbine is defined by the manufacturer. However,
the defined table shows power values only for discrete values. In order to calculate
the output power at any speed between the cut-off and cut-in wind speed, the variable
part of the power curve was fitted into a polynomial curve in the Matlab model.
Power curve values of the reference turbine are given in Appendix A and the plot of
the Matlab curve fitting is given in Appendix B.
As the next step, the power output of the turbine was defined in a separate function
which estimated the power output of the turbine based on the wind speed, and
individual turbine output for each hour was calculated. It was assumed that wind
incident is same on every turbine and each turbine acts independently. Therefore, the
total power output of the wind park was calculated for every hour using the sum of
the power produced by each turbine.
Finally, the capacity factor was calculated by using the hourly power output during
the system peak period and the monthly energy production of the wind park was
calculated after taking into account the losses. Generally, there are several sources
that contribute to the energy loss of a wind plant such as the wake effect, turbine
performance, plant availability, electrical efficiency, curtailments and environmental
losses. Most of these losses are site specific. However, in aggregate, total losses of a
wind power plant is typically in the 10-20 percent range, and in this thesis, the total
energy loss is considered to be 15 percent [46]. The Matlab code of the wind power
output model is included in Appendix C.
33
Historical wind speed time series
• Removed any outliers
• Corrected for surface friction (Tower height adjustment)
Wind turbine power curve

• Variable part is defined by a polynomial curve
Estimated hourly power output

• Capacity factor was calculated by using the power output
during peak hours
• Energy production was calculated after including the losses
Figure 4.3: Flowchart of the wind power output model
4.3.1 Output of the modeled wind park

To evaluate the model, wind data from Nadukuda beach in Mannar was used as per
the information given below, and the expected monthly outputs from the modeled
wind park is plotted in Figures 4.4 and 4.5. However, it has to be noted that the
reference turbine was selected assuming the wind park to be in its initial planning
stages. A more accurate wind park output could be obtained once the wind park
turbine is selected after detailed evaluations.
Wind park capacity – 100MW
Wind measurements – 17th Sep 2002 to 17th Sep 2003, at 40m height in one
hour resolution
Turbine – Vestas V52 – 850kW, 52m Rotor
Cut-in wind speed – 4 m/s
Cut-out wind speed – 25 m/s
Tower height – 65m
Surface friction coefficient – 0.1
Aggregated losses – 15%
34
Figure 4.4: Monthly Capacity Credit of the wind park
Figure 4.5: Monthly energy production of the wind park
35
It is evident that the capacity contribution of the wind park is very high during the
South-West monsoon period, and is quite low during the months of November,
March and April. According to the CEB data, the highest system demand is generally
seen in March and subsequently, the LOLP will be high. As a result, the overall
capacity contribution from this wind park to the system adequacy is expected to be
small due to the low capacity credit during March.
Energy production of the wind park also takes the same pattern as that of the capacity
credit, and a certain amount of thermal generation is likely to be displaced due to
contribution from wind generation when the wind park is added to the system.
Therefore, a separate evaluation was carried out in Chapter 5 to calculate the energy
credit, capacity credit and the emission credit of this modeled wind park.
4.3.2 Analysis of the wind power output model

For an accurate estimation of a wind plant output, the wind resource has to be
evaluated in detail which often requires a lot of effort. While there are several other
uncertainties, such as environmental factors, turbine performance and operational
interruptions, wind condition is the main source of uncertainty and is a major factor
that affects the power generation of a wind power plant. Therefore, every effort has
to be taken to maximize the length, quality and the geographical coverage of the data
collection, because the economic value of a wind plant is highly dependent on the
energy production, which in turn is very much sensitive to the prevailing wind
conditions. However, long term generation planning is usually conducted many years
prior to the actual implementation of a project, and resource assessments of the
proposed wind power projects are often in the initial stages where some compromise
in the predicted performance is inevitable.
In this thesis, hourly wind data of only a year was used to predict the output of the
planned wind park due to the unavailability of data. However, wind data of three
years or more is desirable to capture the annual variability of the wind resource [46].
In addition, wind data with a better resolution such as of 10 minute intervals can
provide more accurate outputs and if available, such data can be readily used with the
developed Matlab wind power output model.
36
For a small wind power plant, wind measurements taken from one metrological mast
is generally sufficient to provide an accurate assessment of the wind resource at the
site. For medium and large projects located in complex terrains or when there is
significant forestry, it is likely that more than one metrological mast is required to
evaluate the wind resource in detail. Nevertheless, this is highly site specific and for
simple terrains, measurements from one mast at the initial stages of a project can
provide reasonably accurate estimations of the wind conditions although
measurements from several masts are required when detailed evaluations are carried
out. Wind data collection by the CEB in Nadukuda beach had been carried out using
a single mast, which is common practice in the initial stages. This data was used to
calculate the wind power output of the planned 100MW wind park in Mannar, and
due to the simple terrain, it was assumed that the same wind conditions will apply to
all the individual turbines of the wind park.
Furthermore, siting of the individual turbines is carried out in such a manner that the
energy production of the wind power plant is optimized by improving independent
operation and reducing wake losses. Since such information is often unavailable and
requires complex dynamic modeling which is not essential for long term generation
planning studies, the wind power output model was kept simple by including the
wake losses in the total aggregated losses of the wind park. However, when detailed
evaluations are carried out, wind measurements are generally collected from several
locations and a more accurate output could be obtained from the model by
calculating the cluster wise wind generation after dividing the project site into a
suitable number of clusters.
The changes in input data such as wind measurement data, turbine power curve data
and tower height could be accommodated towards the latter stages of a project to
improve the output accuracy. However, since the wind power output model is
intended to be used in long term generation planning studies, finer changes such as
the layout changes of individual turbines were not included and cannot be modified
in the model. Such uncertainties were accounted in the total aggregated loss of the
project to simplify the application.
37
Chapter 5
MODELING AND EVALUATING WIND POWER PLANTS IN

WASP IV PACKAGE
5.1 Methodology of Modeling Wind Power in WASP IV
In a previous study titled “Modeling of renewable energy based embedded
generation in long term planning – A Sri Lanka case study”, a considerable effort has
been taken to model NCRE resources in WASP package, and the CEB generation
planning branch successfully uses the findings of the above study to model small
hydro and biomass resources in WASP IV. The above study suggests a Run-Of-River
ROR approach to represent a wind power plant in WASP. However, the capacity
contribution is calculated by simply estimating the average capacity according to the
monthly energy generation. The average capacity method is not generally used to
estimate the capacity contribution of intermittent NCRE plants, since it could provide
inaccurate results. Therefore, this chapter puts forth a modeling methodology for
wind power plants, which utilizes the capacity credit calculation methodology that
was proposed in Chapter 4.
There are two main feasible approaches for modeling NCRE plants in WASP IV:
1. Load Modification Approach (“Negative Load Approach”)

2. Supply-Side Approach (Wind plant is treated as a conventional power plant)
i. Run-Of-River hydro plant
ii. Unreliable thermal plant
At present, the CEB generation planning branch uses the Load Modification
Approach to account for the wind generation added to the system. In this method,
first the system demand is forecasted, and the expected wind energy generation is
then subtracted from the forecasted demand to calculate the net required power
generation from other sources. This modified demand is then used in the
optimization of the long term generation plan. However, this method does not
consider the capacity contribution, nor does it facilitate including wind power plants
as candidate options in the optimization process. In other words, the proposed wind
38
plants are considered as committed plants when deducting the expected wind
generation from the demand forecasts. Since the wind power generation is not
included in the optimization process, WASP results cannot be used for detailed
evaluations and therefore, a supply side approach is proposed in this thesis over the
load modification technique.
In the supply side approach, a wind plant could either be modeled as a ROR hydro
plant or an unreliable thermal plant. The later is simple in application and the forced
outage of the thermal unit is specified to be high to match the expected wind
generation. Fuel costs are specified as zero and the O&M costs represent the real
O&M costs of the wind power plant. Since the generating cost of this unit will be
very low, it will always be loaded when available.
Even though it is not as simple as the thermal plant approach, a better representation
of the wind plant can be achieved by the ROR method due to the following
similarities of the two types of plants:
• Both are non-dispatchable and the power has to be consumed when produced
• Both types can have seasonal variations in the resource availability
• Both resource types have a level of uncertainty
• There is no energy storage available
Seasonal variations in the resource availability can be represented by using the 12

periods in WASP by including the monthly variations in the expected generation, and
the uncertainty in the resource availability can be characterized by the five
hydrological conditions in WASP package by using the probabilities of expected
wind generation. A certain amount of variability and uncertainty are captured in this
method by specifying the expected wind generation and the available capacity by
period, according to their probabilities of occurrence [38]. Hence, this method is
preferred over the unreliable thermal plant method to model a wind power plant and
was used to model the 100MW wind park generation in WASP IV.
39
5.2 Results of the WASP Simulations
The wind generation data obtained from the Matlab wind power output model, which
was used to model the proposed 100MW wind park in WASP IV package is given
below in Table 5.1.
Table 5.1: Estimated output of the proposed wind park
Month Capacity Contribution (MW) Energy Production (GWh)
January 23.66 14.34
February 18.47 8.61
March 13.07 6.18
April 15.94 9.13
May 75.27 45.61
June 78.79 44.51
July 60.94 31.28
August 63.80 33.88
September 71.18 42.76
October 27.90 16.13
November 9.55 6.06
December 19.78 12.62
CEB generation data of other conventional generators were used in the WASP runs
in parallel with the above wind generation data to determine the least-cost path of
system expansion that would meet the forecasted system demand. The expansion
plans were prepared till the year 2030 taking into account 20 years. It was assumed
that the wind park generation will be added to the system only by year 2018, and the
plant lifetime was considered to be 25 years. The discount rate was taken as 10% for
both local and foreign costs [47]. In addition, the 35MW Broadlands and the 120MW
Uma Oya hydro power plants are considered as committed plants from 2015
onwards. Only the 27MW Moragalla and the 49MW Gin ganga projects are
considered as candidate hydro power options, which are to be available by 2018
according to the CEB data. The base load demand forecast of the CEB up to the year
2030 and the list of the thermal candidate plants used in WASP runs are given in
Appendices D and E respectively.
40
Least-cost expansion plans were prepared for the following two scenarios and the
new capacity additions of the same are given in Table 5.2:
1. Case without the wind park generation (Reference case)

2. Case with the wind park generation (Wind power case)
Table 5.2: New capacity additions of the two expansion plans

Year Reference case Wind power case
1x315 MW Puttalam Coal power plant -
2011
(Stage I)
2012 150 MW Upper Kotmale 150 MW Upper Kotmale
2013 - -
2014 1x75 MW Gas Turbine (Auto Diesel) 1x75 MW Gas Turbine (Auto Diesel)
4x5 MW Northern Power (Furnace 4x5 MW Northern Power (Furnace
Oil) Oil)
4x6 MW Chunnakum Power 4x6 MW Chunnakum Power Extension
Extension (Furnace Oil) (Furnace Oil)
2x315 MW Puttalam Coal power plant 2x315 MW Puttalam Coal power plant
2016 35 MW Broadlands 35 MW Broadlands

120 MW Uma Oya 120 MW Uma Oya
2017 2x250 MW Trinco Coal Plant 2x250 MW Trinco Coal Plant

49 MW Gin Ganga 49 MW Gin Ganga
27 MW Moragalla Hydro Plant
100 MW Wind Plant
2020 - -
2021 2x300 MW Coal power plant 2x300 MW Coal power plant
2023 1x300 MW Coal power plant -
2024 - 1x300 MW Coal power plant
PV US$ 7583.354 Million US$ 7560.356 Million
41
As per Table 5.2, majority of the new capacity additions have come from coal power
plants. The 100MW wind park and the two hydro candidates have been selected on
the same year that those plants were available for selection. The present value of the
generation plan has been reduced by US$ 23 Million and as indicated in the WASP
results, addition of more wind power would further reduce the total PV cost of the
power system. In the wind power scenario, a 300MW coal power plant which was
planned to be added in 2023 has been shifted to 2024, and another 300MW coal plant
has been shifted from 2025 to 2026. Further, even though a 250MW LNG plant was
made available for selection from 2017 onwards, it has not been selected in either of
the expansion plans.
5.3 Evaluation of the Modeled Wind Park Using WASP IV

The output of the WASP runs facilitates detailed evaluations of the economic value
of wind park, and this analysis could be carried out by comparing the results of the
reference scenario with the wind scenario. Since the economic value of wind
capacity and energy production could change over the years due to various factors
such as generation mix, the evaluation is generally carried out in terms of net present
worth analysis. In this thesis, the methodology is demonstrated using the modeled
wind park results. However, the same methodology could be applied to evaluate any
NCRE power plant.
In the wind scenario of section 5.2, the 27MW Moragalla hydro power plant was also
selected along with the wind park in the optimum solution of WASP IV. Therefore,
to isolate the effects of wind park introduction into the power system, the Moragalla
hydro power plant was added to the reference scenario in year 2018 and the reference
scenario was re-optimized. Table 5.3 summarizes the capacity additions of two
scenarios up to year the 2030 and the WASP output of new reference scenario was
compared with the wind scenario in economic assessment of the wind park.
42
Table 5.3: Capacity displacements due to the addition of wind park
Year Reference case Wind power case
1x315 MW Puttalam Coal power plant -
2011
(Stage I)
2012 150 MW Upper Kotmale 150 MW Upper Kotmale
2013 - -
4x5 MW Northern Power (Furnace 4x5 MW Northern Power (Furnace
Oil) Oil)
4x6 MW Chunnakum Power Extension 4x6 MW Chunnakum Power Extension
(Furnace Oil) (Furnace Oil)
2x315 MW Puttalam Coal power plant 2x315 MW Puttalam Coal power plant
2016 35 MW Broadlands 35 MW Broadlands

120 MW Uma Oya 120 MW Uma Oya

49 Mw Gin Ganga 49 Mw Gin Ganga
27 MW Moragalla Hydro Plant 27 MW Moragalla Hydro Plant
100 MW Wind Plant
2020 - -
2023 - -

1x35 MW Gas Turbine (Auto Diesel)


2x35 MW Gas Turbine (Auto Diesel) 1x35 MW Gas Turbine (Auto Diesel)
1x105 MW Gas Turbine (Auto Diesel) 1x105 MW Gas Turbine (Auto Diesel)
PV US$ 9008.692 Million US$ 9000.642 Million
43
It can be seen that the PV cost of the expansion plan has been reduced by US$ 8.05
Million due to the addition of the wind park, and the selection of this plant in the
least cost algorithm of WASP IV is therefore justified. When the wind park is added
to the system, a certain amount of the new generating capacity may be displaced due
to the wind generation and the saving in new investments or the capacity credit of the
wind park could be calculated by comparing the optimized expansion plans of the
two scenarios. However, it has to be noted that no new capacity will be avoided from
the expansion schedule if the capacities of candidate options are considerably higher
than the capacity contribution of the wind power plant.
5.3.1 Energy value of the wind park

The energy credit of the wind park is calculated from the differences between the
fuel costs, variable O&M costs and ENS costs of the two scenarios. Wind generation
added to the system will off-load some amount of thermal generation and
consequently, certain thermal units will produce a less amount of electricity that is
equivalent to the amount generated by the wind park. Less thermal generation will
result in savings in fuel and variable O&M costs, and these savings can be directly
attributed to the wind park in the wind power scenario. In addition to the fuel and
variable O&M costs, ENS costs are also considered to compensate for the small
differences in reliability levels of the two scenarios, since small differences are
inevitable when optimizing the two cases. It is important to consider the differences
in ENS costs because lower reliability levels can result in lower generation costs and
it should not be added to the energy credit of the wind park. This effect of different
reliability levels on the energy credit of the wind park is minimized by penalizing
lower reliability levels and crediting higher reliability levels with the ENS cost.
Koritarov et al. [48] suggests the use of the Average Incremental Costs (AIC)
method to calculate the energy credit, which is a common method frequently used by
the World Bank and other financial institutions. The sum of the present values of
differences in fuel, variable O&M and ENS costs are divided by the sum of present
values of wind park generation over the study period to calculate the Average
Incremental Energy Credit of the wind park which is represented in the following
equation.
44
(;3<5 − ;3>5 ) + (?@&B<5 − ?@&B>5 ) + (4CD<5 − 4CD>5 )
∑C
5
E (E + F)5
123 456789 =
>G5
∑C
5
E (E + F)5
Where,
AIC Energy = Average incremental energy credit of wind park over study period
($/MWh)
N = Study period (years)
i = Discount rate
FCRn = Fuel cost in year n under reference case
FCWn = Fuel cost in year n under wind power case
VO&MRn = Variable O&M costs in year n under reference case
VO&MWn = Variable O&M costs in year n under wind power case
ENSRn = ENS costs in year n under reference case
ENSWn = ENS costs in year n under wind power case
WGn = Electricity generation of wind park in year n (MWh)
Following results were obtained for the modeled wind park using the WASP results,
Present value of the sum of savings in fuel costs, variable = US$ 84.94 Million
O&M costs and ENS costs during the study period
Present value of wind energy generation during the = 1009.70 GWh

study period
Average incremental energy credit of wind generation = 84.12 US$/MWh
5.3.2 Capacity value of the wind park

