SU+ @ Strathmore 

University Library 
 

  
 

Electronic Theses and Dissertations 

 

 

This work is availed for free and open access by Strathmore University Library.  

It has been accepted for digital distribution by an authorized administrator of SU+ @Strathmore University. 

For more information, please contact library@strathmore.edu 

 

 
2025 

 

Assessment of sustainable electricity 

generation scenarios in Burundi using 

multi-criteria approaches. 

 

 
Igiraneza, Nice Leslie 
School of Computing and Engineering Sciences 
Strathmore University 

 

 

Recommended Citation 

Igiraneza, N. L. (2025). Assessment of sustainable electricity generation scenarios in Burundi using multi-

criteria approaches [Strathmore University]. http://hdl.handle.net/11071/15975 

 
 
 
 
 

Follow this and additional works at: http://hdl.handle.net/11071/15975 

https://su-plus.strathmore.edu/
https://su-plus.strathmore.edu/
http://hdl.handle.net/11071/2474
mailto:library@strathmore.edu
http://hdl.handle.net/11071/15975
http://hdl.handle.net/11071/15975


 

 

ASSESSMENT OF SUSTAINABLE ELECTRICITY GENERATION SCENARIOS IN 

BURUNDI USING MULTI-CRITERIA APPROACHES  
 

 

 

 

 

 

NICE LESLIE IGIRANEZA 

ADMN. NO: 152495 

 

 

 

 

 

 

 

A DISSERTATION SUBMITTED IN PARTIAL FULFILMENT OF THE 

REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN SUSTAINABLE 

ENERGY TRANSITIONS 

 

OF 

STRATHMORE UNIVERSITY 

 

 

2025  

 



i 

 

DECLARATION 

I so certify that my work has never before been accepted as a degree submission by this university or 

any other university. Except for the references that are appropriately listed in the proposal, the 

dissertation, to the best of my knowledge and belief, does not contain any previously published or 

authored works by other authors. No part of this thesis may be reproduced without the author's and 

Strathmore University's consent.  

Signed and submitted 

Candidate ID: 152495   

Sign:   __________________ ________________ 

Date: 1st April, 2025 

 

Approval  

I have given my permission in my capacity as the academic advisor for Nice Leslie Igiraneza's 

submission of this thesis proposal to the School of Computing and Engineering Sciences. 

Signed: 

___________________________________________________________ 

Date: 1st April 2025  

___________________________________________________________ 

Dr.Eng. Fenwicks S. Musonye., P.E., PhD   

Strathmore University, Kenya 

  



ii 

 

DEDICATION 

To my parents, who taught me that nothing is impossible if I put my mind to it. They have also been 

my support system for many years. I am very grateful to my lecturers, who walked this journey with 

us. I praise the Almighty God for his strength, grace, and never-ending love. 

  



iii 

 

ACKNOWLEDGEMENT 

I am truly grateful to everyone who has played a part in making this research possible. 

First and foremost, I want to express my heartfelt appreciation to my academic advisors and 

professors. In particular, I am deeply thankful to Dr. Eng. Fenwicks Musonye for his invaluable 

guidance, insightful and timely feedback, and unwavering support throughout this study. His 

expertise and encouragement have been instrumental in shaping the direction and quality of this 

work. 

A special thank you to the team at REGIDESO for their generosity in providing critical data, which 

was essential to this research. I also want to acknowledge my colleagues and peers for their enriching 

discussions, constant encouragement, and generously shared knowledge that have contributed to this 

process. 

Above all, I want to express my sincere gratitude to my family and friends for their constant 

encouragement, understanding, and support. Their confidence in me has continuously motivated me 

and kept me engaged and focused. 

This research would not have been possible without God’s grace, strength, wisdom, and guidance, 

for which I am eternally grateful. 

  



iv 

 

ABBREVIATIONS 

     

kW  Kilowatt 

kWh  Kilo-watt Hour 

MW  Megawatt 

RE  Renewable Energy 

BAU  Business As Usual 

IRR  Internal Rate of Return 

NACOSTI  National Commission for Science, Technology, and Innovation 

LCOE  Levelized cost of energy 

NREL  National Renewable Energy Laboratory 

PV  Photovoltaic 

CSP  Concentrated Solar Power 

UNFCCC  United Nations Framework Convention for Climate Change 

NDC  Nationally Determined Contributions 

SDG  Sustainable Development Goals 

LEAP  Low Emissions Analysis Platform 

TIMES  The Integrated MARKAL_EFOM System 

GDP  Gross Domestic Product 

NOTC  Nominal Operating Cell Temperature 

STC  Standard Test Conditions 

AC   Alternating Current 

DC  Direct Current 

DoD  Depth Of Discharge 

PSH  Pumped storage hydropower 

AI  Artificial Intelligence 

ML  Machine Learning 

ARIMA  Autoregressive Integrated Moving Average 

AR  Autoregressive 



v 

 

MA  Moving Average 

ANN  Artificial Neural Network 

AFINN  Fuzzy Inference Neural Network 

MTP  Medium-Term Plan 

LTP  Long-Term Plan 

LPG  Liquified Petroleum Gas 

REGIDESO  Water and Electricity Production and Distribution Authority 

MINHEM  Ministry of Hydraulics, Energy and Mines 

WASP  Wien automatic System Planning 

 

  



vi 

 

ABSTRACT 

With an 11% electrification rate, Burundi is one of the countries in the sub-Saharan region still facing 

significant energy access challenges in the region. This has considerably impacted the economic 

development, energy security and technical advancement of the country. The country primarily relies 

on hydroelectric power, with 49MW installed out of a potential 1700MW, as well as diesel thermal 

plants, solar, biomass, peat, firewood, coal, bagasse, although on a smaller scale. Due to a poor energy 

mix and inadequate maintenance of existing hydro infrastructure, technical issues, like insufficient 

capacity, supply disruptions lead to low-quality electrical supply. Moreover, the increasing reliance 

on traditional biomass has led to deforestation, and environmental degradation.  

There have been nevertheless attempted initiatives to close the energy supply and demand imbalance, 

while promoting renewable energy integration. Burundi is boosting energy generation through public 

and private initiatives, including rehabilitating existing hydropower infrastructure, developing rural 

electrification with mini-hydro, solar, and wind power, extending electric networks and building 

regional plants with neighbors. Feasibility studies for solar and wind power investments are also in 

progress. Despite this, there is no data and structured modeling tool available to support evidence-

based decisions about these investments.  

This study aimed to assess the pathways to electricity generation by developing and testing an energy 

planning tool, that integrated available resources and the future energy demand in Burundi, to close 

the existing gap between supply and demand. It will serve as a resource for the relevant stakeholders 

as they tackle the issues of energy access, affordability, security, decarbonization and 

decentralization, and support potential investors in their decision making. 

 

  



vii 

 

Contents 

CHAPTER ONE INTRODUCTION ......................................................................................... 1 

1.1 Background of Study ........................................................................................................... 1 

1.2 Problem Statement ............................................................................................................... 2 

1.3 Research Objectives ............................................................................................................. 2 

1.3.1 General Objectives ........................................................................................................... 2 

1.3.2 Specific objectives ........................................................................................................... 2 

1.3.3 Research Questions .......................................................................................................... 3 

1.4 Justification .......................................................................................................................... 3 

1.5 Scope of Study ..................................................................................................................... 4 

CHAPTER TWO LITERATURE REVIEW ............................................................................. 5 

2.1 Introduction .......................................................................................................................... 5 

2.2 Theoretical Review .............................................................................................................. 5 

2.2.1 Solar PV Technology Fundamentals ............................................................................... 5 

2.2.2 Hydro Technology Fundamentals .................................................................................... 9 

2.2.3 Wind Energy Fundamentals .......................................................................................... 12 

2.2.4 Demand Forecasting Techniques ................................................................................... 14 

2.3 Empirical Review .............................................................................................................. 25 

2.4 Summary of Gaps .............................................................................................................. 28 

CHAPTER 3 METHODOLOGY ............................................................................................ 29 

3.1 Introduction ........................................................................................................................ 29 

3.2 Data Collection .................................................................................................................. 30 

3.3 Demand forecasting ........................................................................................................... 30 

3.4 WASP Model ..................................................................................................................... 32 

3.5 Scenario Description .......................................................................................................... 36 

3.6 Ethical Considerations ....................................................................................................... 37 

CHAPTER 4 RESULTS AND DISCUSSION........................................................................ 38 



viii 

 

4.1 Introduction ........................................................................................................................ 38 

4.2 Energy resource assessment .............................................................................................. 38 

4.3 Peak Load Demand Forecasting ........................................................................................ 39 

4.4 Optimal Expansion Plan Simulation on WASP................................................................. 40 

4.4.1 Input Modules ................................................................................................................ 40 

4.4.2 Output Modules Results ................................................................................................ 42 

CHAPTER 5 CONCLUSION AND RECOMMENDATIONS .............................................. 47 

5.1 Conclusion ......................................................................................................................... 47 

5.2 Recommendations .............................................................................................................. 47 

APPENDICES ......................................................................................................................... 64 

Appendix A: National Commission for Science, Technology and Innovation (NACOSTI) 

research certificate. ........................................................................................................................ 64 

Appendix B: Ethical Review Approval ......................................................................................... 65 

Appendix C: Similarity Index ........................................................................................................ 66 

Appendix D: DYNPRO Final Simulation Results ......................................................................... 67 

  



ix 

 

LIST OF FIGURES 

Figure 2.1 The photovoltaic effect in a solar cell (A. Khalil, 2015) .................................................... 6 

Figure 2.2 Temperature and irradiance effect on cell properties (A. Khalil, 2015) ............................ 7 

Figure 2.3 Grid Tie PV System(Karchi & Kulkarni, 2022) ................................................................ 8 

Figure 2.4 Stand-alone PV System (Fara & Craciunescu, 2017) ........................................................ 8 

Figure 2.5 A hydroelectric power plant and its basic components (I. U. Khalil et al., 2020) ........... 10 

Figure 2.6: Typical wind turbine generator design (Stephanie Cole, 2022) ...................................... 12 

Figure 2.7 Typical wind turbine power output curve (Stephanie Cole, 2022) .................................. 13 

Figure 2.8 Forecasting Method (Schellong, 2011) ............................................................................ 15 

Figure 2.9 Basics of artificial neural network structure (Sobri et al., 2018) ..................................... 18 

Figure 3.1 Flowchart of the methodological approach ...................................................................... 29 

Figure 3.2. User-computer interaction in running the WASP code for a pre-determined expansion 

plan (adapted from ORNL 73-7759 RI) ............................................................................................ 35 

Figure 4.1 BAU DYNPRO Results on the Recommended Number of Units for an Optimal 

Expansion Strategy ............................................................................................................................ 42 

Figure 4.2  DYNPRO Results on the Recommended Number of Units for an Optimal Expansion 

Strategy .............................................................................................................................................. 43 

 

  



x 

 

LIST OF TABLES 

Table 3.1 WASP Modules ................................................................................................................. 33 

Table 3.2 Scenarios Modeled ............................................................................................................ 36 

Table 4.1 Energy Resources in Burundi ............................................................................................ 38 

Table 4.2 Forecasted Peak Load Demand ......................................................................................... 39 

Table 4.3 Existing Power Plants ........................................................................................................ 40 

Table 4.4 Candidate and Committed Power Plants ........................................................................... 41 

 

 



1 

 

CHAPTER ONE INTRODUCTION 

1.1 Background of Study 

Burundi is one of the countries in sub-Saharan Africa still facing major energy access challenges in 

the region. With only roughly 11% of the country's 11 million residents enjoying access to electricity, 

Burundi is among the least electrified countries in the world (Burundi - Lighting Global, 2022). The 

vast majority of people with access to electricity live in cities, whereas most rural households, which 

constitute 86% of the population as of 2022, and some urban regions rely on conventional biomass. 

Seventy percent of these biomass sources come from firewood, with the remaining 18% supplied by 

agricultural waste, 6% coming from coal, and 1% from bagasse (Macumi & Tacer, 2020). The 

excessive use of fuel wood as a primary energy source has a severe environmental impact and 

highlights the need for quick, sustainable energy solutions.  

Burundi relies on hydroelectric power, which has the biggest energy resource potential, with 49MW 

installed out of a total of 1700MW. On a smaller scale, the country also uses diesel-powered thermal 

plants, solar energy, biomass, firewood, peat, bagasse, coal and wind power (Macumi & Tacer, 2020). 

Nevertheless, some of the hydro power plants are on the brink of meltdowns due to inadequate 

maintenance. The energy sector is grappling with low-quality electrical supply, attributed to 

technical, administrative, as well as institutional issues. Some of the technical causes are: inadequate 

installed capacity, an ageing and inefficient transmission and distribution network, and inadequate 

plant maintenance. The administrative and institutional variables include governance frameworks, 

subsidy redistribution schemes, and political influence over public energy supply (Nsabimana, 2020). 

This has an impact on the chances for expanding the nation's energy resources and limits the scope 

for effective policymaking and the design and operation of institutions that produce, sell, and use 

energy. 

There are rising initiatives, both from the public and private sector, to augment energy generation in 

the country. The Burundi government intends to rehabilitate existing non-functioning power plants, 

renovate existing hydroelectric plants, expand both urban and rural electric networks, build regional 

power plants in collaboration with neighboring countries, and develop rural electrification through 

the construction of mini-hydro plants, solar PV, and wind power. Currently, there are also feasibility 

studies ongoing in various areas of the country for hydroelectric and solar photovoltaic power 

exploitation. 

  



2 

 

1.2 Problem Statement 

The prevailing energy landscape in the country is marked by an imbalance between power supply 

and demand. This is particularly evident in insufficient energy access, especially in rural regions, 

which slows down economic growth. It is predominantly reliant on hydro power and traditional 

biomass sources. The output of the former is reliant on seasonal variations, while the latter leads to 

deforestation, and environmental degradation. Nonetheless, attempts have been launched to address 

the energy supply and demand mismatch, including the restoration of existing hydropower 

infrastructure, the development of new hydroelectric sites, and solar mini-grid projects. 

Despite the above-mentioned initiatives, energy access is still low. There have been suggestions to 

fully exploit the existing energy resources and tap into other technologies. For example, Burundi has 

a hydroelectric potential, estimated at 1700MW that remains untapped. Similarly, solar energy 

represents a significant opportunity, with an estimated potential of 2000kWh/m2/year (African 

Development Bank Group, 2020). Further, a recent study found the Bujumbura region to have a great 

potential in wind energy harvesting (Desai et al., 2022). This notwithstanding, there is no available 

data that can support evidence-based decision making on investments in renewable energy in 

Burundi. There is a need to develop planning tools that incorporate all energy market variables, which 

include energy efficiency and all the available resources. This work aimed to develop an Energy 

Resource Planning tool for Burundi. It will serve as a resource for the relevant stakeholders as they 

tackle issues related to energy access, affordability, energy security, decarbonization and 

decentralization, as well as potential investors in their decision making. 