The capacity benefit of the wind park can be calculated from the differences in the
two optimized expansion schedules obtained for the two scenarios. The addition of
the wind park to the system is likely to substitute a certain amount of new capacity or
delay new capacity additions. However, according to the monthly capacity
contribution results of the wind power model, the wind park in Mannar is expected to
displace only a fraction of the installed capacity since the capacity contribution
during March is only around 13MW which is reckoned to be the most critical period
of the CEB load forecasts. The average amount of displaced capacity by the wind
park during the study period was calculated by comparing the two expansion
schedules using the following equation:
45
C
E
HFIJKLM6N 3LJLMFO9 = (31<5 − 31>5 )
C
5
E
Where,
Displaced Capacity = Average capacity displaced by the wind park during
the study period (MW/year)
CARn = Capacity additions in year n under reference case
CAWn = Capacity additions in year n under wind power case
When the two expansion plans were compared, it was observed that a 35MW gas
turbine plant has been displaced in 2030 due to the addition of the 100MW wind
park. Therefore, there is an additional capacity addition of 3.25MW per year over the
20 years amounting to 65MW. This comes as no surprise since most of the new
capacity additions following the wind park are large coal power plants which are
impossible to be substituted by the wind park. Still, it can be seen that a 300MW coal
power plant has been delayed by one year in 2027 due to the addition of the wind
park.
In addition to the capacity displacements, the savings in investments could be

calculated by comparing the annualized cash flows of the two scenarios. When the
potential savings of the fixed O&M costs are also calculated by comparing the two
scenarios, it is possible to work out the capacity credit of the wind park. Koritarov et
al. [48] suggests the capacity credit of the wind power plant to be expressed in terms
of capital cost savings per MWh of the wind park generation. Again, this is an AIC
approach where the present values of the capital cost and fixed O&M cost savings
are divided by the present value of the wind park generation over the study period, as
expressed in the following equation:
(23<5 − 23>5 ) + (;@&B<5 − ;@&B>5 )

∑C
5
E (E + F)5
123 3LJLMFO9 =
>G5
∑C
5
E (E + F)5
Where,
AIC Capacity = Average incremental capacity credit of wind park over study period
($/MWh)
46
i = Discount rate
ICRn = Capital investment cost in year n under reference case
ICWn = Capital investment cost in year n under wind power case
FO&MRn = Fixed O&M costs in year n under reference case
FO&MWn = Fixed O&M costs in year n under wind power case
WGn = Electricity generation of wind park in year n (MWh)
Following results were obtained for the modeled wind park using the WASP results,
Present value of the sum of savings in capital costs and = US$ -99.82 Million
fixed O&M costs during the study period
Present value of wind energy generation during the = 1009.70 GWh

study period
Average incremental capacity credit of wind generation = -98.86 US$/MWh
The capacity credit of the wind park takes a negative value of US$ -98.86 per MWh
during the study period due to the additional investment cost of the wind park.
However, this result is not unexpected due to the low capacity contribution of the
modeled wind park during March where the system demand is generally highest.
When the AIC approach is used, both the energy and capacity credit components of
the wind park is expressed in $/MWh. This enables to calculate the total average
incremental credits of the wind park by adding the two components. Therefore, the
total wind park credits can be expressed as:
123 >F5N PL7Q ($⁄B>S) = 123 456789 + 123 3LJLMFO9
According to the calculated energy and capacity credit values, the total wind park
credits during the study period amounts to -14.74 US$/MWh which represents an
additional cost of LKR 1.63 per kWh of wind park generation. Nonetheless, only 13
years of wind generation was considered in the WASP runs and therefore this value
does not represent the economic value of the wind park but rather an assessment of
economic value during the study period. The total plant life of 25 years has to be
considered if it is required to evaluate the entire wind park benefits.
47
5.3.3 Emission benefits of the wind park
The addition of the wind generation into the system will displace a certain amount of
thermal generation which would result in a reduction of system emission levels. This
reduction can be attributed to the wind park and can be calculated from the WASP
outputs of the two scenarios in a similar manner as the energy and capacity credits.
The annual emission reductions may vary from year to year depending on the plant
mix, since the wind generation will off-load different generator types throughout the
study period. Therefore, depending on the technology and fuel type of the displaced
generation, the pollutant emissions will also be quite different and the average annual
emission reduction for a particular pollutant can be calculated as:
4UFIIFV5 <6NWMOFV5 = ∑ (4PE<5 − 4PE>5 )

E C
C 5
E
Where,
Emission Reduction = Average annual emission reduction of pollutant 1 (ton/year)
EP1Rn = Total emissions of pollutant 1 in year n under reference case
(ton)
EP1Wn = Total emissions of pollutant 1 in year n under wind power
case (ton)
CO2 was considered as the pollutant 1 in WASP runs and according to the emission
values obtained from the two scenarios, the average annual CO2 emission reduction is
222446 ton/year, which represents a reduction of 1.13% in the annual CO2 emissions.
The present value of the average annual emission reductions can be calculated by
using the following equation:
C
E (4PE<5 − 4PE>5 )
P? 4UFIIFV5 <6NWMOFV5 =
C (E + F)5
5
E
Where,
Emission Reduction = Average annual emission reduction of pollutant 1 (ton/year)
i = Discount rate
EP1Rn = Total emissions of pollutant 1 in year n under reference case
(ton)
48
EP1Wn = Total emissions of pollutant 1 in year n under wind power
case (ton)
As per the WASP results, the present value of the average annual CO2 emission
reduction was found to be 39770.5 ton/year.
The same equations were used to calculate the emission reductions in particulate
matter and the average annual emission reduction of particulate matter is 92 ton/year,
which represents a reduction of 1.18% in the annual particulate matter emissions.
The present value of the average annual emission reduction of particulate matter was
found to be 18 ton/year.
If the external cost of CO2 is known, a monetary value can also be assigned to each
ton of the reduced emissions, and the emission benefits for CO2 can be calculated
using the following equation:
4UFIIFV5 376NFO = D43PE × P? 4UFIIFV5 <6NWMOFV5
Where,
Emission Credit = Present value of average annual emission credit for pollutant
1 ($/year)
SECP1 = Estimated specific emission cost of pollutant 1 ($/ton)
When a moderate carbon price of US$ 30 per tonne of CO2 [6] is used, the emission
credit of the wind park for CO2 was found to be US$ 1.19 Million per year.
Carbon prices are not included in the generation costs of thermal units in the present
practices. However, if included, a level playing field for the NCRE technologies will
be ensured and even thermal technologies, such as natural gas may become highly
competitive with coal, which is the cheapest thermal option under the present
approaches.
49
Chapter 6
EVALUATION OF THE SRI LANKAN GENERATION MIXES

USING MEAN-VARIANCE PORTFOLIO THEORY
6.1 Least Cost Vs Portfolio Based Approaches in Generation Planning
As discussed previously in Chapter 1, long term generation planning is carried out to
identify the future capacity expansion options to meet the forecasted demand at an
acceptable level of system reliability and cost. Generally, the generation planning
horizon may range from 15 to 30 years into the future, and accurate forecasting is
quite challenging amidst the uncertainties in electricity demand, fuel prices,
investment costs, unit operation, regulatory developments, technological advances,
etc. Even though there are limitations, traditional generation planning approaches
often use least-cost methodologies that focus on developing a least-cost expansion
plan simply by adding stand-alone “least-cost” resources instead of evaluating
complete portfolios. Some critics argue that the levelised cost calculation fails to
differentiate between cost streams which have different risks [49]. This may result in
underestimating the cost of generating technologies with relatively risky future cost
streams such as fossil fuel plants and overestimating the costs of technologies with
lower risk future cost streams such as wind turbines. When it comes to generation
expansion planning, renewable technologies may present a higher levelised cost than
the conventional thermal technologies. But that does not imply that the overall cost
of the portfolio of technologies will be more costly, given the "statistical
independence of renewables costs, which do not correlate (or co-vary) with fossil
price movements" [50]. Hence, least cost methodologies are usually biased towards
conventional thermal technologies, which may result in lack of diversification in
generation expansion plans [51].
Least-cost methodologies may have worked in the past decades when there were no
high uncertainty in fuel prices, economic conditions, and low rates of technological
progress [52]. However, modern generation planning has become very challenging
due to a diverse range of resource options and a highly uncertain planning horizon.
Therefore, expanding the system by accurately identifying the least-cost alternatives
50
in today’s uncertain environment is virtually impossible, and given the uncertain
environment, it makes sense to alter traditional planning approaches from evaluating
alternative technologies to evaluating alternative generation portfolios and strategies
[53]. Even though the levelised costs of NCRE sources may be higher than
conventional technologies, the addition of more NCRE sources to a generation
portfolio which have a high share of thermal generation may not increase the overall
generation cost because even if some alternatives in a portfolio may have higher
costs at a given time, others may have lower costs. As a result, the optimal
combination of resources will minimize the overall expected generating cost relative
to the expected risk [54]. In essence, portfolio based approaches evaluate
conventional and NCRE resources not on the basis of their stand-alone costs, but on
the basis of their portfolio cost.
Least-cost planning methodologies may lead to inefficient generation portfolios that

are unnecessarily exposed to volatile fossil fuel risks when such risks are not
explicitly considered in the resource evaluations [55]. The CEB also uses a least-cost
approach when preparing the generation expansion plan which considers the stand-
alone costs of various alternatives. Therefore, this thesis proposes the Mean-Variance
Portfolio Theory as a complementary planning tool to assess portfolio risks, which
are generally ignored in least-cost methodologies.
The foundation of MVPT was laid by Nobel Laureate Harry Markowitz in 1952. In
essence, the theory states that by diversifying a portfolio of assets, the portfolio risk
can be lowered compared to the risk of the individual assets [56]. Since Markowitz’s
original work, portfolio theory is extensively used by financial investors to minimize
the risk and maximize the return of their portfolios by diversification under various
uncertain economic conditions [57]. Over the past decade, MVPT has been applied
by many authors to evaluate generation portfolios. Awerbuch and Berger [58]
performed an analysis on the European Union (EU) generation mix and the results
indicated that the existing and projected generation mixes are not optimal when the
risks are considered. Awerbuch and Yang [55] further evaluated the 2020 EU
generating mix and found that by increasing the share of non-fossil technologies such
as wind and nuclear, the cost and risk of EU generating portfolio could be reduced.
51
Other case studies of smaller generation mixes include Ireland [59], Netherlands
[60], Scotland [54], Cyprus [57] and California’s Generation mix [61]. In common,
all these studies conclude with similar results to support the development of fixed-
cost NCRE resources such as wind, and the inclusion of a CO2 cost in the evaluations
have further strengthened the results as it makes NCRE resources with zero
emissions more attractive compared to fossil fuel based generation.
6.2 Mean Variance Portfolio Theory Basics

Portfolio theory was initially used to manage financial portfolios where it relates the
expected portfolio return to the expected portfolio risk, which is defined as the year-
to-year variation of expected returns. A portfolio is generally defined as a set of
individual securities, and a perfectly diversified portfolio or an efficient portfolio
refers to ones that maximize the expected return for any given level of portfolio risk
while minimizing the portfolio risk for every level of expected returns [62].
Let P be a multi-security portfolio composed of N securities with i ∈ [1, ..., N]. The
\\\\\][ indicates the fractional weights of security i in portfolio P;
allocation vector Z
Z
Z`
\Z ^ =_ ⋮ b
\\\\]
Z
As mentioned above, the performance of a portfolio is characterized by two numbers

in the portfolio theory. The first one is the expected portfolio return and if R
represents the vector of expected returns of the individual securities;
c
c
c\] = _ ` b
⋮
c
Now, the expected portfolio return (c^ ) of a portfolio can be calculated as the sum
of expected returns of the individual securities weighted by their fractional weights;
\\\\\][ . c\]
c^ = Z c = Z
d
52
Secondly, a portfolio can be characterized by the standard deviation of its expected
returns(f[ ), which is the square root of the variance f[` . The standard deviation of
expected returns is a measure of the overall portfolio risk and is a function of not just
the individual security risks, but also the correlation of returns between each security
of the entire portfolio [63]. If the individual security returns were independent
variables, the overall portfolio risk would simply be the weighted sum of all
individual security risks. However, in reality, securities are generally dependant and
hence are correlated variables. Covariance is a measure of the linear association
between two variables and therefore, the covariance matrix of the portfolio (Covp)
which contains the covariance values between the returns of any security i with the
returns of any other security j for all i, j ∈ [1, ..., N] can be represented as;
klm ⋯ klm
Covi = j ⋮ ⋱ ⋮ p
klm ⋯ klm
Now, the variance of expected return (f[` ) of a portfolio can be calculated as;

f[` = Z Zq klm q = Z
\\\\\][ . Covi . Z
\\\\\][
d
Given that ∑ \\\\\]∗

Z = 1 and Z , … , Z ≥ 0, the optimal portfolio allocation Z[ can
be obtained according to the investor’s preferences (there will be different optimal

portfolios for different levels of risk).
The correlation coefficient (x q ) between two securities can be calculated by dividing

the covariance by the standard deviation of security i and the standard deviation of
security j. Correlation coefficient also describes the association between two
variables and a correlation coefficient of +1 corresponds to a perfect positive linear
relationship between the returns of security i and j. A correlation coefficient of -1
corresponds to a perfect negative linear relationship between the returns of two
securities and if the correlation coefficient is 0, the variables are independent.
klm q
x q =
f fq
53
The MVPT analysis is illustrated below considering a portfolio P which contains
only two assets A and B with characteristics as shown in Table 6.1. The same can be
extended to include N number of assets following the same procedure described
previously.
Table 6.1: Expected returns and standard deviation of assets A and B

Assets Returns (R) Standard Deviation (y) Correlation Factor (z{| )
A 10% 12%
-0.7
B 20% 25%
The expected return of the portfolio is a simple weighted average of the expected
returns of the individual assets.
c^ = Z c = Z} c} + Z~ c~ = Z} c} + (1 − Z} )c~

c^ = 0.1Z} + 0.2(1 − Z} )
c^ = −0.1Z} + 0.2
Since the expected returns of the two assets A and B are correlated, the variance of
the portfolio is not a simple weighted sum of the variances of the individual assets.

f[` = Z Zq klm q = Z}` f}` + Z~` f~` + 2Z} Z~ f} f~ x}~

q

f[` = 0.1189Z}` − 0.167Z} + 0.0625
It is evident that when the expected returns, standard deviations and the correlation
between assets are known, the return of the portfolio and the associated risk varies
according to the fractional weights of assets A and B. The results are presented in
Table 6.2 and the risk return curve of this two asset portfolio according to different
fractional weights are shown in Figure 6.1. The end points of this curve represent
portfolios which contain only a single asset. Inner points of the curve represent
alternative portfolios with different risk return values which correspond to different
fractional weights of the two assets.
54
Table 6.2: MVPT analysis results of two-asset portfolio
{ y y
0 0.0625 0.25 0.2
0.1 0.046989 0.216769 0.19
0.2 0.033856 0.184 0.18
0.3 0.023101 0.15199 0.17
0.4 0.014724 0.121342 0.16
0.5 0.008725 0.093408 0.15
0.6 0.005104 0.071442 0.14
0.64 0.004321 0.065738 0.136
0.68 0.003919 0.062605 0.132
0.7 0.003861 0.062137 0.13
0.72 0.003898 0.062432 0.128
0.76 0.004257 0.065243 0.124
0.8 0.004996 0.070682 0.12
0.9 0.008509 0.092244 0.11
1 0.0144 0.12 0.1
0.22
Expected Portfolio Return (RP)
0.2
B
100% Asset B
0.18
The Efficient Frontier
0.16
0.14
C
Minimum Risk
0.12 Portfolio
70% Asset A
0.1
30% Asset B
A
100% Asset A Portfolio Risk (σP)
0.08
0 0.05 0.1 0.15 0.2 0.25 0.3
Figure 6.1: Expected return and risk of two-asset portfolio
55
When constructing a portfolio, an investor may select any point on the curve
depending on the portfolio risk he is willing to take. However, for the portfolio return
to be optimal, the selected point has to be in the upper part of the curve between
points C and B. The points which lie in between points C and A cannot produce
optimal portfolios because for each value of risk along this part of the curve, a higher
portfolio return could be obtained at the same risk level by choosing a point between
C and B. Therefore, the portion of the curve between points C and B is called the
efficient frontier and it includes the points which offer the highest possible returns
for each possible value of portfolio risks.
6.3 Application of MVPT to Electricity Generation Assets