1.3  Research Objectives 

1.3.1 General Objectives 

This research aimed to assess sustainable electricity generation scenarios in Burundi using multi-

criteria approaches.  

1.3.2 Specific objectives 

1. Conduct an energy resource assessment for Burundi. 

2. Forecast energy demand for Burundi. 

3. Model energy investments scenarios considering the demand and supply side characteristics 

for Burundi. 

 

 



3 

 

1.3.3 Research Questions  

 The research will seek to answer the following research inquiries:  

1. What are the available energy resources in Burundi, and how can they be assessed in terms 

of their sustainability, availability, capital expenditure required, operational costs and 

potential to contribute to the national energy mix?  

2. What is the projected energy demand in Burundi over the next 15 years, considering variables 

like population increase, urbanization, economic development, and technological 

improvements? 

3. What cost-effective renewable energy investment scenarios could be pursued to optimize 

resource use? 

1.4 Justification 

Developing an Energy Resource Planning tool for Burundi is key due to the current energy insecurity, 

environmental degradation, and socio-economic challenges. The country's reliance on intermittent 

hydro power plants and traditional biomass worsens energy poverty, while inefficient energy 

practices contribute to environmental harm. Access to reliable, affordable, and sustainable energy is 

essential for promoting economic growth, achieving social development goals (SDGs), and 

mitigating climate change risks. 

As part of the efforts to boost the attainment of the United Nation’s Sustainable Development Goal 

7, to “ensure access to affordable, reliable and modern energy for all by 2030”(Team & Roser, 2023), 

Burundi has set itself the target of increasing the national rate of access to electricity to 30% by 2030 

(Ministère de l’Energie et des Mines, Burundi, 2013). These research findings will serve as a roadmap 

to utility planners and the government in their attempts to maintain an optimal energy mix, while 

promoting energy efficiency to achieve environmental and economic goals. It can also be used by 

the Ministry of the Environment in prioritizing programs and policy options. 

Burundi, as a participant in United Nations Framework Convention for Climate Change (UNFCCC), 

submitted its NDCs (Nationally Determined Contributions) in 2021. According to NDC, (2021), the 

government outlined its goal of the battle against climate change as follows: "A state that promotes 

development that is resilient to the harmful effects of climate change.". The updated submitted NDC 

has an unconditional emissions reduction target of 3.04% and a conditional target of 12.61% by 2030, 

as part of their commitment to lower greenhouse gas emissions. Additionally, the carbon market 

presents itself as a source of revenue through carbon financing. This study will assist policy makers 

and government officials in coming up with carbon friendly guidelines and protocols while 

maintaining a stable grid. 



4 

 

1.5 Scope of Study 

This research focused on designing an energy planning tool for sustainable power generation in 

Burundi and aimed to develop a model that will guide potential energy investors, private and public 

on the factors to consider as they assess the performance rate and economic viability of power 

generation projects. This study assumed that energy transportation will be done via the forthcoming 

transmission lines and considered technical losses due to transmission as negligible. 

  



5 

 

CHAPTER TWO LITERATURE REVIEW 

2.1 Introduction 

This literature review chapter delves into a comprehensive exploration of both theoretical 

frameworks and empirical studies relevant to this research. In the first section, a theoretical review 

explores the fundamental concepts of the main intended electricity sources for exploitation in 

Burundi, namely solar, hydro, and wind. This section also introduces the carious existing energy 

demand forecasting models. The second section presents an empirical review that highlights a few 

studies that have been done within the area of research. The aim is to identify the study gaps, limits, 

and areas for further exploration. The last part of this chapter includes a summary of these study gaps, 

providing grounds for investigating the research problem at hand. 

2.2 Theoretical Review 

2.2.1 Solar PV Technology Fundamentals 

The environmental advantages and falling costs of solar energy have made it a popular renewable 

and sustainable energy source that is reducing dependency on fossil fuels. It is, however, an 

intermittent energy source and is usually integrated with other power sources. Despite the recent 

considerable price reductions, the initial investment for installation is still relatively high. On the 

other hand, compared to conventional energy sources, solar modules present significantly lower 

operation costs and demand less maintenance after installation. Rooftop, ground, or floating 

installations are possible for solar PV panels. Large-scale solar installations are mostly ground 

mounted and require significant land area, which may not be feasible in densely populated or 

metropolitan locations and could result in disputes over land use. In some cases, solar projects may 

intrude on agricultural land, natural habitats, or territories belonging to indigenous communities, 

raising concerns about environmental justice and the rights of affected communities.  

The process of transforming light (photons) into electricity (voltage) is known as the solar 

photovoltaic effect, and it is how solar energy is captured. A voltage or electric current is produced 

in the cell upon exposure to sunlight. Fig. 1 is a representation of several solar PV cells, which, when 

combined form a PV panel. 



6 

 

 

Figure 2.1 The photovoltaic effect in a solar cell (A. Khalil, 2015) 

Solar cells are up of two different types of semiconductors, p-type and n-type, connected to form a 

p-n junction. When the two are joined, this junction produces an electric field that makes it easier for 

electrons to move freely, traveling toward the positive side (p) and holes toward the negative side 

(n). As a result, particles with a positive charge migrate in the opposite direction from the ones with 

a negative charge due to this field. The cell generates an electric current as a result of this operation. 

The size of individual cells and the intensity of the light affect this current (A. Khalil, 2015). 

Photovoltaic cells are available in many different semiconductor materials. The most prominent 

materials are copper-indium selenide, amorphous silicon, polycrystalline silicon, and 

monocrystalline silicon. Depending on the voltage or current required, the individual solar cells are 

usually combined in series and in parallel. 

To increase the power output of a PV array, solar panels are also interconnected in series or parallel. 

When multiple panels are connected in series, they produce a PV module string, with the positive 

and negative terminals of each other interconnected, increasing the voltage of the PV array. The 

number of panels on the string is determined by the required voltage level. A parallel connection, on 

the other hand, increases the current by connecting the negative and positive terminals of one module 

to the respective modules' negative and positive terminals. The necessary current level for the system 

is then obtained by connecting numerous PV strings in parallel (Yusof & Baharuddin, 2020). The 

output of a solar PV system also depends on a few other factors, such as solar irradiance, temperature, 

panels mounting tilt and orientation, shading, system design and components, among others.  



7 

 

2.2.1.1 Factors Affecting the Output of Solar PV Systems. 

Solar irradiance is referred to as the power of solar radiation collected from the sun, per unit area at 

a specific location on Earth's surface. It is a gauge of how much sunlight is reaching the surface of 

the Earth at any one time. Solar irradiance is measured perpendicular to the direction of sunlight and 

is commonly given in watts per square meter (W/m²). Seasons, air quality, and sun angle are some 

of the variables that can cause variations in solar irradiance values during the day. A solar PV panel's 

current rises proportionately to increased irradiance (Majid et al., 2013)  

Solar panels also operate less efficiently at higher temperatures. While sunlight is essential for solar 

PV production, excessively high temperatures can reduce the efficiency of the panels, affecting 

electricity output. Similarly, an increase in temperature at the cell level affects the PV module output. 

The industry standard test settings known as Standard Test conditions (STC) are used to evaluate all 

solar PV panels in order to ascertain their rated power and other attributes. Nominal Operating Cell 

Temperature (NOCT) is a standard measurement used in the solar energy industry to rate the 

operating temperature of photovoltaic (PV) modules under specific settings. These are a standard 

sunlight intensity of 800 W/m², 20°C air temperature and 1m/s wind conditions for NOTC and cell 

25°C temperature, 1000 W/m² irradiance and air mass value of 1.5 for STC. Tsai & Tsai (2012) 

further differentiates the NOTC and STC. 