When applying the portfolio theory to electricity generating assets, the return of a
generation technology has to be defined. If the generating costs are measured in
US$/MWh, the corresponding return is simply the inverse of the generating cost,
measured in MWh/US$ [58]. Therefore, when it comes to generating assets,
maximization of the return essentially means the minimization of generating costs. In
addition, the following analogies are considered when applying MVPT to generation
planning;
• Generation technologies represent securities or assets in a portfolio

• Fractional weights represent the shares of electricity generation by different
types of generating technologies in a generation mix
• Instead of expected return, expected generation cost of each technology will
be considered
Let Cp be the expected generation cost of the generation mix, wi be the fractional
weight of the energy generated by the ith technology in the generation mix and Ci be
its expected levelised generation cost. Now we can define the expected generation
`
cost of the generation mix and variance of the expected generation cost (f[ ) as;
^ = Z

56

f[
`
= Z Zq klm q = Z Zq x q f fq

q

q

Where, f represents the technology risk of ith technology and x q is the correlation
coefficient between the generation costs of technologies i and j. Therefore, the
expected risk of a generation mix is a weighted average of the risks of individual
technology costs, tempered by their correlations or covariances. The generation cost
of a technology mainly consists of three components; investment cost, fuel cost and
operation and maintenance cost. When external costs are considered, a CO2 cost is
also included. Since each technology itself consists of a portfolio of cost
components, the risk for an individual technology is the portfolio risk for those cost
components.
When applied to generation planning, MVPT can account for correlations between
generating costs of various technologies. For example, fossil fuel prices are generally
correlated with each other and a generation mix is exposed to fuel price risks if it is
dominated by fossil fuel technologies. On the contrary, renewable technologies and
nuclear power can diversify the generation mix and reduce its expected risk because
the generation costs of renewable technologies and nuclear power are not correlated
with fossil fuel prices.
Application of MVPT in generation planning is illustrated in Figure 6.2 which shows

the portfolio diversification effect of a two technology portfolio. Suppose technology
A is a renewable technology with a higher generation cost and lower cost risk, such
as wind. It has an expected generation cost (illustrative) of around US$ 135 per
MWh and an expected generation cost risk of 7 percent. Suppose technology B is a
thermal technology with a lower generation cost but with a higher generation cost
risk, such as gas fired generation. Its generation cost is US$ 85 per MWh and the risk
is 14 percent. The correlation factor between the generation costs of the two
technologies is assumed to be zero. The variation of portfolio cost and portfolio risk
according to different fractional weights of generation of the two technologies is
shown in Figure 6.2.
57
145
Expected Portfolio Cost (Cp)
Portfolio C:
80% Asset A
135
A 20% Asset B
100% Asset A
125
C
Portfolio Y:
60% Asset A
115 40% Asset B
Y
105
The Efficient Frontier
Z
95
B
85
100% Asset B
Portfolio Risk (σp)
75
0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15
Figure 6.2: Portfolio effect for a two asset generation portfolio
It can be seen that the total portfolio risk decreases from Portfolio A to Portfolio C
when technology B is added to the portfolio even though technology B has a higher
risk level than technology A. Portfolio C is the minimum risk portfolio with a risk of
around 6.2 percent which is lower than the risk of technology A and this illustrates
the effect of diversification.
The lower part of the curve between points B and C is the efficient frontier and any
generation mix above point C is inefficient because for any point between C and A, a
superior mix could be obtained at the same risk level with a lower generation cost.
For example, portfolio Y is superior to portfolio A since it has a lower generation
cost at the same risk level of portfolio A. Further it can be seen that at point Z, the
portfolio risk is reduced by 28 percent for a generation cost increase of 15 percent
when compared with portfolio B. When only the stand-alone costs are considered,
technology A has a higher generation cost than B. However, mixes such as Y and Z
show that properly combined generation mixes can produce efficient results, and
these diversification effects or optimal mixes are not captured when generation
planning is based solely on least cost methodologies.
58
6.4 Data Required for the Application of MVPT to Generation Portfolios
The application of MVPT to a generation mix requires three data sets; the levelised
cost of generation for each technology (US$/MWh) with respective fractional
weights of cost components, the risk or standard deviation of each cost component
and the correlation coefficients between all cost components.
The levelised costs used in this thesis, except for wind and natural gas are based on
the CEB generation expansion plan of 2010. Levelised cost of generation for wind
and natural gas are based on the publication “Projected Costs of Generating
Electricity, 2010 Edition” by the International Energy Agency [64]. Since a carbon
cost is not used in the present planning approaches, the levelised costs were
calculated including and exccluding carbon costs. The cost of CO2 was considered as
the economic cost of abatement to reduce emissions and the amount was considered
to be US$ 30/tCO2 [6]. Further, it was assumed that the electricity generation of coal
was done using only steam turbines, for wind using only onshore wind plants and for
diesel and natural gas, using only Combined Cycle Gas Turbines (CCGT). Table 6.3
summarizes the levelised costs with and without the inclusion of CO2 costs.
Table 6.3: Levelised generating costs (US$/MWh)

With the inclusion of CO2 Costs
Capital Fuel O&M CO2 Total
Coal 25.63 38.15 4.14 25.37 93.29
Combined Cycle - Diesel 17.60 159.11 5.05 17.22 198.98
Natural Gas (LNG) 15.97 61.12 4.48 10.54 92.11
Hydro 95.73 0.00 1.87 0.00 97.60
Wind 115.24 0.00 21.92 0.00 137.16
Biomass (Dendro) 30.51 100.04 6.37 0 136.92
Without the inclusion of CO2 Costs
Capital Fuel O&M CO2 Total
Coal 25.63 38.15 4.14 0.00 67.92
Combined Cycle - Diesel 17.60 159.11 5.05 0.00 181.76
Natural Gas (LNG) 15.97 61.12 4.48 0.00 81.57
Hydro 95.73 0.00 1.87 0.00 97.60
Wind 115.24 0.00 21.92 0.00 137.16
Biomass (Dendro) 30.51 100.04 6.37 0 136.92
59
The standard deviation values of capital, fuel, O&M and CO2 costs for each
technology were obtained from Awerbuch and Yang (2007) and are given in
Appendix F [55].
The correlation coefficients of fuel, O&M and CO2 costs between different
technologies were obtained from [55] and are shown in Appendix G. According to
Awerbuch and Berger (2003), the correlation between the capital costs of any two
thermal technologies is assumed to be high, and a coefficient of 0.7 was used. A
value of 0.1 was used for the correlation between the capital costs of thermal and
renewable technologies. The correlation between capital and O&M cost is assumed
to be small and thus a value of 0.1 was used. Further, it was assumed that the
correlations between O&M and fuel, capital and fuel, O&M and CO2, and fuel and
CO2 costs are zero for each pair of technologies [58].
6.5 Evaluation of the Sri Lankan Generation Mixes of 2012 and 2025
The preceding section described the generating cost and risk values for different
technologies used in the evaluation. Since a CO2 cost is not considered in the present
planning approaches, two scenarios were evaluated separately with and without the
inclusion of CO2 costs.
Theoretically, any single technology could potentially generate the total electricity
demand of the generation mix. However, in practice every technology has certain
technical and economical limitations which must be recognized and included in the
MVPT model in order to develop realistic portfolios. For example, renewable
technologies may be limited in resource availability and may have certain practical
difficulties when it comes to project developments. While conventional technologies
such as coal, oil and natural gas may not have upper bounds due to resource
limitations, upper bounds may still be required to account for lead times to construct
new power plants. Lower bounds may be applied in both renewable and conventional
technologies to prevent the model from retiring existing power plants which are not
scheduled to be retired by the target year. In this thesis, maximum wind and biomass
power generation is therefore limited to 10% considering the resource limitations and
construction lead times. Since majority of the hydro power resources are already
60
developed, the maximum hydro power share is limited to 30% and the minimum
hydro power share is limited to 18% to prevent the retiring of already developed
hydro power plants. Accordingly, the maximum coal power share is limited to 81.5%
and the minimum coal power share is 6.5%. The maximum oil and natural gas share
is therefore limited to 75%.
Table 6.4 shows the target generation mixes of 2012 and 2025 as per the CEB
generation expansion plan of 2010 and Table 6.5 shows the lower and upper bounds
of alternative technologies applied in MVPT analysis.
Table 6.4: Target generation mixes of 2012 and 2025

Technology Expected Generation (GWh) Energy Share (%)
2012 2025 2012 2025
Coal 1821 19547 14.61 70.95
Oil 4944 306 39.67 1.11
Gas (LNG) 0 0 0.00 0.00
Hydro 5456 6878 43.78 24.96
Wind 140 378 1.12 1.37
Biomass (Dendro) 102 443 0.82 1.61
Source: CEB Generation Expansion Plan 2010
Table 6.5: Lower and upper bounds for the alternative technologies
Technology Lower bound (%) Upper bound (%)
Coal 6.5% 81.5%
Oil 0% 75%
Gas (LNG) 0% 75%
Hydro 18% 30%
Wind 0.5% 10%
Biomass (Dendro) 0% 10%
Figure 6.3 illustrates the generating costs per MWh and risks associated with each of
the generating technologies with a CO2 cost of 30$ per tonne while Figure 6.4 shows
the individual cost and risk values without the inclusion of a CO2 cost. For
comparison, the cost-risk combinations of target Sri Lankan generation mixes of
2012 and 2025 are also shown in the two figures.
61
200
Oil
180
Generating Cost ($/MWh)
160
140
Wind Biomass
120
CEB 2012
100
CEB 2025 Coal Hydro

Gas
80
4% 6% 8% 10% 12% 14% 16% 18% 20% 22%
Risk: year-to-year standard deviation
Figure 6.3: Costs and risks of generating technologies (with CO2 costs)
180
Oil
160
140
Wind Biomass
o
120
CEB 2012
100
Hydro
80
Gas
CEB 2025
Coal
60
4% 6% 8% 10% 12% 14% 16% 18% 20% 22%
Figure 6.4: Costs and risks of generating technologies (without CO2 costs)
62
Figure 6.4 shows that compared to the 2012 generation mix, the 2025 mix’s
electricity generating cost has reduced from 111.60 US$ to 75.29 US$/MWh. In
addition, the expected risk of the generation mix is also reduced from 10.32 percent
to 8.88 percent. This is attained by retiring most of the costly and risky oil power
generation and by increasing the share of coal based generation.
As shown in Figure 6.3, the inclusion of a CO2 cost has increased the overall cost of
generation due to the increase in cost of fossil fuel generation. The generation cost of
2025 mix has been increased by 21.43 percent due to the high share of coal based
generation. However, the generation cost of the 2012 mix has been increased by only
8.64 percent due to the high share of hydro power generation. Nevertheless, the 2025
mix shows a lower cost of 95.82 US$/MWh, and a risk of 7.12 percent compared to
the cost of 122.14 US$/MWh and risk of 9.77 percent of the 2012 generation mix.
Furthermore, it has to be noted that with the inclusion of a CO2 cost of 30 US$ per
tonne, the generation cost of coal power generation increases by 27.17 percent from
67.92 US$/MWh to 93.29 US$/MWh, while the cost of gas based generation
increases by only 11.44 percent from 81.57 US$/MWh to 92.11 US$/MWh. This
made the cost of gas based generation highly competitive with the cost of coal power
generation, and it is evident that higher CO2 costs would make the cost of gas based
generation considerably cheaper than that of coal based generation.
6.5.1 Results and Discussion

A MVPT analysis of the CEB generation mixes was carried out to evaluate the
position of the target CEB generation mixes against the Efficient Frontier (EF). Two
scenarios were analyzed with and without the inclusion of CO2 costs, and further, the
possibility of reducing the portfolio cost or the risk of the target CEB mixes was
examined.
Figures 6.5 and 6.6 illustrates the efficient frontier and the position of the target CEB
mixes for scenario one which includes CO2 costs. Figures 6.7 and 6.8 represent the
second scenario which does not include CO2 costs. The blue curves in all four figures
represent the efficient frontier or in other words, the locus of all optimal generation
mixes. Although there can be an infinite number of generation mixes which could
63
meet the 2025 electricity demand, the optimal mixes lie on the EF. There are no
feasible mixes below the EF and the mixes which lie above the EF are sub-optimal or
inefficient because it is possible to further reduce the cost or risk of those generation
mixes by finding mixes on the EF either by moving below or to the left.
In the evaluation of the CEB generation mixes, four typical optimal mixes were
selected along the EF for easier comparison;
Mix A1/A2: Least risk/highest cost optimal portfolio, usually the most diverse
Mix B1/B2: Same cost as the CEB mix, but with a lower risk
Mix C1/C2: Same risk as the CEB mix, but with a lower cost
Mix D1/D2: Least cost/highest risk optimal portfolio, usually the least diverse
6.5.1.1 Scenario one (With CO2 costs)

The efficient frontier, costs and risks of each mix and the respective weights of each
technology, along the EF were obtained using a Matlab model given in Appendix H
subjected to the lower and upper technology bounds as mentioned above in Table
6.5.
103
A1
102
101
100
99
98
97
96
B1 CEB 2025
95
C1
94
93
D1
92
5% 6% 7% 8% 9% 10% 11% 12%
Figure 6.5: Efficient frontier for the CEB Generation mix - Scenario 1
64
125
CEB 2012
120
115
110
105
A1
100
95 B1 CEB 2025
C1
E1 D1
90
5% 6% 7% 8% 9% 10% 11% 12%
Figure 6.6: CEB generation mixes of 2012 and 2025 against EF - Scenario 1
Figure 6.5 displays the EF for the Sri Lankan generation mix. It starts at A1 with the
minimum-risk portfolio and while moving forward along the EF, the model
minimizes the generation cost of the mix and takes more risk up to point D1 which
represents the minimum-cost optimal portfolio.
Table 6.6: Technology shares of optimal portfolios – Scenario 1

Cost Risk Coal Oil Gas Hydro Wind Biomass
$/MWh % Share in electricity generation (%)
CEB 2025 95.82 7.12 70.95 1.11 0.00 24.96 1.37 1.61
CEB 2012 122.15 9.77 14.61 39.67 0.00 43.78 1.12 0.82
Mix A1 102.56 5.54 40.82 2.66 6.53 30.00 10.00 10.00
Mix B1 95.82 6.71 56.28 0.00 9.21 30.00 0.50 4.01
Mix C1 94.60 7.12 60.62 0.00 8.83 30.00 0.50 0.04
Mix D1 93.28 11.75 6.50 0.00 75.00 18.00 0.50 0.00
Mix E1 93.54 9.77 28.11 0.00 53.39 18.00 0.50 0.00
65
As shown in Table 6.6, the CEB 2025 mix has an overall generating cost of 95.82
US$/MWh and a risk of 7.12 percent. Mix B1 is an equal cost/low risk generating
mix which reduces the risk up to 6.71 percent. On the other hand, mix C1 is an equal
risk/low cost mix which reduces the generating cost up to 94.60 US$/MWh.
Mix A1 is the minimum-risk portfolio which reduces the risk from 7.12 percent to
5.54 percent, and it happens to be the most diversified mix with the highest share of
renewables where all hydro, wind and biomass has hit the upper limits of respective
energy shares. But this comes at a cost increase of 6.57 percent which is unattractive
when compared with mix B1. Similarly, mix D1 represents the minimum-cost mix. It
reduces the cost by 2.58 US$/MWh, but this cost reduction comes with a
considerable increase in portfolio risk from 7.12 percent to 11.75 percent, which is
almost a 40 percent increase. Therefore, it can be seen that the practical range of
generation mixes lie between mixes B1 and C1. In addition, it is interesting to note
that both the mixes; B1 and C1, have reduced shares of coal power generation, no oil
based generation and a considerable amount of natural gas generation.
Even though the CEB 2025 mix can be further fine tuned, it is generally close to the
EF and shows a considerable improvement when compared with the CEB 2012 mix.
As per Figure 6.6, CEB 2012 mix is noticeably away from the EF and has an overall
generating cost of 122.15 US$/MWh and a risk of 9.77 percent. The CEB 2025 mix
shows a cost reduction of 21.55 percent and a risk reduction of 27.12 percent when
compared with the 2012 mix. Further, mix E1 shows that the generating cost of the
CEB 2012 mix can be reduced by 23.42 percent while maintaining the same risk
level. This involves a high share of gas generation, but this will not hold true if the
CO2 costs are not considered since the removal of CO2 costs makes the coal
generation considerably cheaper than natural gas generation. However, if the climate
change mitigation is also an objective of long term generation planning, the carbon
costs and emission reductions should also be considered along with the costs and
risks of the generating mixes. Since the present generation planning approaches does
not consider a CO2 cost, the Sri Lankan generation mixes were evaluated without the
inclusion of CO2 costs for comparison purposes and the results are presented in the
following section.
66
6.5.1.2 Scenario two (Without CO2 costs)
90
A2
88
86
84
82
80
78
76 CEB 2025
B2
74 C2
72
D2
70
6% 6.5% 7% 7.5% 8% 8.5% 9% 9.5% 10%
Figure 6.7: Efficient frontier for the CEB Generation mix - Scenario 2
115
CEB 2012
110
105
100
95
90
A2
85
80
CEB 2025
75
B2 C2 D2
70
6% 6.5% 7% 7.5% 8% 8.5% 9% 9.5% 10% 10.5%
Figure 6.8: CEB generation mixes of 2012 and 2025 against EF - Scenario 2
67
Figure 6.7 displays the EF of the Sri Lankan generation mix and the four selected
generation mixes as discussed in the previous section. Figure 6.8 shows the position
of the two target CEB mixes compared to the EF and Table 6.7 summarizes the
energy shares of each technology for the six portfolios shown in Figures 6.7 and 6.8.
Table 6.7: Technology shares of optimal portfolios – Scenario 2

Cost Risk Coal Oil Gas Hydro Wind Biomass
$/MWh % Share in electricity generation (%)
CEB 2025 75.29 8.88 70.95 1.11 0.00 24.96 1.37 1.61
CEB 2012 111.60 10.32 14.61 39.67 0.00 43.78 1.12 0.82
Mix A2 89.16 6.13 30.23 2.85 16.93 30.00 10.00 10.00
Mix B2 75.29 8.69 67.99 0.00 6.27 21.19 0.50 4.06
Mix C2 74.73 8.88 69.84 0.00 5.75 20.80 0.50 3.11
Mix D2 72.05 9.85 81.50 0.00 0.00 18.00 0.50 0.00
When the CO2 costs are not included, the overall generating cost of the CEB 2025
mix amounts to 75.29 US$/MWh and has a risk of 8.88 percent. Mix A2 represents
the minimum-risk portfolio with a risk of 6.13 percent and a cost of 89.16
US$/MWh, which is a cost increase of 15.55 percent from the CEB 2025 mix.
Similar to scenario one, A2 is the most diversified generation mix and has the highest
share of renewable energy generation with all hydro, wind and biomass shares at
upper limits. Mix D2 represents the minimum-cost portfolio with a generation cost of
72.05 US$/MWh and a risk of 9.85 percent. Here, the cost is reduced by 4.30 percent
while the risk is increased by 9.84 percent and not unexpectedly, this mix is the least
diversified mix with the highest share of coal generation. Even though mix D2 seems
attractive when the cost-risk trade-off is considered, practically it may be difficult to
have this combination due to operational constraints.
Mix B2 is an equal cost/low risk generating mix which reduces the risk to 8.69
percent with a risk reduction of only 2.14 percent. Mix C2 represents an equal
risk/low cost mix which reduces the generating cost to 74.73 US$/MWh with a cost
reduction of 0.74 percent. Therefore, it is evident that when CO2 costs are not taken
into account, the CEB 2025 mix comes very close to the EF, and further it should be
68
noted that both B2 and C2 have a substantial share of generation which comes from
natural gas and similar to scenario one, there is no oil based generation in either of
the two mixes.
Though the CEB 2025 mix lies close to the EF, the CEB 2012 mix on the other hand,
lies clearly away from the EF with a generating cost of 111.60 US$/MWh and a risk
of 10.32 percent. This shows that the 2012 mix is highly inefficient in cost-risk terms
where significant cost reductions are possible with different generation mixes. For
example, the CEB 2025 mix shows a cost reduction of 32.54 percent and a risk
reduction of 13.95 percent when compared to the 2012 mix.
Oil based generation has been largely reduced or have even been removed from the
generation mixes in both the scenarios when the portfolios move close to the EF. In
addition, although the MVPT analysis indicates a sizeable amount of natural gas
generation in optimal portfolios, the least-cost approaches would almost never select
natural gas over coal generation if CO2 costs are not included due to the higher stand-
alone costs of natural gas generation. Nevertheless, the portfolio approach has shown
that the addition of natural gas and renewable resource based generation to the mix is
certainly beneficial due to the cost stability it brings without increasing the overall
generation cost unnecessarily. Therefore, based on the above discussion, it is
apparent that least-cost based generation planning might possibly lead to
economically inefficient portfolios.
In brief, the portfolio analysis presented in this chapter does not produce a specific
capacity expansion plan. Nonetheless, it is capable of identifying the technologies
that efficiently diversify the generation mix while increasing the energy security and
thus, is significantly more detailed and robust than arbitrarily mixing different
technologies. Therefore, MVPT is limited to providing a direction for generation
planners which when utilized, would enable them to evaluate portfolio costs and
risks, and could be used as a complementary planning tool alongside present
practices to evaluate different technology alternatives and generation mixes.
69
Chapter 7
SUMMARY AND CONCLUSION

Since generation planning plays a major role in a country’s efforts of mainstreaming
NCRE developments, the long term generation planning methodologies adhered and
utilized by Sri Lanka were examined in this thesis, and several conceptual
modifications were proposed to successfully integrate and evaluate NCRE resources
in the country’s present planning process. In addition, the proposed concepts were
applied to the Sri Lankan system to evaluate their applicability, and the results were
analyzed and discussed in each chapter.
7.1 Summary and Key Findings

Chapter 1 provided an overview of long term generation planning, highlighted the
importance of mainstreaming NCRE resources of a country and identified the
complexities associated with NCRE planning. The present status of the NCRE sector
in Sri Lanka and relevant policy objectives and targets were also discussed.
In Chapter 2, the constraints to mainstream adoption of NCRE technologies in Sri

Lanka were discussed along with related policy and regulatory interventions. The
establishment of SLSEA, introduction of technology specific, cost based NCRE tariff
and provision of long term SPPAs were recognized as the main regulatory
interventions that contributed to the development of NCRE resources in Sri Lanka.
Even though Sri Lanka has taken several constructive measures to develop NCRE
resources in the country, a clear lack of technology and performance standards were
identified, and the need to establish a proper framework to formulate the relevant
standards were highlighted.
The CEB’s generation planning process was reviewed and issues related to NCRE
planning were discussed in Chapter 3. Since the planning process is least-cost based,
risks associated with different technologies are not recognized when alternative
technologies are compared. Capacity contribution of intermittent NCRE sources such
as wind is not calculated, and such plants are not considered as candidate options.
The benefits of NCRE plants are difficult to be evaluated in detail when those are not
70
treated and included as a conventional power plant in WASP IV. These issues were
addressed conceptually in Chapter 3 and the proposed concepts were applied to the
Sri Lankan system in the subsequent chapters.
The methods of calculating capacity credit of NCRE generators were discussed in

Chapter 4. Several recommended methods in published literature such as ELCC,
peak period capacity factor, multi state unit representation and capacity factor during
high LOLP hours were evaluated, and the peak period capacity factor method was
found to be the most suitable and effective method to be applied to the Sri Lankan
context, when the characteristics of the system load profile of Sri Lanka, input data
requirement of the model and accuracy of the prediction is considered. The function
and structure of the wind power output model was explained and model limitations
were also identified.
The modeling and evaluation of wind power plants in WASP package was described
in Chapter 5. Approaches of modeling NCRE plants in WASP were discussed and
the shortcomings of the present practices were identified. Subsequently, a ROR
method was proposed and two generation plans were optimized in WASP with and
without the hypothetical wind park. The modeled wind park was selected by the
WASP optimization tool on the same year it was available for selection, and the
WASP results indicated further cost reductions with addition of more wind power. A
methodology to evaluate the modeled wind park was also presented in Chapter 5.
Chapter 6 described the application of MVPT as a planning tool to assess portfolio

risks and evaluated the Sri Lankan generation mixes of 2012 and 2025. Position of
the CEB mixes was evaluated against the efficient frontier in two cases with and
without the inclusion of CO2 costs, and the possibility of reducing the portfolio cost
and the risk was also examined. Results showed that the CEB 2025 mix comes
considerably close to the EF, and the CEB 2012 mix was noticeably away in both the
cases, which is an indication of the inefficiency of the 2012 mix. For example, when
the CO2 costs were not considered, the 2025 mix showed a cost reduction of 32.54
percent and a risk reduction of 13.95 percent when compared to the 2012 mix.
71
With the inclusion of a CO2 cost of 30 US$ per tonne, the generation cost of coal
power generation increased by 27.17 percent, while the cost of gas based generation
increased only by 11.44 percent. This made the cost of gas generation highly
competitive with the cost of coal generation, and it was evident that higher CO2 costs
would make the cost of gas generation considerably cheaper than that of coal power
generation.
Furthermore, it was noted that oil based generation has been largely reduced or have
even been removed from the generation mixes in both scenarios when portfolios
move close to the EF. Even though the MVPT evaluation indicated a sizeable
amount of natural gas generation in optimal portfolios, using least-cost approaches
would almost never select natural gas over coal generation due to the differences in
stand-alone costs. Nevertheless, the portfolio approach has shown that the addition of
natural gas and NCRE based generation to the mix is certainly desirable due to the
portfolio risk reduction it brings without increasing the overall generation costs
unnecessarily.
7.2 Recommendations and Contributions of the Thesis

This research was carried out to examine the ways to mainstream NCRE
technologies in the traditional planning efforts, and the main recommendations and
contributions of this thesis are summarized below;
• There is a clear lack of NCRE technology and performance standards in Sri

Lanka, and establishment of a proper framework to formulate relevant
standards is essential.
• For long term capacity expansion planning, the use of peak period capacity
factor method is recommended to calculate the capacity credit of NCRE
generation in Sri Lanka.
• A ROR hydro equivalent model is proposed to model wind power plants in
WASP IV package.
• An Average Incremental Cost method is recommended to evaluate the energy
and capacity cost savings from NCRE power plants.
72
• Target generation mixes shall be evaluated using MVPT to assess the
generation portfolio risks and to capture the inherent risks in different
alternative technologies.
In this research, a wind power output model was developed using Matlab, and the
model was tested using hourly wind data of 12 months. However, wind data of
several years with a better resolution such as that of 10 or 30 minute intervals would
provide a better output prediction. For medium and large projects in complex
terrains, wind measurements collected from several metrological masts shall be used
once available and the project site shall be clustered appropriately to obtain a better
estimation of the project output.
Another Matlab model was developed in this research for portfolio risk evaluation of
the Sri Lankan generation mixes using MVPT. At present, the risks are evaluated via
sensitivity studies and the MVPT analysis presents a new perspective to the risk
assessment of the generation portfolios. The developed model can be readily used
alongside the present practices, and could be regarded as a significant contribution of
this thesis.
Naturally, the output consistency of the two models depends on the accuracy of the
input data and thus, periodic revision of the model inputs is highly recommended.
7.3 Scope for Future Work

The focus of this research was on integrating and evaluating small hydro power,
wind and biomass power which are considered to be the three main NCRE sources in
Sri Lanka, and the present work can be further extended to represent other NCRE
sources such as photovoltaic and geothermal power as well.
In the wind power output model used in this thesis, the outputs from individual units
are aggregated into an equivalent wind capacity that is equal to the sum of all
individual outputs. However, in reality, wind incident will not be equal for all the
turbines at a given time since the turbines will be spread over a large geographical
area. Therefore, dynamic modeling and detailed analysis could be carried out as
future work to aggregate the individual turbine outputs more accurately.
73
The impacts of intermittency will have to be evaluated as NCRE penetration
increases, and for successful integration, scenarios such as back up generation and
energy storage techniques will have to be studied in depth. It is believed that energy
storage techniques such as pumped hydro and hydrogen storage will be instrumental
in dampening the intermittency effects of various NCRE sources. Therefore, further
developments in modeling with the inclusion of the above storage techniques will
provide a better understanding on the NCRE integration concerns.
Transmission planning was not included in the present work due to time constraints.
Nevertheless, good transmission infrastructure is vital for successful integration of
NCRE generation and could be added in future work.
7.4 Conclusion
In essence, this research presented methodologies to model a wind power plant in
WASP IV and to evaluate the benefits of the modeled NCRE plants in the present
generation planning process of Sri Lanka. Furthermore, this thesis provided a new
insight into the capacity contribution of NCRE plants and also discussed policy and
regulatory interventions in Sri Lanka in relation to NCRE developments.
Even though the application of the proposed concepts are mostly done on a wind
power plant, the methodologies used for the calculation of capacity credit and for the
economic evaluation of the modeled wind park are broadly applicable to other NCRE
resource types as well.
In conclusion, it is expected that the methodologies, models, results and discussions

presented in this thesis would provide valuable resources to the generation planners
and would help in successful development and grid integration of NCRE resources in
Sri Lanka.
74
APPENDICES
75
APPENDIX A
POWER CURVE VALUES OF THE REFERENCE TURBINE
Turbine VESTAS V52 – 850kW

Rotor Diameter 52m
Cut-in wind speed 4 m/s
Nominal wind speed 16 m/s
Stop wind speed 25 m/s
Area swept 2,124 m2
Wind Speed (m/s) Power (kW)

1 0
2 0
3 0
4 25.5
5 67.4
6 125
7 203
8 304
9 425
10 554
11 671
12 759
13 811
14 836
15 846
16 849
17 850
18 850
19 850
20 850
21 850
22 850
23 850
24 850
25 850
76
VESTAS V52 - 850kW Turbine Power Curve
900
800
700
600
Power (kW)
500
400
300
200
100
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Wind Speed (m/s)
77
APPENDIX B
CURVE FITTING OF REFERENCE TURBINE POWER CURVE
Curve fitting of turbine power curve (Variable part)
turbine_out_var vs. turbine_spd_var

800
Curve fit
700
600
Linear model Poly7:

500 f(x) = p1*x 7 + p2*x 6 + p3*x 5 + p4*x 4 + p5*x 3 +
Power (kW)
p6*x 2 + p7*x + p8
Coefficients (with 95% confidence bounds):
400 p1 = -0.0007301 (-0.0008532, -0.000607)
p2 = 0.05024 (0.04162, 0.05886)
p3 = -1.407 (-1.658, -1.155)
p4 = 20.65 (16.69, 24.6)
300 p5 = -171.6 (-207.6, -135.5)
p6 = 819.8 (629.8, 1010)
p7 = -2066 (-2601, -1531)
200 p8 = 2117 (1497, 2736)
Goodness of fit:
SSE: 1.315
100 R-square: 1
Adjusted R-square: 1
RMSE: 0.5129
4 6 8 10 12 14 16
Wind Speed (m/s)
78
APPENDIX C
MATLAB CODE OF WIND POWER OUTPUT MODEL