 

Figure 2.2 Temperature and irradiance effect on cell properties (A. Khalil, 2015) 

Figure 2(a) illustrates how a rise in solar radiation causes the VOC (open circuit voltage) to rise and 

the ISC (short circuit current) to rise linearly. The effects of temperature on cell characteristics are 

shown in Figure 2(b). Reduced efficiency results from the open circuit voltage decreasing linearly 



8 

 

with increasing cell temperature. A slight increase in short circuit current is observed when the cell 

temperature rises (A. Khalil, 2015). 

Other factors that affect a solar PV array's production include azimuth, tilt, and shading. Reduced 

sunlight exposure from buildings, trees, and other barriers can lower the electricity production of 

solar modules. When connecting a solar module, two main angles are considered to have an impact 

on PV module performance. These are the module's tilt angle and orientation angle, or azimuth. 

Generally, solar panels should be positioned to face the true south for sites located in the northern 

hemisphere and true north for the ones in the southern hemisphere. The mounting location's angle 

with respect to the equator should serve as guidance for the panels' tilt (Dewi et al., 2019). 

2.2.1.2 Components of a Solar PV System 

In addition to solar panels, the second component of a typical solar system is an inverter. Inverters in 

PV applications fall under two categories: 

• Grid-tie inverters in grid tied systems. 

• Hybrid/ Stand-alone inverters in off grid systems.

 

Figure 2.3 Grid Tie PV System(Karchi & 

Kulkarni, 2022) 

  

 

Figure 2.4 Stand-alone PV System (Fara & 

Craciunescu, 2017) 

Figure 3 is a schematic for a simple grid-tied PV installation. The DC electricity from the solar panels 

is fed into an inverter, which converts it into alternating current (AC) electricity. For such systems, 

the grid-tied inverter is connected to the utility grid, which allows the system to inject surplus 

electricity generated by the solar PV array back into the grid, where that option is available. These 

systems do not typically incorporate energy storage, relying instead on the grid to provide power 

when sunlight is insufficient.  

In contrast, standalone PV systems, also known as off-grid systems, hybrid/ stand-alone inverters 

operate independently of the grid and allow integration with battery storage to store surplus energy 

produced during sunny periods for use during periods of low or no solar production. A typical stand-

alone system should also have an AC engine generator to supply power to the AC loads and allow 

the batteries to charge when the PV system is producing less than the required demand (Mayfield, 



9 

 

2010). Standalone systems are commonly used in remote areas where grid connection is impractical 

or expensive, providing a reliable and sustainable source of electricity. 

A battery bank makes up the third important part. Batteries keep energy in reserve to provide energy 

throughout the night or on autonomous days (also known as no-sun days or dark days) when the sun 

does not provide enough radiation. Battery autonomy, which refers to how long a battery can operate 

without being recharged by solar panels or another source matters when sizing the battery capacity 

required in a solar system. It depends on factors such as battery capacity, energy consumption, solar 

generation, depth of discharge (DoD), efficiency losses, and temperature. These batteries, which can 

be lead-acid, lithium-ion, Nickel-Cadmium or Ni-Metal hydride batteries are made to be 

progressively discharged and recharged hundreds of times up to 80% of their capacity. (Al-Shamani, 

2015). Lifetime depends on factors such as the number of charging and discharging cycles, 

temperature, and other relevant parameters. 

2.2.2 Hydro Technology Fundamentals 

According to Singh & Singal (2017) research, impulse momentum is the basic idea underlying the 

production of hydropower. Through a turbine's rotation, water potential is transformed into 

mechanical energy. A generator is then used to further convert this mechanical energy into electrical 

energy. Because they can produce continuously as long as there is a consistent supply of water and 

can react quickly to changes in the demand for electricity, hydroelectric power plants provide 

operational flexibility. Because of this, hydroelectric plants are well-suited to deliver base load power 

since they can continuously produce electricity to fulfill the grid's minimal need (Lejeune, 2012).  

Hydroelectric power generation is regarded as cost-effective in terms of cost per kilowatt-hour (kWh) 

in comparison to other RE sources including geothermal, biomass, solar, and wind (I. U. Khalil et 

al., 2020). On the flip side, hydro projects, even those considered nominally successful, frequently 

come with adverse environmental impacts. These can manifest as loss of biodiversity, interference 

with fish migration patterns, significant land inundation, displacement of human communities, and 

various other consequences (Tahseen & Karney, 2017). 

Fig 2.5 presents the basic components of a hydro powerplant.  



10 

 

 

Figure 2.5 A hydroelectric power plant and its basic components (I. U. Khalil et al., 2020) 

The process of producing hydroelectric electricity entails gathering or storing water at one location 

and directing it downward to a lower level via enormous pipes or tunnels known as penstocks. The 

head is the vertical distance that separates these two altitudes. When the water reaches the bottom of 

its course through the pipes, it causes turbines to rotate. Generators are then powered by the turbines, 

converting the mechanical energy from the turbines into electrical energy. The alternating voltage 

produced by the generators is stepped up by transformers for long-distance transmission. The 

powerhouse is the building that houses the generators and turbines and provides access for the pipes 

or penstocks (Singh & Singal, 2017).  

There are three categories of hydropower plants: impoundment, diversion or "run of the river," and 

pumped storage. The latter is the most prevalent kind of hydroelectric power plant; it is usually a 

large hydroelectric system that stores river water in a reservoir using a dam. When the reservoir's 

water is discharged, it passes through a rotating turbine, starting a generator that generates energy 

(Tkáč, 2018). 

A diversion or run-of-the-river facility for hydropower plants relies on a river’s natural flow rather 

than a big reservoir. These hydroelectric facilities direct a portion of the water flow either through a 

penstock, or through a canal. Diversion facilities have the option to utilize a dam for the purpose of 

redirecting water, although it is not mandatory. While they may possess a limited capacity for storage, 

known as pondage, their ability to store water is far lower compared to an impoundment facility. 

Typically, storage only affects short-term changes in water flow and does not significantly impact 

the flow of rivers downstream (Tsuanyo et al., 2023).  



11 

 

Pumped storage hydropower is a large energy storage mechanism. Such a plant can store electricity 

generated from non-renewable energy sources, like solar, wind, and nuclear power, for later use. 

Pumped hydro storage facilities store energy by using a system of two connected reservoirs, one 

higher than the other in altitude. Water is pumped to the upper reservoir when there is an excess of 

energy available. On the other hand, in periods of excess demand, water is released from the higher 

to the lower reservoir via reversible Francis turbines, producing energy along the way. The process 

is then repeated, resulting in a cycle efficiency of about 80% overall. Pumped-storage facilities are 

highly suitable for effectively handling surges in energy demand and ensuring backup power 

generation (Lejeune, 2012). 

2.2.2.1 Hydropower Systems Design 

A hydroelectric project necessitates the availability of a river. The energy that can be extracted from 

the river is subject to two factors: the water’s flow rate of and the vertical distance between the 

riverbed and the water surface, commonly referred to as the head of water. According to Breeze, 

(2005) a river with a high flow rate would generate more electricity compared to a river with a low 

flow rate, even if they are of the same size. 

After identifying a suitable location for a hydropower project, there are typically two methods for 

harnessing its potential. One approach is to construct a dam and establish a reservoir behind it, which 

will be used to extract water for powering hydraulic turbines in the powerhouse. The other approach, 

known as a run-of-river plan, operates with no need for a reservoir, but it typically includes a barrage 

of some kind. Instead, it directly extracts water from the river and transports it to the powerhouse, 

where the turbines are erected (Breeze, 2005). 

The main components required to convert the potential energy of the river into electricity are turbines, 

which transform kinetic and potential energy into mechanical energy, and a generator, which 

transforms mechanical energy into electrical energy. The monthly hydropower output can be 

determined using an equation that takes into account the water volume released, gross head, and 

efficiency of the turbine generator, which typically ranges between 0.7 and 0.85 (Jeyalitha et al., 

2008). 

 

 



12 

 

2.2.3 Wind Energy Fundamentals 

Kalmikov (2017) defines wind as atmospheric air in motion. It is caused by temperature fluctuations 

in different places. The temperature properties of land and sea differ due to their respective 

compositions. Generally, there is more wind in the water than on land. Wind turbines convert wind 

power into electricity.  

A wind turbine converts the wind's kinetic energy into rotational motion in the shaft, generating 

electricity. The amount of power that can be harnessed from the wind is determined by both its 

velocity and the area covered by the turbine's moving blades. Hence, a higher wind speed or larger 

rotor blades result in greater wind energy extraction (Stephanie Cole, 2022). Figure 6 illustrates the 

primary components of a wind turbine, including the tower, the rotor blades, high speed shaft, drive 

shaft, generator, and gearbox. 

 

Figure 2.6: Typical wind turbine generator design (Stephanie Cole, 2022) 

The wind speed fluctuates hourly, although the overall average remains constant throughout the day. 

The average wind velocity is utilized for the analysis of performance and power output. Wind power 

is usually exploited in conjunction with another power source to provide a consistent production 

(Zafar, 2018). The efficiency and safety of the wind turbine are determined by the wind speed. There 

are three different types of wind speeds used to evaluate performance and reliability of a wind turbine. 

These are: the cut-in speed, rated output wind speed and cut-out speed. 

The cut-in speed refers to the threshold velocity at which a wind turbine’s blades initiate rotation and 

subsequently produce electrical energy. Wind turbine blades are unable to function at extremely low 

wind speeds because of their substantial weight and insufficient power supply. Nevertheless, as the 



13 

 

velocity increases, the turbine produces enough torque to revolve and generate electricity. The normal 

cut-in speed ranges between 3 and 4 m/s. (Stephanie Cole, 2021) 

The electrical output power quickly increases as the wind speed surpasses the cut-in speed. 

Nonetheless, the power output reaches the electrical generator's maximal capacity between 12 and 

17 m/s. The generator's rated power output is its maximum power output, and the rated output wind 

speed is the wind speed at which this maximum output is reached (Stephanie Cole, 2021).  

The maximum speed at which a wind turbine can function is known as the cut-out speed. Beyond 

this point, the turbine is exposed to extraordinarily strong stresses that could jeopardize its stability. 

Turbines are usually equipped with a braking system that can be operated manually or automatically. 

The braking system kicks in and brings the object to a stop when the speed hits the cut-out threshold. 

About 25 m/s is the usual threshold velocity, according to Stephanie Cole, (2021). 

 

Figure 2.7 Typical wind turbine power output curve (Stephanie Cole, 2022) 

Pfaffel et al., (2017) notes a few performance indicators of wind turbines. The most common metric, 

the capacity factor, represents the proportion of the turbines' actual electrical output over a specific 

timeframe to its ideal (or rated) power output.  It depends heavily on the available wind conditions. 

Among other performance indicators the author highlights are the time-based availability, energetic 

availability, technical availability, failure rate, and average down time. 

Betz's Limit, which establishes the highest theoretical efficiency of a wind turbine, is a key concept 

in wind energy conversion. According to this theory, which was put forth by German physicist Albert 

Betz in 1919, no wind turbine is capable of converting more than 59.3% of the wind's kinetic energy 



14 

 

into mechanical energy. The reason for this limit is that if all of the wind's energy were extracted, it 

would cease to move at all, which would halt more air from passing through the turbine. Depending 

on variables including wind speed, system losses, and blade design, contemporary wind turbines can 

reach efficiencies of anywhere between 35% and 45% in real-world applications. In order to 

maximize wind turbine performance and raise the viability of wind energy systems overall, it is 

essential to comprehend Betz's Limit. 

Wind resources can be classified in two main categories based on their geographical location: onshore 

and offshore wind power. Offshore wind power refers to the production of energy using wind farms 

placed in bodies of water, usually in the sea, whereas onshore wind farms are normally constructed 

on land. The reason offshore farms produce more power is because wind speeds are higher offshore 

compared to onshore. Because of their location, offshore conversion technologies need a more 

durable bearing structure than onshore systems. Additionally, the transmission of electric power 

requires the use of specialized cables, and equipment both for installation and maintenance, which 

makes offshore wind power much more expensive compared to onshore wind power. However, 

offshore wind farms are less impactful on the environment and people, making them less 

controversial. (Desalegn et al., 2023).  

2.2.4 Demand Forecasting Techniques 

Forecasting involves predicting future trends by analyzing patterns in past and present data. It 

consists of three key components: input variables, which encompass historical and current data; 

forecasting methods, which rely on trend analysis; and output variables, which represent future 

projections. As illustrated in Figure 2.8, a typical energy demand forecasting process begins with a 

well-organized and accessible historical energy database, serving as the foundation for accurate 

predictions. 



15 

 

 

Figure 2.8 Forecasting Method (Schellong, 2011) 

The database should contain energy demand data and other factors influencing energy use. This data 

is fed into a forecast model, which generates a prediction of energy demand. This is typically an 

ongoing process, in which new data is added to update the energy database and continually provide 

new forecasts with time. 

2.2.4.1 Forecasting Methods 

The choice of the ideal forecasting technique is dependent on the availability of data, the objectives 

of the forecasting tool, and the planning processes. Debnath & Mourshed (2018) categorize demand 

forecasting strategies into two primary groups: stand-alone and hybrid methods. This classification 

is based on the number of methodologies employed for pattern analysis. Stand-alone approaches 

employ a singular strategy for analyzing trends, while hybrid methods incorporate multiple stand-

alone techniques. Hybridization is primarily used to streamline and enhance the accuracy of forecast 

output, resulting in more dependable projections.  

Energy demand forecasting methods include a variety of techniques utilized to predict future energy 

consumption patterns, crucial for effective energy resource planning, infrastructure development, and 

policy formulation. Conventional tools include time series analysis, such as regression analysis and 

moving averages, which examine historical consumption data to detect trends and patterns. 

Econometric models integrate economic indicators and demographic variables to forecast energy 

demand based on economic activity and population changes. Additionally, advanced techniques like 

artificial intelligence and machine learning algorithms utilize complex data sets to enhance accuracy 

and adaptability in forecasting. These methods account for dynamic factors such as weather patterns, 

technological advancements, and policy shifts, enabling more robust and precise predictions of 



16 

 

energy demand at various temporal and spatial scales, essential for informed decision-making in the 

energy sector. 

Suganthi & Samuel (2012) categorize energy demand forecasting models into various categories. In 

the field of economics, different models are used, including regression, time series, econometric, 

cointegration, and decomposition models. Other approaches include AutoRegressive Integrated 

Moving Average (ARIMA) models, artificial systems such as expert systems and Artificial Neural 

Network (ANN) models, grey prediction models, input-output models, and fuzzy logic models. 