%==========Wind Power Output Model==================================
clear
data = xlsread('man.xls');
dates = data(:, 1);
%times = data(:,2);
windspd_ms = data(:, 2);
filtered_spds = windspd_ms(windspd_ms>=0);
filtered_dates = dates(windspd_ms>=0);
%filtered_times = times(windspd_ms>=0);
initial_spds = filtered_spds;
twr_height_corrected_spds = initial_spds * ((65/40)^ 0.1);% tower
height correction, roughness is assumed to be 0.1
spd_ms = twr_height_corrected_spds;
%===================================================================
%Month and Hour selection from the wind speed measurement time stamp
date_string = num2str(filtered_dates);
%time_string = num2str(filtered_times);
%dt_string = strcat(date_string,time_string);
hour_string = date_string(:,9:10); % peak hour selection
month_string = date_string(:,5:6); % month selection for monthly
capacity factor calculation
for i = 1:length(hour_string)
hour_double(i) = str2double(hour_string(i,:));
end
for k = 1:length(month_string)
month_double(k) = str2double(month_string(k,:));
end
%===================================================================
%Peak wind value selection from the valid wind speed data set
peak_wind = spd_ms(hour_double>=18 & hour_double <=22); % Wind

speeds during peak times
%hist(peak_wind, 18), xlabel('Wind Speed m/s'),ylabel('Frequency');
peak_dates = filtered_dates(hour_double>=18 & hour_double <=22);
%Peak wind speeds in January

peak_wind_jan = spd_ms(month_double==01 & hour_double>=18 &
hour_double <=22);
peak_dates_jan = filtered_dates(month_double==01 & hour_double>=18 &
hour_double <=22);
79
%Peak wind speeds in February
peak_wind_feb = spd_ms(month_double==02 & hour_double>=18 &
hour_double <=22);
peak_dates_feb = filtered_dates(month_double==02 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in March
peak_wind_mar = spd_ms(month_double==03 & hour_double>=18 &
hour_double <=22);
peak_dates_mar = filtered_dates(month_double==03 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in April
peak_wind_apr = spd_ms(month_double==04 & hour_double>=18 &
hour_double <=22);
peak_dates_apr = filtered_dates(month_double==04 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in May
peak_wind_may = spd_ms(month_double==05 & hour_double>=18 &
hour_double <=22);
peak_dates_may = filtered_dates(month_double==05 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in June
peak_wind_june = spd_ms(month_double==06 & hour_double>=18 &
hour_double <=22);
peak_dates_june = filtered_dates(month_double==06 & hour_double>=18
& hour_double <=22);
%Peak wind speeds in July
peak_wind_july = spd_ms(month_double==07 & hour_double>=18 &
hour_double <=22);
peak_dates_july = filtered_dates(month_double==07 & hour_double>=18
& hour_double <=22);
%Peak wind speeds in August
peak_wind_aug = spd_ms(month_double==08 & hour_double>=18 &
hour_double <=22);
peak_dates_aug = filtered_dates(month_double==08 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in September
peak_wind_sep = spd_ms(month_double==09 & hour_double>=18 &
hour_double <=22);
peak_dates_sep = filtered_dates(month_double==09 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in October
peak_wind_oct = spd_ms(month_double==10 & hour_double>=18 &
hour_double <=22);
peak_dates_oct = filtered_dates(month_double==10 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in November
peak_wind_nov = spd_ms(month_double==11 & hour_double>=18 &
hour_double <=22);
peak_dates_nov = filtered_dates(month_double==11 & hour_double>=18 &
hour_double <=22);
%Peak wind speeds in December
peak_wind_dec = spd_ms(month_double==12 & hour_double>=18 &
hour_double <=22);
peak_dates_dec = filtered_dates(month_double==12 & hour_double>=18 &
hour_double <=22);
%===================================================================
80
%Wind turbine powe curve data of Vestas V52 850kW turbine
turbine_pc = xlsread ('vestasv52_850.xls'); %W.Tur power curve data
turbine_spd = turbine_pc (:,1);
turbine_out = turbine_pc (:,2);
turbine_spd_var = turbine_pc (4:16,1);%variable part of power curve
turbine_out_var = turbine_pc (4:16,2);
%scatter(turbine_spd,turbine_pc(:,2)); %Scatter plot of power curve
%===================================================================
%Discrete points of the power curve is fitted to a polynomial curve
to determine the variable power output of the turbine (When wind spd
is between Vcut-in and Vrated)
coeff = polyfit(turbine_spd_var,turbine_out_var,7);%coefficients of
the polynomial fit of turbine power curve
%scatter(turbine_spd_var,turbine_out_var);
p1 = coeff(1,1);
p2 = coeff(1,2);
p3 = coeff(1,3);
p4 = coeff(1,4);
p5 = coeff(1,5);
p6 = coeff(1,6);
p7 = coeff(1,7);
p8 = coeff(1,8);%These coefficients are used in the var_coeff Func.
%===================================================================
%Wind Power calculation during peak hours (Annual)_Wind Park
for j=1:length(peak_wind)
peak_hourly_wind_power (j,:) = power_func(peak_wind(j));%Wind power
during peak hours in MW
end
%Monthly Wind power calculation during peak hours

for jan=1:length(peak_wind_jan) %January
peak_hourly_wind_power_jan (jan,:) =
power_func(peak_wind_jan(jan));%Wind Pwr during peak hours_MW_Jan
end
for feb=1:length(peak_wind_feb) %February
peak_hourly_wind_power_feb (feb,:) =
power_func(peak_wind_feb(feb));%Wind Pwr during peak hours_MW_Feb
end
for mar=1:length(peak_wind_mar) %March
peak_hourly_wind_power_mar (mar,:) =
power_func(peak_wind_mar(mar));% Wind Pwr during peak hours_MW_Mar
end
for apr=1:length(peak_wind_apr) %April
peak_hourly_wind_power_apr (apr,:) =
power_func(peak_wind_apr(apr));% Wind Pwr during peak hours_MW_Apr
end
for may=1:length(peak_wind_may) %May
peak_hourly_wind_power_may (may,:) =
power_func(peak_wind_may(may));% Wind Pwr during peak hours_MW_May
end
for june=1:length(peak_wind_june) %June
peak_hourly_wind_power_june (june,:) =
power_func(peak_wind_june(june));% Wind Pwr during peak hours_MW_Jun
end
for july=1:length(peak_wind_july) %July
81
peak_hourly_wind_power_july (july,:) =
power_func(peak_wind_july(july));% Wind Pwr during peak hours_MW_Jul
end
for aug=1:length(peak_wind_aug) %August
peak_hourly_wind_power_aug (aug,:) =
power_func(peak_wind_aug(aug));% Wind Pwr during peak hours_MW_Aug
end
for sep=1:length(peak_wind_sep) %September
peak_hourly_wind_power_sep (sep,:) =
power_func(peak_wind_sep(sep));% Wind Pwr during peak hours_MW_Sep
end
for oct=1:length(peak_wind_oct) %October
peak_hourly_wind_power_oct (oct,:) =
power_func(peak_wind_oct(oct));% Wind Pwr during peak hours_MW_Oct
end
for nov=1:length(peak_wind_nov) %November
peak_hourly_wind_power_nov (nov,:) =
power_func(peak_wind_nov(nov));% Wind Pwr during peak hours_MW_Nov
end
for dec=1:length(peak_wind_dec) %December
peak_hourly_wind_power_dec (dec,:) =
power_func(peak_wind_dec(dec));% Wind Pwr during peak hours_MW_Dec
end
%===================================================================
%Peak Period Capacity Factor calculation (Capacity Credit)
annual_avg_peak_wind_power = mean(peak_hourly_wind_power);
annual_cap_factor = (annual_avg_peak_wind_power/100.3)*100;
%Monthly Peak Period Capacity Factor calculation

avg_peak_wind_power_jan = mean(peak_hourly_wind_power_jan); %Jan
cap_factor_jan = (avg_peak_wind_power_jan/100.3)*100;
avg_peak_wind_power_feb = mean(peak_hourly_wind_power_feb); %Feb
cap_factor_feb = (avg_peak_wind_power_feb/100.3)*100;
avg_peak_wind_power_mar = mean(peak_hourly_wind_power_mar); %Mar
cap_factor_mar = (avg_peak_wind_power_mar/100.3)*100;
avg_peak_wind_power_apr = mean(peak_hourly_wind_power_apr); %Apr
cap_factor_apr = (avg_peak_wind_power_apr/100.3)*100;
avg_peak_wind_power_may = mean(peak_hourly_wind_power_may); %May
cap_factor_may = (avg_peak_wind_power_may/100.3)*100;
avg_peak_wind_power_june = mean(peak_hourly_wind_power_june); %Jun
cap_factor_june = (avg_peak_wind_power_june/100.3)*100;
avg_peak_wind_power_july = mean(peak_hourly_wind_power_july); %Jul
cap_factor_july = (avg_peak_wind_power_july/100.3)*100;
avg_peak_wind_power_aug = mean(peak_hourly_wind_power_aug); %Aug
cap_factor_aug = (avg_peak_wind_power_aug/100.3)*100;
avg_peak_wind_power_sep = mean(peak_hourly_wind_power_sep); %Sept
cap_factor_sep = (avg_peak_wind_power_sep/100.3)*100;
avg_peak_wind_power_oct = mean(peak_hourly_wind_power_oct); %Oct
cap_factor_oct = (avg_peak_wind_power_oct/100.3)*100;
avg_peak_wind_power_nov = mean(peak_hourly_wind_power_nov); %Nov
cap_factor_nov = (avg_peak_wind_power_nov/100.3)*100;
avg_peak_wind_power_dec = mean(peak_hourly_wind_power_dec); %Dec
cap_factor_dec = (avg_peak_wind_power_dec/100.3)*100;
monthly_cap_factors (1,:) = cap_factor_jan;

monthly_cap_factors (2,:) = cap_factor_feb;
monthly_cap_factors (3,:) = cap_factor_mar;
82
monthly_cap_factors (4,:) = cap_factor_apr;
monthly_cap_factors (5,:) = cap_factor_may;
monthly_cap_factors (6,:) = cap_factor_june;
monthly_cap_factors (7,:) = cap_factor_july;
monthly_cap_factors (8,:) = cap_factor_aug;
monthly_cap_factors (9,:) = cap_factor_sep;
monthly_cap_factors (10,:) = cap_factor_oct;
monthly_cap_factors (11,:) = cap_factor_nov;
monthly_cap_factors (12,:) = cap_factor_dec;
%bar(monthly_cap_factors); Monthly cap. contribution from wind plant
%===================================================================
% Annual energy production calculation
for en=1:length(spd_ms)
hourly_pwr (en,:) = power_func(spd_ms(en));%Hourly wind power in MW
for filtered out wind speeds
end
eng_production = trapz (hourly_pwr); % Energy production in MWh

% If aggregated losses are 15%
expected_eng_prod = (eng_production * 0.85)/1000; %Expected annual
energy production in GWh
%===================================================================
%Monthly energy production calculation
wind_jan = spd_ms(month_double==01); %January

for jani=1:length(wind_jan)
hourly_pwr_jan (jani,:) = power_func(wind_jan(jani));
end
eng_production_jan = trapz (hourly_pwr_jan);
expected_eng_prod_jan = (eng_production_jan * 0.85)/1000; %GWh
wind_feb = spd_ms(month_double==02); %February
for febi=1:length(wind_feb)
hourly_pwr_feb (febi,:) = power_func(wind_feb(febi));
end
eng_production_feb = trapz (hourly_pwr_feb);
expected_eng_prod_feb = (eng_production_feb * 0.85)/1000;
wind_mar = spd_ms(month_double==03); %March
for mari=1:length(wind_mar)
hourly_pwr_mar (mari,:) = power_func(wind_mar(mari));
end
eng_production_mar = trapz (hourly_pwr_mar);
expected_eng_prod_mar = (eng_production_mar * 0.85)/1000;
wind_apr = spd_ms(month_double==04); %April
for apri=1:length(wind_apr)
hourly_pwr_apr (apri,:) = power_func(wind_apr(apri));
end
eng_production_apr = trapz (hourly_pwr_apr);
expected_eng_prod_apr = (eng_production_apr * 0.85)/1000;
wind_may = spd_ms(month_double==05); %May
for mayi=1:length(wind_may)
hourly_pwr_may (mayi,:) = power_func(wind_may(mayi));
end
eng_production_may = trapz (hourly_pwr_may);
expected_eng_prod_may = (eng_production_may * 0.85)/1000;
83
wind_june = spd_ms(month_double==06); %June
for junei=1:length(wind_june)
hourly_pwr_june (junei,:) = power_func(wind_june(junei));
end
eng_production_june = trapz (hourly_pwr_june);
expected_eng_prod_june = (eng_production_june * 0.85)/1000;
wind_july = spd_ms(month_double==07); %July
for julyi=1:length(wind_july)
hourly_pwr_july (julyi,:) = power_func(wind_july(julyi));
end
eng_production_july = trapz (hourly_pwr_july);
expected_eng_prod_july = (eng_production_july * 0.85)/1000;
wind_august = spd_ms(month_double==08); %August
for augusti=1:length(wind_august)
hourly_pwr_august (augusti,:) =
power_func(wind_august(augusti));
end
eng_production_august = trapz (hourly_pwr_august);
expected_eng_prod_august = (eng_production_august * 0.85)/1000;
wind_sep = spd_ms(month_double==09); %September
for sepi=1:length(wind_sep)
hourly_pwr_sep (sepi,:) = power_func(wind_sep(sepi));
end
eng_production_sep = trapz (hourly_pwr_sep);
expected_eng_prod_sep = (eng_production_sep * 0.85)/1000;
wind_oct = spd_ms(month_double==10); %October
for octi=1:length(wind_oct)
hourly_pwr_oct (octi,:) = power_func(wind_oct(octi));
end
eng_production_oct = trapz (hourly_pwr_oct);
expected_eng_prod_oct = (eng_production_oct * 0.85)/1000;
wind_nov = spd_ms(month_double==11); %November
for novi=1:length(wind_nov)
hourly_pwr_nov (novi,:) = power_func(wind_nov(novi));
end
eng_production_nov = trapz (hourly_pwr_nov);
expected_eng_prod_nov = (eng_production_nov * 0.85)/1000;
wind_dec = spd_ms(month_double==12); %December
for deci=1:length(wind_dec)
hourly_pwr_dec (deci,:) = power_func(wind_dec(deci));
end
eng_production_dec = trapz (hourly_pwr_dec);
expected_eng_prod_dec = (eng_production_dec * 0.85)/1000;
monthly_eng_prod (1,:) = expected_eng_prod_jan;

monthly_eng_prod (2,:) = expected_eng_prod_feb;
monthly_eng_prod (3,:) = expected_eng_prod_mar;
monthly_eng_prod (4,:) = expected_eng_prod_apr;
monthly_eng_prod (5,:) = expected_eng_prod_may;
monthly_eng_prod (6,:) = expected_eng_prod_june;
monthly_eng_prod (7,:) = expected_eng_prod_july;
monthly_eng_prod (8,:) = expected_eng_prod_august;
monthly_eng_prod (9,:) = expected_eng_prod_sep;
monthly_eng_prod (10,:) = expected_eng_prod_oct;
monthly_eng_prod (11,:) = expected_eng_prod_nov;
monthly_eng_prod (12,:) = expected_eng_prod_dec;
%bar(monthly_eng_prod); Monthly energy production of wind park
84
%====================Power Function=================================
function [wind_power] = power_func(x)
if x<=3 || x>=26
wind_power = 0;
elseif x>=17 && x<=25

wind_power = 850*118;%Wind power in MW
else
wind_power = (var_coeff(x)*118)/1000;%Wind power in MW
end
end
%====================Coefficients of polynomial=====================
function [var_pwr] = var_coeff(x)
p1 = -0.000730103444883633;
p2 = 0.050242948056187;
p3 = -1.406879944953474;
p4 = 20.645523570313510;
p5 = -1.715904264728373e+02;
p6 = 8.198030199834666e+02;
p7 = -2.066155295423886e+03;
p8 = 2.116667132848899e+03;
var_pwr = p1*x^7 + p2*x^6 + p3*x^5 + p4*x^4 + p5*x^3 + p6*x^2

+ p7*x + p8;
end
%================================End================================
85
APPENDIX D
BASE LOAD DEMAND FORECAST OF THE CEB UP TO YEAR 2030
Year Demand Growth Gross Generation Growth Peak

(GWh) Rate (%) Losses (%) (GWh) Rate (%) (MW)
2010 9190 8.9 14.4 10740 8.7 2170
2011 10036 9.2 14.3 11715 9.1 2356
2012 10698 6.6 14.2 12464 6.4 2503
2013 11402 6.6 14.9 13402 7.5 2688
2014 12149 6.6 15.1 14315 6.8 2853
2015 12941 6.5 15.1 15238 6.4 3035
2016 13773 6.4 15.1 16220 6.4 3211
2017 14630 6.2 14.8 17168 5.8 3397
2018 15530 6.2 14.6 18188 5.9 3604
2019 16481 6.1 14.4 19257 5.9 3820
2020 17489 6.1 14.3 20397 5.9 4051
2021 18563 6.1 14.6 21741 6.6 4258
2022 19708 6.2 14.4 23019 5.9 4513
2023 20932 6.2 14.3 24436 6.2 4796
2024 22242 6.3 14.2 25922 6.1 5092
2025 23647 6.3 14.2 27559 6.3 5418
2026 25153 6.4 14.1 29284 6.3 5684
2027 26768 6.4 14.1 31149 6.4 6051
2028 28503 6.5 14.0 33142 6.4 6443
2029 30365 6.5 14.0 35308 6.5 6869
2030 32367 6.6 14.0 37615 6.5 7323
Source: Long Term Generation Expansion Plan 2011-2025, CEB
86
APPENDIX E
DETAILS OF THE CANDIDATE THERMAL PLANTS USED IN THE WASP