Additionally, integrated models such as support vector regression, autoregressive models, and 

particle swarm optimization models are utilized. Lastly, bottom-up models like MARKAL, The 

Integrated MARKAL-EFOM System (TIMES), and the Long-range Energy Alternatives Planning 

System (LEAP) are also employed. 

A time series energy demand forecasting model utilizes historical energy consumption data to predict 

future demand patterns based on observed trends and seasonality. This model usually employs 

numerical approaches such as autoregressive integrated moving average (ARIMA) or exponential 

smoothing to discover and extrapolate trends in time series data. The model can catch repeating 

trends, such as daily and seasonal variations, and project them into the future by analyzing previous 

energy consumption data at regular time intervals (hourly, daily, monthly). Additionally, integrating 

exogenous variables like weather data or economic indicators can improve prediction accuracy by 

accounting for external factors that influence energy demand changes (Ghalehkhondabi et al., 2017). 

A study by García-Ascanio & Maté (2010) shows that using interval time series forecasting models 

can lower operational risk and improve power system planning. 

Regression models establish a forecasting function by computing the value of a dependent variable 

using one or multiple independent variables. The terms "response variable" and "predictor variable" 

refer to the dependent and independent variables, respectively. The Simple linear regression predicts 

the Y variation as X varies by a specific amount. With regression, the Y variable is predicted from X 

using a linear equation: Y=b0+b1X. In this equation, b0 is the intercept (or constant), and b1 is the 

slope of X. Because it is dependent on X, the Y variable is referred to as the response. The X variable 

is either the independent variable or the predictor. The machine learning experts commonly refers to 

Y as the target and X as the feature (Peter Bruce, Andrew Bruce, and Peter Gedeck, 2020). 

Econometric models for energy demand forecasting use economic theory and statistical methods to 

assess the link between energy use and major economic variables including GDP, industrial output, 

and population growth. These models typically utilize time-series or regression analysis to identify 



17 

 

the effect of economic factors on energy demand. By incorporating key determinants of energy 

consumption, econometric models provide insights into the underlying drivers of energy demand and 

enable policymakers to forecast future energy consumption levels under different economic 

scenarios. Rao & Parikh (1996) demonstrated that econometric models are ideal in energy demand 

forecasting for developing countries. 

AutoRegressive Integrated Moving Average (ARIMA) models are mostly utilized to forecast energy 

demand because of their ability to detect temporal trends and seasonality in past energy consumption 

data. In an ARIMA model, the historical values of the energy demand series are regressed on lagged 

values and error terms, which include autoregressive (AR) and moving average (MA) modules. The 

"integrated" component represents differencing to attain stationarity in time series data. Fattah et al. 

(2018) argue that when the seasonal adjustment order is large or diagnostics fail to prove that the 

time series is stable after seasonal adjustment, the ARIMA approach becomes expensive and, in many 

circumstances, impossible to create a model. In such instances, the static variables of the standard 

ARIMA model are viewed as the primary barrier to forecasting highly fluctuating seasonal demand. 

Another limitation of the standard ARIMA approach is that it necessitates a high number of 

observations to obtain the best fit model for a given dataset. 

 Olivencia Polo et al., (2015) by the form and function of biological neural networks that learn 

patterns and relationships from historical energy consumption data via training methods while 

dealing with complicated non-linear behaviors. Figure 8 illustrates the basics of ANN, which 

comprises of three distinct layers: input layer, hidden layer, and output layer. 

 



18 

 

 

Figure 2.9 Basics of artificial neural network structure (Sobri et al., 2018) 

Artificial Neural Network (ANN) models are increasingly utilized in demand forecasting due to their 

ability to handle complex data as well as nonlinear relationships (MALCOLM et al., 1999).These 

models can capture complex interactions between various factors influencing energy demand, 

including socio-economic variables, weather patterns, and technological advancements. These are 

then utilized to predict productivity, giving the ability to foresee current and potential problems 

(including fault, failure, production losses, etc.) based on data collected from sensors (Ferrero 

Bermejo et al., 2019). This provides more accurate and adaptive forecasts compared to traditional 

statistical methods. According to Miller et al. (1990), one of the key disadvantages of ANN models 

is the necessity for extensive information and sufficient data quality, which is not frequently available 

or affordable. 

The Grey Prediction Model, a subset of Grey System Theory, offers a forecasting technique 

particularly suitable for situations with limited data or when traditional statistical methods may not 

be applicable. This notion is referred to as "grey" due to the existence of unclear data. In energy 

demand forecasting, the Grey Prediction Model utilizes a small dataset to establish the underlying 

relationships between energy consumption and influential factors (Julong, 2018.). Grey prediction is 

commonly employed due to its attribute of higher forecasting accuracy than other methods and the 

typically lower requirement in terms of data pieces to develop a forecasting model. It employs a Grey 

Differential Equation to model the dynamic evolution of energy demand over time, effectively 

capturing both linear and nonlinear trends. 



19 

 

Fuzzy sets are used to make decisions based on ambiguous data, interact with human cognitive 

processes, and assess uncertainty at different stages of a process. Fuzzy sets theory allows systems 

to express their principles using "if-then" statements, eliminating the requirement for mathematical 

analysis in modelling (Haji & Assadi, 2009). A study by Iyatomi & Hagiwara (2004) proves that 

fuzzy logic use fewer analytical observations than a number of other approaches and yields 

anticipated values from imperfect data, however their output is not always satisfactory. When the 

number of input dimensions increases, conventional fuzzy systems frequently fail to complete the 

task effectively. In the same study, Iyatomi & Hagiwara, (2004) offer an adaptive fuzzy inference 

neural network (AFINN) that can construct itself, estimate parameters, and extract rules. Fuzzy sets 

enable modelers to condense large quantities of data into fewer variable rules, which are then used 

by fuzzy forecasting models. 

Forecasting is an indispensable tool, even in operations management, enabling organizations to 

anticipate future demand and optimize decision-making processes. The Forecasting Fundamentals, a 

study on Mighty Mechanical provides a systematic exploration of forecasting techniques, 

emphasizing their application in demand prediction. A critical distinction is drawn between 

qualitative methods, which rely on expert judgment and subjective inputs, and quantitative methods, 

which leverage historical data and mathematical models for objective analysis. Given the operational 

focus of demand forecasting, quantitative approaches, particularly time-series models, are often 

prioritized for their ability to extract meaningful patterns from past. 

Time-series forecasting encompasses several techniques, each suited to different demand 

characteristics. The naïve method, for instance, uses the most recent observation as the forecast, 

offering simplicity but limited accuracy in volatile environments. Conversely, moving averages 

smooth out fluctuations by averaging past data points, with the option to weight recent observations 

more heavily, using weighted moving averages or equally as in simple moving averages. Exponential 

smoothing further refines this approach by applying exponentially decreasing weights to older data, 

controlled by a smoothing constant, α. Lower α values enhance stability, while higher values improve 

responsiveness to recent changes. Trend projection, a regression-based technique, is highlighted for 

its effectiveness in modeling steady growth or decline, and this is done through an equation which 

forecasts demand based on a linear trend. 

The study also highlights a significant challenge in forecasting which is accounting for seasonality. 

The study addresses this through decomposition and seasonal indices. By isolating trend, seasonal, 

cyclical, and random components, forecasts can be adjusted to reflect periodic demand variations. 

For example, quarterly forecasts derived from annual projections are refined using seasonal indices 



20 

 

such as 0.80 for Q1, 1.20 for Q2, ensuring alignment with historical patterns. This approach is 

contrasted with applying separate trend lines to each quarter’s data, illustrating how methodological 

choices influence forecast precision. 

Finally, the study also underscores the importance of evaluating forecast accuracy through metrics 

such as Mean Absolute Deviation (MAD), Mean Squared Error (MSE), and Mean Absolute Percent 

Error (MAPE). MAD measures average error magnitude, while MSE penalizes larger errors more 

severely, making it useful for high-stakes applications. MAPE, expressed as a percentage, facilitates 

cross-comparisons across datasets. A hypothetical comparison of two forecasting methods reveals 

that identical MFE or MAD values may mask critical differences in consistency, which MSE and 

MAPE can uncover. This aligns with broader literature emphasizing the need for multi-metric 

assessment to balance bias, precision, and practicality (Hyndman & Koehler, 2006). 

2.2.4.2 Energy Modeling Tools 

The transition toward renewable energy systems necessitates robust modeling tools to address 

challenges such as grid stability, variable renewable energy (VRE) integration, and decarbonization 

pathways. The flowchart as illustrated in Figure 2.10 provides a structured framework for selecting 

an appropriate energy modeling tool based on various criteria. It begins with the general logic of 

available models, which are categorized by approach, purpose, and methodology. The resolution of 

the model is then considered in both temporal and spatial dimensions. Additionally, the inclusion of 

technical and economic parameters, such as conventional and renewable generation, energy storage, 

grid management, demand elasticity, and market factors, plays a crucial role in model selection. 

Finally, the availability of the model—whether it is open-source, commercial, or defined by a specific 

software type—guides the final model choice. 



21 

 

 

Figure 2.10 Classification for model choice, adapted from (Ringkjøb et al., 2018) 

 

Sommer, (2023) highlights several open-source and commercial tools that facilitate these analyses, 

each with distinct methodological approaches and applications. Among these, Python for Power 

System Analysis (PyPSA) emerges as a prominent tool for optimizing power systems with high 



22 

 

shares of VRE. Utilizing linear programming (LP), PyPSA enables cost-effective grid expansion and 

sector coupling, such as electricity, heat, hydrogen, while maintaining hourly temporal resolution. 

However, its limitations in non-linear power flow analysis restrict its utility for detailed grid stability 

assessments (Brown et al., 2018).  

For technical grid analysis, pandapower offers specialized capabilities, including non-linear power 

flow calculations and short-circuit analysis, making it indispensable for distribution system studies. 

Its open-source nature and Python integration enhance accessibility, though its focus on operational 

rather than long-term economic modeling narrows its scope (Thurner et al., 2018). In contrast, 

RAPSim employs agent-based simulation (ABS) to model microgrids and decentralized energy 

systems, providing granular insights into community-scale renewable integration. While effective for 

localized projects, its scalability to national grids remains constrained (Pochacker et al., 2014). A 

comparative evaluation reveals that PyPSA and pandapower are particularly synergistic: PyPSA 

excels in economic optimization and renewable integration, while pandapower addresses technical 

grid constraints. 

Commercial and hybrid tools such as OpenDSS complement these open-source options by offering 

advanced distribution system simulations, including quasi-static time-series analysis for smart grids. 

Despite its robustness, OpenDSS's complexity and partial open-source availability pose barriers to 

widespread adoption (D. Montenegro et al., 2022). GridLAB-D, another open-source tool, focuses 

on demand-side modeling with sub-second resolution, incorporating behavioral and weather data to 

simulate residential and commercial energy use. While valuable for DSM studies, its limited spatial 

scope reduces its applicability to transmission-scale analyses (GridLAB-D, 2022). 

Ringkjøb et al., (2018) reviewed different modeling tools encompassing a diverse array of approaches 

tailored to address the complexities of energy and electricity systems with high shares of variable 

renewable energy sources (VRES). These tools are broadly categorized into power system analysis 

tools, operation and investment decision-support models, and long-term scenario analysis 

frameworks. Power system analysis tools, such as CYME, DIgSILENT PowerFactory, and 

HYPERSIM, focus on granular temporal resolutions to simulate grid stability, power flows, and 

dynamic responses, making them indispensable for assessing short-term technical challenges like 

frequency regulation and fault management. In contrast, operation and investment-oriented models, 

including BALMOREL, EMPS, and PLEXOS, employ optimization techniques such as linear or 

mixed-integer programming to balance supply-demand dynamics, optimize unit commitment, and 

evaluate capacity expansion over decadal horizons, often incorporating stochastic methods to account 

for renewable variability. 



23 

 

The author adds that Long-term energy system models, such as TIMES, GCAM, and MESSAGE, 

adopt a broader perspective, integrating macroeconomic and policy-driven scenarios with 

technological pathways to explore decarbonization strategies. These models typically use time-

slicing or representative periods to manage computational constraints, though this simplification risks 

underestimating the operational flexibility required for high VRES penetration. Hybrid tools like 

OSeMOSYS and Calliope bridge this gap by coupling high-resolution operational constraints with 

long-term planning, while agent-based models (e.g., EMLab-Generation) simulate market actor 

behavior to capture emergent dynamics in liberalized energy markets. 

 

One of the most sophisticated tools for utility-scale energy planning is PLEXOS, a power system 

optimization model that employs mixed-integer linear programming (MILP) for generation dispatch 

and expansion planning. PLEXOS is widely used by utilities and market operators for simulating 

power generation, fuel dispatch, and energy market interactions. It allows for a comprehensive 

analysis of thermal and renewable energy systems while optimizing transmission and storage 

deployment. Its ability to integrate multiple time scales and policy constraints makes it suitable for 

long-term energy planning and operational forecasting (Energy Exemplar, 2021). 

Another widely utilized utility-scale energy system model is the Model for Energy Supply Strategy 

Alternatives and their General Environmental Impact (MESSAGE). Developed by the International 

Atomic Energy Agency (IAEA), MESSAGE is a long-term optimization tool that enables planners 

to evaluate investment strategies for power generation infrastructure. It is particularly effective for 

multi-sectoral energy planning, as it incorporates detailed cost-benefit analyses for different 

expansion scenarios. The model also supports decision-making related to the integration of renewable 

energy and conventional power sources (Isik, 2016). 

Similarly, the Wien Automatic System Planning Package (WASP) is an established tool used for 

least-cost generation expansion planning, particularly in developing countries. Developed by the 

IAEA, WASP utilizes dynamic programming to determine the most cost-effective expansion strategy 

while ensuring system reliability and adequacy. The model is particularly effective for hydro-thermal 

power planning, as it allows for the evaluation of different capacity expansion options under varying 

demand and resource constraints (IAEA, 2001). 

The Integrated MARKAL-EFOM System (TIMES) is another widely applied optimization tool that 

extends the capabilities of the original MARKAL model. TIMES is based on a bottom-up approach 

and is used for national and regional energy system planning. The model integrates economic 



24 

 

optimization techniques to assess energy technology evolution, policy impacts, and carbon pricing 

mechanisms. TIMES is particularly useful for assessing long-term energy transitions, given its ability 

to incorporate various technological pathways and demand-side measures (Ryder et al., 2023) 

For researchers and policymakers seeking an open-source alternative, Balmorel provides a flexible 

modeling framework for electricity and district heating system analysis. This tool employs MILP 

optimization techniques to determine the least-cost expansion of generation, transmission, and 

storage infrastructure. Balmorel is particularly advantageous for cross-border power system studies, 

as it accounts for market interactions and regulatory constraints across multiple regions (Wiese et al., 

2018). 