RUNS
Type of Power Economic FOR Total Fixed O&M Variable

Plant Life (%) Construction Cost (US$/kW- O&M Cost
Cost (US$/kW) Month) (US$/MWh)
150 MW Oil 30 Years 2.74 1512.4 0.940 5.15
fired steam
plants
300 MW Oil 30 Years 2.74 1278.4 0.659 2.58
fired steam
plants
300 MW Coal 30 Years 2.74 1542.0 0.704 2.76
fired steam
plants
35MW Auto 20 Years 8.0 737.4 0.638 5.12
diesel fired gas
turbines
75MW Auto 20 Years 8.0 602.5 0.549 4.38
diesel fired gas
turbines
105MW Auto 20 Years 8.0 501.3 0.483 3.83
diesel fired gas
turbines
150MW Auto 30 Years 8.0 1070.0 0.484 4.15
diesel fired
Combined
cycle plants
300MW Auto 30 Years 8.0 864.3 0.365 3.13
diesel fired
Combined
cycle plants
Source: Long Term Generation Expansion Plan 2011-2025, CEB
87
APPENDIX F
STANDARD DEVIATION VALUES OF CAPITAL, FUEL, O&M AND CO2

COSTS
Construction Fuel O&M CO2

Coal 0.23 0.14 0.054 0.26
Oil 0.23 0.25 0.242 0.26
Gas 0.15 0.19 0.105 0.26
Hydro 0.1 0 0.153 0
Wind 0.05 0 0.08 0
Biomass 0.2 0.18 0.108 0
Source: Awerbuch and Yang (2007) [55]
88
APPENDIX G
CORRELATION COEFFICIENTS OF FUEL, O&M AND CO2 COSTS
1. Fuel and CO2 Correlation Coefficients

Coal Oil Gas Biomass CO2
Coal 1 0.27 0.47 -0.38 -0.49
Oil 0.27 1 0.49 -0.17 0.19
Gas 0.47 0.49 1 -0.44 0.68
Biomass -0.38 -0.17 -0.44 1 0
CO2 -0.49 0.19 0.68 0 1
2. O&M Correlation Coefficients

Technology Coal Gas Oil Hydro Wind Biomass
Coal 1 0.25 -0.18 0.03 -0.22 0.18
Gas 0.25 1 0.09 -0.04 0 0.32
Oil -0.18 0.09 1 -0.27 -0.58 0.01
Hydro 0.03 -0.04 -0.27 1 0.29 -0.18
Wind -0.22 0 -0.58 0.29 1 -0.18
Biomass 0.18 0.32 0.01 -0.18 -0.18 1
89
APPENDIX H
MATLAB CODE OF MVPT ANALYSIS MODEL

%============Mean Variance Portfolio Theory Analysis================
clear;
%==========Calculation of Technology Risks==========================
% fractional weights of cost components with CO2 cost
%Cap Fuel O&M CO2
FW1= [0.2747 0.4089 0.0444 0.2719 %Coal
0.0885 0.7996 0.0254 0.0865 %Oil
0.1734 0.6636 0.0486 0.1144 %Gas
0.9808 0.0000 0.0192 0.0000 %Hydro
0.8402 0.0000 0.1598 0.0000 %Wind
0.2228 0.7306 0.0465 0.0000]; %Biomass
SQRFW1 = realpow(FW1,2);
% fractional weights of cost components without CO2 cost

%Cap Fuel O&M CO2
FW2= [0.3774 0.5617 0.0610 0.0000 %Coal
0.0968 0.8754 0.0278 0.0000 %Oil
0.1958 0.7493 0.0549 0.0000 %Gas
0.9808 0.0000 0.0192 0.0000 %Hydro
0.8402 0.0000 0.1598 0.0000 %Wind
0.2228 0.7306 0.0465 0.0000]; %Biomass
SQRFW2 = realpow(FW2,2);
% Technology standard deviations of Cost Components

%Cap Fuel O&M CO2
STDCC= [0.23 0.14 0.054 0.26 %Coal
0.23 0.25 0.242 0.26 %Oil
0.15 0.19 0.105 0.26 %Gas
0.10 0 0.153 0 %Hydro
0.05 0 0.08 0 %Wind
0.20 0.18 0.108 0]; %Biomass
SQRSTDCC = realpow(STDCC,2);
%===================================================================
%Tech risk WITH CO2 - TR1
INT1= bsxfun(@times,SQRFW1,SQRSTDCC);
%Fuel and CO2 Correlation

FCO2Corr= [-0.49
0.19
0.68
0
0
0];
X1= FW1(:,2);
X2= FW1(:,4);
X3= STDCC(:,2);
X4= STDCC(:,4);
X5= bsxfun(@times,X1,X2);
X8= bsxfun(@times,X7,FCO2Corr);
X9= 2*X8; %Correlation element array
90
X10 = INT1(1,:); %Tech risk with CO2 - COAL
X11 = sum(X10);
X12 = X11+X9(1,1);
TRCOAL=sqrt(X12); %Technology risk of coal
X13 = INT1(2,:); %Tech risk with CO2 - OIL
X14 = sum(X13);
X15 = X14+X9(2,1);
TROIL=sqrt(X15); %Technology risk of oil
X16 = INT1(3,:); %Tech risk with CO2 - GAS
X17 = sum(X16);
X18 = X17+X9(3,1);
TRGAS=sqrt(X18); %Technology risk of gas
X19 = INT1(4,:); %Tech risk with CO2 - HYDRO
X20 = sum(X19);
X21 = X20+X9(4,1);
TRHYDRO=sqrt(X21); %Technology risk of hydro
X22 = INT1(5,:); %Tech risk with CO2 - WIND
X23 = sum(X22);
X24 = X23+X9(5,1);
TRWIND=sqrt(X24); %Technology risk of wind
X25 = INT1(6,:); %Tech risk with CO2 - BIOMASS
X26 = sum(X25);
X27 = X26+X9(6,1);
TRDENDRO=sqrt(X27); %Technology risk of Biomass
%================End of Tech risk WITH CO2==========================
%===================================================================
%Tech risk WITHOUT CO2 - TR3
INT2= bsxfun(@times,SQRFW2,SQRSTDCC);
%No Fuel and CO2 Correlation
YX10 = INT2(1,:); %Tech risk without CO2 - COAL
YX11 = sum(YX10);
YTRCOAL=sqrt(YX11); %Technology risk of coal
YX13 = INT2(2,:); %Tech risk without CO2 - OIL
YX14 = sum(YX13);
YTROIL=sqrt(YX14); %Technology risk of oil
YX16 = INT2(3,:); %Tech risk without CO2 - GAS
YX17 = sum(YX16);
YTRGAS=sqrt(YX17); %Technology risk of gas
YX19 = INT2(4,:); %Tech risk without CO2 - HYDRO
YX20 = sum(YX19);
YTRHYDRO=sqrt(X20); %Technology risk of hydro
YX22 = INT2(5,:); %Tech risk without CO2 - WIND
YX23 = sum(YX22);
YTRWIND=sqrt(YX23); %Technology risk of wind
YX24 = INT2(6,:); %Tech risk without CO2 - BIOMASS
YX25 = sum(YX24);
YTRDENDRO=sqrt(YX25); %Technology risk of Biomass
%===============End of Tech risk WITHOUT CO2========================
TR1=[TRCOAL TROIL TRGAS TRHYDRO TRWIND TRDENDRO]; %Technology stds
with CO2 cost risk for new plants
TR3=[YTRCOAL YTROIL YTRGAS YTRHYDRO YTRWIND YTRDENDRO]; %Technology
stds without CO2 cost risk for new plants
%===============End of Calculation of Technology risks==============
%===Calculation of Correlation coefficients between technologies====
%====Correlations between technologies with CO2 costs===============
C1 = [0.23 0 0.23 0
0 0.14 0 0
0.054 0 0.054 0
91
0 0 0 0.26];%Stds of Coal
C2 = [0.2747 0 0.2747 0
0 0.4089 0 0
0.0444 0 0.0444 0
0 0 0 0.2719];%Weights of Coal
O1 = [0.23 0 0.242 0
0 0.25 0 0
0.23 0 0.242 0
0 0 0 0.26];%Stds of Oil
O2 = [0.0885 0 0.0254 0
0 0.7996 0 0
0.0885 0 0.0254 0
0 0 0 0.0865];%Weights of Oil
G1 = [0.15 0 0.105 0
0 0.19 0 0
0.15 0 0.105 0
0 0 0 0.26];%Stds of Gas
G2 = [0.1734 0 0.0486 0
0 0.6636 0 0
0.1734 0 0.0486 0
0 0 0 0.1144];%Weights of Gas
% Correlation between Coal and Oil

%Cap %Fuel %O&M %CO2
CO = [0.7 0 0.1 0 %Capital
0 0.27 0 0 %Fuel
0.1 0 -0.18 0 %O&M
0 0 0 1]; %CO2
CO1= bsxfun(@times,CO,C1);
CO2= bsxfun(@times,CO1,O1);
CO3= bsxfun(@times,CO2,C2);
CO4= bsxfun(@times,CO3,O2);
CO5= sum(CO4);
CO6= sum(CO5);
CorCO= CO6/(TRCOAL*TROIL); % Tech. Corr. Coeff. between Coal and Oil
% Correlation between Coal and Gas

CG = [0.7 0 0.1 0 %Capital
0 0.47 0 0 %Fuel
0.1 0 0.25 0 %O&M
0 0 0 1]; %CO2
CG1= bsxfun(@times,CG,C1);
CG2= bsxfun(@times,CG1,G1);
CG3= bsxfun(@times,CG2,C2);
CG4= bsxfun(@times,CG3,G2);
CG5= sum(CG4);
CG6= sum(CG5);
CorCG= CG6/(TRCOAL*TRGAS); % Tech. Corr. Coeff. between Coal and Gas
% Correlation between Coal and Hydro

CH = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 0.03 0 %O&M
0 0 0 0]; %CO2
92
H1 = [0.1 0 0.153 0
0 0 0 0
0.1 0 0.153 0
0 0 0 0];%Stds of Hydro
H2 = [0.9808 0 0.0192 0
0 0 0 0
0.9808 0 0.0192 0
0 0 0 0];%Weights of Hydro
CH1= bsxfun(@times,CH,C1);
CH2= bsxfun(@times,CH1,H1);
CH3= bsxfun(@times,CH2,C2);
CH4= bsxfun(@times,CH3,H2);
CH5= sum(CH4);
CH6= sum(CH5);
CorCH= CH6/(TRCOAL*TRHYDRO); % Tech. Corr. Coeff. btwn Coal & Hydro
% Correlation between Coal and Wind

CW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.22 0 %O&M
0 0 0 0]; %CO2
W1 = [0.05 0 0.08 0
0 0 0 0
0.05 0 0.08 0
0 0 0 0];%Stds of Wind
W2 = [0.8402 0 0.1598 0
0 0 0 0
0.8402 0 0.1598 0
0 0 0 0];%Weights of Wind
CW1= bsxfun(@times,CW,C1);
CW2= bsxfun(@times,CW1,W1);
CW3= bsxfun(@times,CW2,C2);
CW4= bsxfun(@times,CW3,W2);
CW5= sum(CW4);
CW6= sum(CW5);
CorCW= CW6/(TRCOAL*TRWIND); % Tech. Corr. Coeff. btwn Coal & Wind
% Correlation between Coal and Biomass

CD = [0.1 0 0.1 0 %Capital
0 -0.38 0 0 %Fuel
0.1 0 0.18 0 %O&M
0 0 0 0]; %CO2
D1 = [0.2 0 0.108 0
0 0.18 0 0
0.2 0 0.108 0
0 0 0 0];%Stds of Biomass
D2 = [0.2228 0 0.0465 0
0 0.7306 0 0
0.2228 0 0.0465 0
0 0 0 0];%Weights of Biomass
CD1= bsxfun(@times,CD,C1);
CD2= bsxfun(@times,CD1,D1);
CD3= bsxfun(@times,CD2,C2);
CD4= bsxfun(@times,CD3,D2);
93
CD5= sum(CD4);
CD6= sum(CD5);
CorCD= CD6/(TRCOAL*TRDENDRO); % Tech.Corr.Coeff. Btwn Coal & Biomass
% Correlation between Oil and Gas

OG = [0.7 0 0.1 0 %Capital
0 0.49 0 0 %Fuel
0.1 0 0.09 0 %O&M
0 0 0 1]; %CO2
O3 = O1';
O4 = O2';
OG1= bsxfun(@times,OG,O3);
OG2=bsxfun(@times,OG1,G1);
OG3=bsxfun(@times,OG2,O4);
OG4=bsxfun(@times,OG3,G2);
OG5= sum(OG4);
OG6= sum(OG5);
CorOG= OG6/(TROIL*TRGAS); % Tech.Corr.Coeff. between Oil and Gas
% Correlation between Oil and Hydro

OH = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.27 0 %O&M
0 0 0 0]; %CO2
OH1= bsxfun(@times,OH,O3);
OH2=bsxfun(@times,OH1,H1);
OH3=bsxfun(@times,OH2,O4);
OH4=bsxfun(@times,OH3,H2);
OH5= sum(OH4);
OH6= sum(OH5);
CorOH= OH6/(TROIL*TRHYDRO); % Tech.Corr.Coeff. between Oil and Hydro
% Correlation between Oil and Wind

OW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.58 0 %O&M
0 0 0 0]; %CO2
OW1= bsxfun(@times,OW,O3);
OW2=bsxfun(@times,OW1,W1);
OW3=bsxfun(@times,OW2,O4);
OW4=bsxfun(@times,OW3,W2);
OW5= sum(OW4);
OW6= sum(OW5);
CorOW= OW6/(TROIL*TRWIND); % Tech.Corr.Coeff.between Oil and Wind
% Correlation between Oil and Biomass

OD = [0.1 0 0.1 0 %Capital
0 -0.17 0 0 %Fuel
0.1 0 0.01 0 %O&M
0 0 0 0]; %CO2
OD1= bsxfun(@times,OD,O3);
OD2= bsxfun(@times,OD1,D1);
OD3= bsxfun(@times,OD2,O4);
94
OD4= bsxfun(@times,OD3,D2);
OD5= sum(OD4);
OD6= sum(OD5);
CorOD= OD6/(TROIL*TRDENDRO); % Tech.Corr.Coeff. btwn Oil and Biomass
% Correlation between Gas and Hydro

GH = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.04 0 %O&M
0 0 0 0]; %CO2
G3 = G1';
G4 = G2';
GH1= bsxfun(@times,GH,G3);
GH2=bsxfun(@times,GH1,H1);
GH3=bsxfun(@times,GH2,G4);
GH4=bsxfun(@times,GH3,H2);
GH5= sum(GH4);
GH6= sum(GH5);
CorGH= GH6/(TRGAS*TRHYDRO); % Tech.Corr.Coeff. btwn Gas and Hydro
% Correlation between Gas and Wind

GW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 0 0 %O&M
0 0 0 0]; %CO2
GW1= bsxfun(@times,GW,G3);
GW2=bsxfun(@times,GW1,W1);
GW3=bsxfun(@times,GW2,G4);
GW4=bsxfun(@times,GW3,W2);
GW5= sum(GW4);
GW6= sum(GW5);
CorGW= GW6/(TRGAS*TRWIND); % Tech.Corr.Coeff. between Gas and Wind
% Correlation between Gas and Biomass

GD = [0.1 0 0.1 0 %Capital
0 -0.44 0 0 %Fuel
0.1 0 0.32 0 %O&M
0 0 0 0]; %CO2
GD1= bsxfun(@times,GD,G3);
GD2=bsxfun(@times,GD1,D1);
GD3=bsxfun(@times,GD2,G4);
GD4=bsxfun(@times,GD3,D2);
GD5= sum(GD4);
GD6= sum(GD5);
CorGD= GD6/(TRGAS*TRDENDRO); % Tech.Corr.Coeff. btwn Gas and Biomass
% Correlation between Hydro and Wind

HW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 0.29 0 %O&M
0 0 0 0]; %CO2
95
H3 = H1';
H4 = H2';
HW1= bsxfun(@times,HW,H3);
HW2= bsxfun(@times,HW1,W1);
HW3= bsxfun(@times,HW2,H4);
HW4= bsxfun(@times,HW3,W2);
HW5= sum(HW4);
HW6= sum(HW5);
CorHW= HW6/(TRHYDRO*TRWIND); % Tech.Corr.Coeff. btwn Hydro and Wind
% Correlation between Hydro and Biomass

HD = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.18 0 %O&M
0 0 0 0]; %CO2
HD1= bsxfun(@times,HD,H3);
HD2= bsxfun(@times,HD1,D1);
HD3= bsxfun(@times,HD2,H4);
HD4= bsxfun(@times,HD3,D2);
HD5= sum(HD4);
HD6= sum(HD5);
CorHD= HD6/(TRHYDRO*TRDENDRO); % Tech.Corr.Coeff. btwn Hyd. & BMS.
% Correlation between Wind and Biomass

WD = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.18 0 %O&M
0 0 0 0]; %CO2
W3 = W1';
W4 = W2';
WD1= bsxfun(@times,WD,W3);
WD2= bsxfun(@times,WD1,D1);
WD3= bsxfun(@times,WD2,W4);
WD4= bsxfun(@times,WD3,D2);
WD5= sum(WD4);
WD6= sum(WD5);
CorWD= WD6/(TRWIND*TRDENDRO); % Tech.Corr.Coeff. bten Wind & Biomass
%====End of Correlations between technologies with CO2 costs========

%====Correlations between technologies without CO2 costs============
XC1 = [0.23 0 0.23 0

0 0.14 0 0
0.054 0 0.054 0
0 0 0 0];%Stds of Coal
XC2 = [0.3774 0 0.3774 0
0 0.5617 0 0
0.0610 0 0.0610 0
0 0 0 0];%Weights of Coal
XO1 = [0.23 0 0.242 0
0 0.25 0 0
0.23 0 0.242 0
0 0 0 0];%Stds of Oil
XO2 = [0.0968 0 0.0278 0
0 0.8754 0 0
96
0.0968 0 0.0278 0
0 0 0 0];%Weights of Oil
XG1 = [0.15 0 0.105 0
0 0.19 0 0
0.15 0 0.105 0
0 0 0 0];%Stds of Gas
XG2 = [0.1958 0 0.0549 0
0 0.7493 0 0
0.1958 0 0.0549 0
0 0 0 0];%Weights of Gas
% Correlation between Coal and Oil without CO2

XCO = [0.7 0 0.1 0 %Capital
0 0.27 0 0 %Fuel
0.1 0 -0.18 0 %O&M
0 0 0 0]; %CO2
XCO1= bsxfun(@times,XCO,XC1);
XCO2= bsxfun(@times,XCO1,XO1);
XCO3= bsxfun(@times,XCO2,XC2);
XCO4= bsxfun(@times,XCO3,XO2);
XCO5= sum(XCO4);
XCO6= sum(XCO5);
XCorCO= XCO6/(YTRCOAL*YTROIL); %TCorrCoeff btwn Coal&Oil W/O CO2
% Correlation between Coal and Gas without CO2

XCG = [0.7 0 0.1 0 %Capital
0 0.47 0 0 %Fuel
0.1 0 0.25 0 %O&M
0 0 0 0]; %CO2
XCG1= bsxfun(@times,XCG,XC1);
XCG2= bsxfun(@times,XCG1,XG1);
XCG3= bsxfun(@times,XCG2,XC2);
XCG4= bsxfun(@times,XCG3,XG2);
XCG5= sum(XCG4);
XCG6= sum(XCG5);
XCorCG= XCG6/(YTRCOAL*YTRGAS); % TCorrCoeff btwn Coa&Gas W/O CO2
% Correlation between Coal and Hydro without CO2

XCH = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 0.03 0 %O&M
0 0 0 0]; %CO2
XH1 = H1;
XH2 = H2;
XCH1= bsxfun(@times,XCH,XC1);
XCH2= bsxfun(@times,XCH1,XH1);
XCH3= bsxfun(@times,XCH2,XC2);
XCH4= bsxfun(@times,XCH3,XH2);
XCH5= sum(XCH4);
XCH6= sum(XCH5);
XCorCH= XCH6/(YTRCOAL*YTRHYDRO);%TCorrCoeff btwn Coal&Hyd W/O CO2
97
% Correlation between Coal and Wind without CO2
XCW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.22 0 %O&M
0 0 0 0]; %CO2
XW1 = W1;
XW2 = W2;
XCW1= bsxfun(@times,XCW,XC1);
XCW2= bsxfun(@times,XCW1,XW1);
XCW3= bsxfun(@times,XCW2,XC2);
XCW4= bsxfun(@times,XCW3,XW2);
XCW5= sum(XCW4);
XCW6= sum(XCW5);
XCorCW= XCW6/(YTRCOAL*YTRWIND);%TCorrCoeff btwn Coal&Wind W/O CO2
% Correlation between Coal and Biomass without CO2

XCD = [0.1 0 0.1 0 %Capital
0 -0.38 0 0 %Fuel
0.1 0 0.18 0 %O&M
0 0 0 0]; %CO2
XD1 = D1;
XD2 = D2;
XCD1= bsxfun(@times,XCD,XC1);
XCD2= bsxfun(@times,XCD1,XD1);
XCD3= bsxfun(@times,XCD2,XC2);
XCD4= bsxfun(@times,XCD3,XD2);
XCD5= sum(XCD4);
XCD6= sum(XCD5);
XCorCD= XCD6/(YTRCOAL*YTRDENDRO);%TCorrCoeff btwn Coal&BMS W/O CO2
% Correlation between Oil and Gas without CO2
XOG = [0.7 0 0.1 0 %Capital
0 0.49 0 0 %Fuel
0.1 0 0.09 0 %O&M
0 0 0 0]; %CO2
XO3 = XO1';
XO4 = XO2';
XOG1= bsxfun(@times,XOG,XO3);
XOG2= bsxfun(@times,XOG1,XG1);
XOG3= bsxfun(@times,XOG2,XO4);
XOG4= bsxfun(@times,XOG3,XG2);
XOG5= sum(XOG4);
XOG6= sum(XOG5);
XCorOG= XOG6/(YTROIL*YTRGAS);%TCorrCoeff btwn Oil&Gas W/O CO2
% Correlation between Oil and Hydro without CO2

XOH = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.27 0 %O&M
0 0 0 0]; %CO2
XOH1= bsxfun(@times,XOH,XO3);
XOH2= bsxfun(@times,XOH1,XH1);
XOH3= bsxfun(@times,XOH2,XO4);
XOH4= bsxfun(@times,XOH3,XH2);
98
XOH5= sum(XOH4);
XOH6= sum(XOH5);
XCorOH= XOH6/(YTROIL*YTRHYDRO);%TCorrCoeff btwn Oil&Hydro W/O CO2
% Correlation between Oil and Wind without CO2

XOW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.58 0 %O&M
0 0 0 0]; %CO2
XOW1= bsxfun(@times,XOW,XO3);
XOW2= bsxfun(@times,XOW1,XW1);
XOW3= bsxfun(@times,XOW2,XO4);
XOW4= bsxfun(@times,XOW3,XW2);
XOW5= sum(XOW4);
XOW6= sum(XOW5);
XCorOW= XOW6/(YTROIL*YTRWIND);%TCorrCoeff btwn Oil&Wind W/O CO2
% Correlation between Oil and Biomass without CO2

XOD = [0.1 0 0.1 0 %Capital
0 -0.17 0 0 %Fuel
0.1 0 0.01 0 %O&M
0 0 0 0]; %CO2
XOD1= bsxfun(@times,XOD,XO3);
XOD2= bsxfun(@times,XOD1,XD1);
XOD3= bsxfun(@times,XOD2,XO4);
XOD4= bsxfun(@times,XOD3,XD2);
XOD5= sum(XOD4);
XOD6= sum(XOD5);
XCorOD= XOD6/(YTROIL*YTRDENDRO);%TCorrCoeff btwn Oil&BMS W/O CO2
% Correlation between Gas and Hydro without CO2

XGH = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.04 0 %O&M
0 0 0 0]; %CO2
XG3 = XG1';
XG4 = XG2';
XGH1= bsxfun(@times,XGH,XG3);
XGH2= bsxfun(@times,XGH1,XH1);
XGH3= bsxfun(@times,XGH2,XG4);
XGH4= bsxfun(@times,XGH3,XH2);
XGH5= sum(XGH4);
XGH6= sum(XGH5);
XCorGH= XGH6/(YTRGAS*YTRHYDRO);%TCorrCoeff btwn Gas&Hydro W/O CO2
% Correlation between Gas and Wind without CO2

XGW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 0 0 %O&M
0 0 0 0]; %CO2
XGW1= bsxfun(@times,XGW,XG3);
XGW2= bsxfun(@times,XGW1,XW1);
XGW3= bsxfun(@times,XGW2,XG4);
99
XGW4= bsxfun(@times,XGW3,XW2);
XGW5= sum(XGW4);
XGW6= sum(XGW5);
XCorGW= XGW6/(YTRGAS*YTRWIND);%TCorrCoeff btwn Gas&Wind W/O CO2
% Correlation between Gas and Biomass without CO2

XGD = [0.1 0 0.1 0 %Capital
0 -0.44 0 0 %Fuel
0.1 0 0.32 0 %O&M
0 0 0 0]; %CO2
XGD1= bsxfun(@times,XGD,XG3);
XGD2= bsxfun(@times,XGD1,XD1);
XGD3= bsxfun(@times,XGD2,XG4);
XGD4= bsxfun(@times,XGD3,XD2);
XGD5= sum(XGD4);
XGD6= sum(XGD5);
XCorGD= XGD6/(YTRGAS*YTRDENDRO);%TCorrCoeff btwn Gas&Biomass W/O CO2
% Correlation between Hydro and Wind without CO2

XHW = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 0.29 0 %O&M
0 0 0 0]; %CO2
XH3 = XH1';
XH4 = XH2';
XHW1= bsxfun(@times,XHW,XH3);
XHW2= bsxfun(@times,XHW1,XW1);
XHW3= bsxfun(@times,XHW2,XH4);
XHW4= bsxfun(@times,XHW3,XW2);
XHW5= sum(XHW4);
XHW6= sum(XHW5);
XCorHW= XHW6/(YTRHYDRO*YTRWIND);%TCorrCoeff btwn Hydro&Wind W/O CO2
% Correlation between Hydro and Biomass without CO2

XHD = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.18 0 %O&M
0 0 0 0]; %CO2
XHD1= bsxfun(@times,XHD,XH3);
XHD2= bsxfun(@times,XHD1,XD1);
XHD3= bsxfun(@times,XHD2,XH4);
XHD4= bsxfun(@times,XHD3,XD2);
XHD5= sum(XHD4);
XHD6= sum(XHD5);
XCorHD= XHD6/(YTRHYDRO*YTRDENDRO);%TCorrCoeff btwn Hydro&BMS W/O CO2
% Correlation between Wind and Biomass without CO2

XWD = [0.1 0 0.1 0 %Capital
0 0 0 0 %Fuel
0.1 0 -0.18 0 %O&M
0 0 0 0]; %CO2
XW3 = XW1';
XW4 = XW2';
100
XWD1= bsxfun(@times,XWD,XW3);
XWD2= bsxfun(@times,XWD1,XD1);
XWD3= bsxfun(@times,XWD2,XW4);
XWD4= bsxfun(@times,XWD3,XD2);
XWD5= sum(XWD4);
XWD6= sum(XWD5);
XCorWD= XWD6/(YTRWIND*YTRDENDRO);%TCorrCoeff btwn Wind&BMS W/O CO2
%===End of Correlations between technologies without CO2 costs======

%Tech Correlation coefficients with CO2 cost
TCC = [1 CorCO CorCG CorCH CorCW CorCD
CorCO 1 CorOG CorOH CorOW CorOD
CorCG CorOG 1 CorGH CorGW CorGD
CorCH CorOH CorGH 1 CorHW CorHD
CorCW CorOW CorGW CorHW 1 CorWD
CorCD CorOD CorGD CorHD CorWD 1];
%Tech Correlation coefficients without CO2 cost
XTCC = [1 XCorCO XCorCG XCorCH XCorCW XCorCD
XCorCO 1 XCorOG XCorOH XCorOW XCorOD
XCorCG XCorOG 1 XCorGH XCorGW XCorGD
XCorCH XCorOH XCorGH 1 XCorHW XCorHD
XCorCW XCorOW XCorGW XCorHW 1 XCorWD
XCorCD XCorOD XCorGD XCorHD XCorWD 1];
%End of Calculation of Correlation coefficients between technologies
%=======MVPT Analysis of 2025 Generation Mix========================

D = diag(TR1);
XD = diag(TR3);
COV=D*TCC*D; %covariance matrix (with CO2)
XCOV=XD*XTCC*XD; %covariance matrix (without CO2)
%Coal Oil Gas Hydro Wind Biomass

LCOE = [93.29 198.98 92.11 97.60 137.16 136.92]; %LCOE
XLCOE = [67.92 181.76 81.57 97.60 137.16 136.92]; %LOCE
without CO2 cost
%=============Scenario (1) with CO2 Cost============================

RTN = 1./LCOE;%Return (with CO2 cost)
bounds = [0.065 0 0 0.18 0.005 0; 0.815 0.75 0.75 0.30 0.10
0.10];% Upper and lower bounds of technologies
[risk,returns,wts] = frontcon(RTN,COV,5000,[],bounds);
cost = 1./returns;
%plot(risk*100,cost);%Efficient frontier
hold on;
portwts_ceb = [0.709458 0.011106 0.0000 0.249637 0.01372
0.016079];%CEB Energy Mix 2025_with CO2
[risk_ceb,ret_ceb] = portstats(RTN,COV,portwts_ceb);
cost_ceb = 1./ret_ceb;
%scatter(risk_ceb*100,cost_ceb);
hold on;
yportwts_ceb = [0.146112 0.396694 0.000000 0.437776
0.011233 0.008184];%CEB Energy Mix 2012_with CO2
[yrisk_ceb,yret_ceb] = portstats(RTN,COV,yportwts_ceb);
ycost_ceb = 1./yret_ceb;
%scatter(yrisk_ceb*100,ycost_ceb);
hold on;
%scatter(TR1*100,LCOE);%Individual technology cost and risk
101
portwts1 = wts(4231,:);%Same risk portfolio_2025 mix
[risk1,ret1] = portstats(RTN,COV,portwts1);
cost1 = 1./ret1;
%scatter(risk1*100,cost1);
hold on;
portwts2 = wts(3535,:);% Same cost less risk portfolio_2025mix
cost2 = 1./ret2;
hold on;
portwts3 = wts(1,:);% Least risk portfolio
cost3 = 1./ret3;
hold on;
portwts4 = wts(5000,:);% Least cost portfolio
cost4 = 1./ret4;
hold on;
portwts5 = wts(4847,:);% Same risk less cost portfolio_2012mix
cost5 = 1./ret5;
hold on;
%Cost risk matrix of scenario 1
Cost_risk_1 = [cost_ceb risk_ceb*100 %CEB 2025
ycost_ceb yrisk_ceb*100 %CEB 2012
cost3 risk3*100 %A1
cost2 risk2*100 %B1
cost1 risk1*100 %C1
cost4 risk4*100 %D1
cost5 risk5*100]; %E1
efportwts = 100*[portwts1 %Same risk_2025
portwts2 %Same cost_2025
portwts3 %Least risk_2025
portwts4 %Least cost_2025
portwts5]; %Same risk_2012
%=============End of Scenario (1)===================================

%=============Scenario (2) without CO2 Cost=========================
XRTN = 1./XLCOE;% Return (Without CO2 Cost)

[xrisk,xreturns,xwts] = frontcon(XRTN,XCOV,5000,[],bounds);
xcost = 1./xreturns;
%plot(xrisk*100,xcost);%Efficient frontier
hold on;
xportwts_ceb = [0.709458 0.011106 0.0000 0.249637 0.01372
0.016079];%CEB Energy Mix 2025
[xrisk_ceb,xret_ceb] = portstats(XRTN,XCOV,xportwts_ceb);
xcost_ceb = 1./xret_ceb;
%scatter(xrisk_ceb*100,xcost_ceb);
hold on;
xxportwts_ceb = [0.146112 0.396694 0.000000 0.437776
0.011233 0.008184];%CEB Energy Mix 2012
[xxrisk_ceb,xxret_ceb] = portstats(XRTN,XCOV,xxportwts_ceb);
xxcost_ceb = 1./xxret_ceb;
102
%scatter(xxrisk_ceb*100,xxcost_ceb);
hold on;
%scatter(TR3*100,XLCOE);
xportwts1 = xwts(4066,:);%Same risk portfolio_2025 mix
[xrisk1,xret1] = portstats(XRTN,XCOV,xportwts1);
xcost1 = 1./xret1;
%scatter(xrisk1*100,xcost1);
hold on;
xportwts2 = xwts(3878,:);% Same cost less risk portfolio_2025mix
xcost2 = 1./xret2;
hold on;
xportwts3 = xwts(1,:);% Least risk portfolio
xcost3 = 1./xret3;
hold on;
xportwts4 = xwts(5000,:);% Least cost portfolio
xcost4 = 1./xret4;
hold on;
%Cost risk matrix of scenario 2
Cost_risk_2 = [xcost_ceb xrisk_ceb*100 %CEB 2025
xxcost_ceb xxrisk_ceb*100 %CEB 2012
xcost3 xrisk3*100 %A2
xcost2 xrisk2*100 %B2
xcost1 xrisk1*100 %C2
xcost4 xrisk4*100]; %D2
xefportwts = 100*[xportwts1 %Same risk_2025
xportwts2 %Same cost_2025
xportwts3 %Least risk_2025
xportwts4]; %Least cost_2025
%==============End of Scenario (2)==================================
%=========End of Mean Variance Portfolio Theory Analysis============
103
REFERENCES
[1] National Energy Policy and Strategies of Sri Lanka, Ministry of Power and
Energy, Government of Sri Lanka, October 2006.
[2] Generation Performance in Sri Lanka – 2012, Public Utilities Commission of
Sri Lanka, July 2013.
[3] Ceylon Electricity Board, Long Term Generation Expansion Plan 2011-2025,
April 2011.
[4] Giebel, G. (2005). Wind power has a capacity credit a catalogue of 50+
supporting studies. WindEng EJournal, windeng.net.
[5] Awerbuch, S., Jansen, J. C., & Beurskens, L. (2004). Building capacity for
portfolio-based energy planning in developing countries. The Renewable
Energy & Energy Efficiency Partnership (REEEP) and UNEP, London-Paris,
ECN project, (77565).
[6] Wijayatunga, P. D. C., Fernando, W. J. L. S., & Shrestha, R. M. (2004).
Impact of distributed and independent power generation on greenhouse gas
emissions: Sri Lanka. Energy conversion and management, 45(20), 3193-
3206.
[7] Fernando, K. S., Kariyawasam, P. L. G., & Alwis, A. M. A. (2002). Wind
Energy Resource Assessment – Puttalam and Central Regions of Sri Lanka.
Ceylon Electricity Board, Colombo, Sri Lanka.
[8] Elliott, D., Schwartz, M., Scott, G., Haymes, S., Heimiller, D., & George, R.
(2003). Wind energy resource atlas of Sri Lanka and the Maldives (No.
NREL/TP-500-34518). National Renewable Energy Laboratory (NREL),
Golden, CO..
[9] Wijayatunga, P. D.C., Daranagama, U., & Ariyadasa, K. P. (2005). Techno-
economic feasibility of biomass-based electricity generation in Sri Lanka.
Bioenergy-Realizing the Potential, 29, 141.
[10] Cochran, J., Bird, L., Heeter, J., & Arent, D. J. (2012). Integrating variable
renewable energy in electric power markets: Best practices from international
experience (No. NREL/TP-6A00-53732). National Renewable Energy
Laboratory.
[11] Wijayatunga, P. D. C., Siriwardena, K., Fernando, W. J. L. S., Shrestha, R.
M., & Attalage, R. A. (2006). Strategies to overcome barriers for cleaner
generation technologies in small developing power systems: Sri Lanka case
study. Energy conversion and management, 47(9), 1179-1191.
[12] Sri Lanka Sustainable Energy Authority, On-grid renewable energy
development: A guide to the project approval process for On-grid renewable
energy project development, July 2011.
[13] Tanaka, N. (2009). World energy outlook 2009. International Energy Agency.
104
[14] Wijayatunga, P. D. C. (2012). Regulation for renewable energy development:
Lessons from Sri Lanka experience. Renewable Energy, 61, 29-32.
[15] Sawin, J. L. (2004). Mainstreaming renewable energy in the 21st century
(Vol. 169). Worldwatch Institute.
[16] Sawin, J. L. (2006). National policy instruments: Policy lessons for the
advancement & diffusion of renewable energy technologies around the world.
Renewable Energy. A Global Review of Technologies, Policies and Markets.
[17] Wijayatunga, P. D. C., & Prasad, D. (2009). Clean energy technology and
regulatory interventions for Greenhouse Gas emission mitigation: Sri Lankan
power sector. Energy Conversion and Management, 50(6), 1595-1603.
[18] Goldstein, L. L., Mortensen, J., & Trickett, D. (1999). Grid-connected
renewable-electric policies in the European Union (No. NREL/TP-620-
26247). National Renewable Energy Laboratory.
[19] Krohn, S. (2002). Danish wind turbines: an industrial success story. Available
at the www-site of the Danish Wind Industry Association: www.windpower.
org.
[20] International Atomic Energy Agency, Wien Automatic System Planning
(WASP) Package: A Computer Code for Power Generating System
Expansion Planning, Version WASP-IV with User Interface - User’s Manual,
Vienna, 2006.
[21] Birman, K., Ganesh, L., Renessee, R. V., Ferris, M., Hofmann, A., Williams,
B., ... & Smith, B. (2011). Computational Needs for the Next Generation
Electric Grid Proceedings (No. LBNL-5105E). Ernest Orlando Lawrence
Berkeley National Laboratory, Berkeley, CA (US).
[22] Connolly, D., Lund, H., Mathiesen, B. V., & Leahy, M. (2010). A review of
computer tools for analysing the integration of renewable energy into various
energy systems. Applied Energy, 87(4), 1059-1082.
[23] Foley, A. M., Ó Gallachóir, B. P., Hur, J., Baldick, R., & McKeogh, E. J.
(2010). A strategic review of electricity systems models. Energy, 35(12),
4522-4530.
[24] Bhattacharyya, S., & Timilsina, G. R. (2009). Energy demand models for
policy formulation: a comparative study of energy demand models. World
Bank Policy Research Working Paper Series, Policy Research Working Paper
No. 4866.
[25] Rastler, D. (2011). Midcontinent Independent System Operator (MISO)
Energy Storage Study, Electric Power Research Institute (EPRI) and MSIO,
Report No. 1024489.
[26] Corbus, D., King, J., Mousseau, T., Zavadil, R., Heath, B., Hecker, L., ... &
Moland, G. (2010). Eastern wind integration and transmission study (No.
NREL/ CP-550-46505). National Renewable Energy Laboratory.
105
[27] Central Electricity Authority, National Electricity Plan - 2012, Ministry of
Power, Government of India, January 2012.
[28] Midcontinent Independent System Operator Inc., MVP Business Case
Metrics: Capacity Cost Savings from New Transmission Investment, Planning
Advisory Committee, May 2011.
[29] Deane, J. P., Chiodi, A., Gargiulo, M., & Ó Gallachóir, B. P. (2012). Soft-
linking of a power systems model to an energy systems model. Energy, 42(1),
303-312.
[30] Nweke, C. I., Leanez, F., Drayton, G. R., & Kolhe, M. (2012, October).
Benefits of chronological optimization in capacity planning for electricity
markets. In Power System Technology (POWERCON), 2012 IEEE
International Conference on (pp. 1-6). IEEE.
[31] Energy Exemplar, PLEXOS for Power Systems, www.energyexemplar.com.
[32] William E, L., Luke J, R., Liam D, W., Colin F, A., & Anthony R, S. (2012).
An economic evaluation of the potential for distributed energy in Australia.
Energy Policy, 51, 277–289.
[33] Deane, J. P., Drayton, G., & Ó Gallachóir, B. P. (2014). The impact of sub-
hourly modelling in power systems with significant levels of renewable
generation. Applied Energy, 113, 152-158.
[34] Mc Garrigle, E. V., Deane, J. P., & Leahy, P. G. (2013). How much wind
energy will be curtailed on the 2020 Irish power system?. Renewable Energy,
55, 544-553.
[35] Deane, J. P., Dalton, G., & Ó Gallachóir, B. P. (2012). Modelling the
economic impacts of 500MW of wave power in Ireland. Energy Policy, 45,
614-627.
[36] Price, J. E. (2011, July). Management of available transfer capability with
growth in renewable resources. In Power and Energy Society General
Meeting, 2011 IEEE (pp. 1-8). IEEE.
[37] Mudannayake, N. A., Modelling of renewable energy based embedded
generation in long term planning: A Sri Lanka case study, Master’s thesis,
University of Moratuwa, 2011.
[38] Koritarov, V., Modeling wind energy resources in generation expansion
models, FERC Technical conference on planning models and software,
Washington DC, June 2010.
[39] Garver, L. L. (1966). Effective load carrying capability of generating units.
Power Apparatus and Systems, IEEE Transactions on, (8), 910-919.
[40] Milligan, M. & Porter, K. (2005). Determining the capacity value of wind: A
survey of methods and implementation (No. NREL/ CP-500-38062), National
Renewable Energy Laboratory.
106
[41] Milligan, M. & Porter, K. (2008). Determining the capacity value of wind: An
updated survey of methods and implementation (No. NREL/ CP-500-43433),
National Renewable Energy Laboratory.
[42] Keane, A., Milligan, M., Dent, C. J., Hasche, B., D'Annunzio, C., Dragoon,
K., ... & O'Malley, M. (2011). Capacity value of wind power. Power Systems,
IEEE Transactions on, 26(2), 564-572.
[43] Milligan, M. R. (2000). Modelling utility‐scale wind power plants. Part 2:
Capacity credit. Wind Energy, 3(4), 167-206.
[44] Milligan, M., & Parsons, B. (1999). A comparison and case study of capacity
credit algorithms for wind power plants. Wind Engineering, 23(3), 159-166.
[45] Masters, G. M. (2005) Wind Power Systems, in Renewable and Efficient
Electric Power Systems, John Wiley & Sons, Inc., Hoboken, NJ, USA.
[46] European Wind Energy Association. (2009). Wind energy-the facts: a guide to
the technology, economics and future of wind power. Earthscan, London, UK.
[47] Public Utilities Commission of Sri Lanka, Least-Cost Generation Expansion
Planning Code, April 2011.
for the Evaluation of Wind Power using the WASP‐IV Computer Model, The
[48] Koritarov, V., Conzelmann, G., Veselka, T. & Hamilton, B., A Methodology
World Bank, January 2005.

[49] Gross, R., Blyth, W., & Heptonstall, P. (2010). Risks, revenues and
investment in electricity generation: why policy needs to look beyond costs.
Energy Economics, 32(4), 796-804.
[50] Awerbuch, S. (2006). Portfolio-based electricity generation planning: policy
implications for renewables and energy security. Mitigation and adaptation
strategies for Global Change, 11(3), 693-710.
[51] Costello, K. (2007). Diversity of generation technologies: implications for
decision-making and public policy. The Electricity Journal, 20(5), 10-21.
[52] Awerbuch, S., Beurskens, L., Drennen, T., & Jansen, J. C. (2005). The cost of
geothermal energy in the western US region: A portfolio-based approach: A
mean-variance portfolio optimization of the region’s generating mix to 2013
(No. SAND2005-5173). Sandia National Laboratories.
[53] Awerbuch, S., Dillard, J., Mouck, T., & Preston, A. (1996). Capital
budgeting, technological innovation and the emerging competitive
environment of the electric power industry. Energy Policy, 24(2), 195-202.
[54] Awerbuch, S., Bazilian, M., & Roques, F. A. (2008). Analytical methods for
energy diversity and security: Portfolio optimization in the energy sector : a
tribute to the work of Dr. Shimon Awerbuch. Amsterdam, Elsevier Science.
[55] Awerbuch, S., & Yang, S. (2007). Efficient electricity generating portfolios
for Europe: maximising energy security and climate change mitigation. EIB
papers, 12(2), 8-37.
107
[56] Markowitz, H. (1952). Portfolio selection. The journal of finance, 7(1), 77-
91.
[57] Rodoulis, N. (2010). Evaluation of Cyprus’ Electricity Generation Planning
Using Mean-Variance Portfolio Theory. Cyprus Economic Policy Review,
4(2), 25-42.
[58] Awerbuch, S., & Berger, M. (2003). Energy security and diversity in the EU:
a mean-variance portfolio approach. IEA Report Number EET/2003/03.
[59] Doherty, R., Outhred, H., & O'Malley, M. (2006). Establishing the role that
wind generation may have in future generation portfolios. Power Systems,
IEEE Transactions on, 21(3), 1415-1422.
[60] Jansen, J. C., Beurskens, L. W. M., & Van Tilburg, X. (2006). Application of
portfolio analysis to the Dutch generating mix. Energy research Center at the
Netherlands (ECN) report C-05-100.
[61] Bates White LLC., (2007). A mean-variance portfolio optimization of
California’s generation mix to 2020: Achieving California’s 33 percent
renewable portfolio standard goal. Report for California Energy Commission
CEC-300-2007-009-D.
[62] Delarue, E., De Jonghe, C., Belmans, R., & D'haeseleer, W. (2011). Applying
portfolio theory to the electricity sector: Energy versus power. Energy
Economics, 33(1), 12-23.
[63] Van, H. J. C., & Wachowicz, J. M. (2008). Fundamentals of financial
management. Prentice Hall, Harlow, England.
[64] International Energy Agency, & Nuclear Energy Agency, Projected Costs of
Generating Electricity, 2010
108

Full-Thesis 9

Uploaded by

Copyright:

Available Formats

Full-Thesis 9

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Full-Thesis 9

Uploaded by

Copyright:

Available Formats

MAINSTREAMING RENEWABLE ENERGY

DEVELOPMENTS INTO TRADITIONAL PLANNING

J.H.K. Kanchana Chathuranga

Degree of Master of Science in Electrical Engineering

Department of Electrical Engineering

J.H.K. Kanchana Chathuranga

Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of

Department of Electrical Engineering

Also, I hereby grant to University of Moratuwa the non-exclusive right to reproduce

Signature of the author Date:

Signature of the supervisor Date:

Signature of the co-supervisor Date:

First and foremost, I would like to express my sincere gratitude to my supervisors

J.H.K. Kanchana Chathuranga

Chapter 2. Mainstreaming NCRE Developments in Sri Lanka: 7

Chapter 3. Review of the CEB Generation Planning Process 15

Chapter 4. Capacity Credit of Renewable Generation and Wind 26

Chapter 5. Modeling and Evaluating Wind Power Plants in WASP IV 38

Chapter 6. Evaluation of the Sri Lankan Generation Mixes Using 50

Chapter 7. Summary and Conclusion 70

Figure 1.1 NCRE Project development from 1999 to 2012 2

AIC Average Incremental Cost

Appendix A Power Curve Values of the Reference Turbine 76

Figure 1.1: NCRE Project development from 1999 to 2012

However, with the increasing levels of penetration and interest in NCRE

The aim of traditional generation expansion planning has been to provide an

1.1 Objectives of the Research

WASP IV package uses a least cost methodology to evaluate different alternative

1.3 Outline of the Thesis

Methods of calculating the capacity credit of wind generation is discussed in Chapter

MAINSTREAMING NCRE DEVELOPMENTS IN SRI LANKA:

2.1 Constraints to Mainstream Adoption of NCRE Technologies in Sri

2.2 Policy and Regulatory Interventions for NCRE Developments

2.2.1 Policy and regulatory interventions in Sri Lanka

2.2.2 Policy lessons from around the world

• Pricing laws and grid access

It is mandatory in some countries for a certain percentage of electricity to be

• Information dissemination, training, education and public awareness

On the contrary, there is a clear lack of standards in Sri Lanka. No institution

• Stakeholder involvement in NCRE developments

REVIEW OF THE CEB GENERATION PLANNING PROCESS

• Macroeconomic data (Electricity demand, Fuel prices, Economic growth, etc)

• Technological innovations (Cost reductions, Efficiency improvements,

3.1 Overview of the CEB Generation Planning Methodology

3.2 Wien Automatic System Planning Package (WASP)

Table 3.1 : Principal capabilities and limitations of WASP IV

Periods per year 12

Cosine terms in the Fourier representation of the load duration 100

Types of plants candidates for system expansion, of which:

Environmental pollutants (materials) 2

System configuration s in all the study period 5,000

3.3.1 Electric Generation Expansion Analysis System

New features, such as demand-side management and economic dispatch under

3.3.2 PLEXOS for Power Systems

In addition to the above mentioned features, PLEXOS has robust renewable

2022 5MW Biomass 1x300 MW Coal power - 0.035

3.5 Identified Issues in the Present Generation Planning Process

3.6 Proposed Concepts to Modify the Existing Planning Process

In addition to modeling a wind plant in WASP IV, a methodology to evaluate the

LOLE′ = P[( C + IG ) < L ]

P( C < L ) = P[( C + IG ) < (L + AL )]

Given that ∑ \\\\\]∗

f[` = Z Zq klm q = Z}` f}` + Z~` f~` + 2Z} Z~ f} f~ x}~