Finally, the Long-range Energy Alternatives Planning System (LEAP) offers a scenario-based 

approach to energy system modeling. Unlike optimization-driven models, LEAP employs simulation 

techniques to project energy demand, supply, and emissions under different policy scenarios. It is 

mostly used in developing countries for national energy planning, particularly in assessing the long-

term impacts of energy policies and sustainability measures. Its user-friendly interface and 

adaptability make it a preferred tool for government agencies and research institutions involved in 

low-carbon energy planning (Davis et al., 2020). 

Among utility-scale energy planning and optimization tools, each presents strengths and applications 

depending on the specific needs of a power system. PLEXOS and WASP are particularly suited for 

detailed utility-scale expansion planning, while TIMES and MESSAGE offer broader economy-wide 

energy system modeling. Balmorel provides a valuable open-source option for electricity market 

analysis, whereas LEAP is best suited for scenario-based policy evaluation. The selection of an 

appropriate energy modeling tool depends on the scope, scale, and objectives of the analysis, as well 

as the level of technical expertise required for implementation. 

 

 

 

  



25 

 

2.3 Empirical Review 

There has been great interest by scholars in developing sustainable energy plans for various countries. 

More recent studies focused on energy planning tools aiming towards more RE integration and CO2 

mitigation. 

Icaza et al., (2022) assessed the South American Electric Energy System, particularly focusing on 

integrating REs to transition away from fossil fuel-based systems. Their study emphasizes the 

Ecuadorian case, notable for its initial reliance on fossil fuels, particularly damaging in the Amazon 

region. Using an analysis in EnergyPlan, the research proposes realistic RE sources and their 

distribution to meet the projected 2050 demand, envisioning a 100% RE generation system in 

Ecuador. This transition is expected to have positive effects, including economic growth, increased 

production levels, and enhanced citizen well-being. By 2050, hydro, solar PV, and wind are expected 

to contribute considerably to the energy matrix, with 20 GW of installed electricity and 72.24 TWh 

of yearly production. The average production cost per MWh is anticipated to be approximately 18 

US cents. This study highlights Ecuador's initial reliance on fossil fuels and proposes significant new 

technologies for a 100% renewable energy transition. However, the findings are limited to the 

technological dimension, excluding financial feasibility. 

Simsek et al., (2020) compared energy scenarios for a low-carbon transition in Chile by 2030 using 

LEAP model to forecast the energy demand, supply, and emissions. The study offers insights into 

long-term energy planning, exploring alternatives and evaluating the impact of existing policies, 

NDCs, and SDGs. Results show that scenarios with considerable energy demand reductions lead to 

notable emission cuts, especially from the demand sector, with a need for energy efficiency and RE 

policies in transport, mining, and industries. Increased deployment of wind, solar, and hydropower 

could achieve over 80% RE generation by 2030, reducing emissions and diversifying the energy mix. 

While the paper focuses on demand reduction strategies, leading to significant CO2 emission 

decrease and increase of RE generation in Chile, it does not directly tackle the issue energy access. 

Following the launch of a program called (REFIT) to facilitate the introduction of renewable energy 

(RE) generation into South Africa's power network, the National Energy Regulator of South Africa 

(NERSA) launches a RE program aimed at generating 17.8 GW of RE by 2030 through sources such 

as wind power, solar PV, concentrated solar power (CSP), landfill, and mini hydro. With an 

expected significant increase in generators to be connected to Eskom's transmission and distribution 

networks expected, Bello et al. (2013) examines the implications for South Africa's distribution 

networks, particularly under Eskom's (the national power utility) oversight in rural areas and smaller 

towns and municipal control in cities. It proposes a standardized approach, leveraging the SMART 



26 

 

Grid concept, to optimize the planning, design, and operation of distribution networks across the 

country while implementing initiatives like as distributed generation and enhanced demand 

management. This research is a great addition to this literature review, as it breaks down a program 

designed to increase RE generation and fight energy monoculture. It, however, is mainly a 

recollection and critique of the composition of the REFIT program and doesn't address electricity 

access issues. 

In Kenya, Okomol et al. (2021) provide a detailed roadmap for both national and county governments 

on how to attain 100% energy access and prioritize sustainable energy transition, emphasizing the 

creation of green jobs and enterprises. Utilizing the LEAP energy modeling software, the article 

presents the Medium-Term Plan (MTP) and Long-Term Plan (LTP) updates from 2019 to 2030, 

focusing on expanding the country's power system while ensuring compliance with planning criteria 

and frameworks. The author analyzes 4 different scenarios, focusing on BAU, the need to reduce the 

wood fuel technologies in households, a transition to clean biogas fuels, LPG and solar, and finally, 

an ambitious goal to reach zero biomass use countrywide, with high adaptation of LPG, electricity 

and solar. Recommendations include sustaining broad-based development by ensuring stable and 

sufficient energy supply to meet growing demand, emphasizing the need for the Government of 

Kenya to prioritize sustainable energy initiatives. 

This study serves as a foundational document for investors, particularly in energy generation and 

distribution, with recommendations regarding RE projects to invest in. The paper does not however 

quantify the level of investments or guide on the least cost of production for the recommended 

projects.  

Ugwoke et al. (2021)   underline the necessity for an integrated energy planning framework in low-

income countries with rural populations and limited finances, aiming to utilize RE for economic 

development. They introduce a framework for integrated rural energy planning to enhance energy 

access in rural areas through renewable energy. Using case studies from rural Nigeria, they apply this 

framework to identify suitable sites, estimate energy demand, and assess renewable energy resources 

using GIS-based tools. The results provide insights into localized energy policies combining rural 

electrification with energy conservation measures, with potential relevance in similar contexts across 

the sub-Saharan Africa region and developing Asia. While the approach offers a unified roadmap for 

energy planning and RE integration to enhance rural energy access, it focuses solely on community-

scale projects and does not address the complexities of urban areas with higher energy demands from 

industrial and commercial activities. 



27 

 

The Philippines has gradually demonstrated a strong commitment to RE sources, to combat climate 

change, surpassing many countries in their adoption and distribution of renewable energy in primary 

energy and electricity. Roxas & Santiago (2016) argue in their study on an alternative framework for 

renewable energy planning in the Philippines that the emphasis on aggressively increasing RE 

capacity, primarily through grid-connected applications, has neglected the potential benefits of off-

grid RE solutions, which could replace costly diesel generation and improve community livelihoods.  

This paper suggests a shift in focus from simply increasing megawatts to considering the broader 

impact of renewable energy (RE) on communities. It advocates for a planning approach that evaluates 

both the livelihood benefits and energy delivery effectiveness of different RE technologies in off-

grid and on-grid connected regions. Additionally, it presents a real-life off-grid RE application 

addressing energy and economic poverty on a kW scale. The author stresses the importance of 

tailoring an Energy Reform Program (ERP) to the economic context and goals of the country, 

highlighting RE's potential to alleviate rural poverty. Key strategies include diversifying the energy 

mix for security and adopting a demand-driven approach to expand energy access, particularly on 

off-grid islands. It is important to note that these recommendations serve as suggestions rather than 

outlining a definitive energy plan for the Philippines. 

Using the case study of Cambodia and Thailand, Chhay (2018) performed a Long-Term electricity 

planning, aiming at reducing the GHG emissions caused by an increasing reliance on fossil fuels to 

meet rising energy demands in Cambodia and Thailand. The study employs the LEAP model to 

analyze different scenarios from 2015 to 2050, including Business-as-Usual (BAU), Renewable 

Energy (RE), Carbon Capture Storage (CCS), Carbon Tax (CT), and Electric Vehicle (EV) scenarios. 

Results indicate that scenarios such as the CT500 (referring to a carbon tax increase from $20/tCO2 

in 2020 to $500/tCO2 by 2050) show promise in significantly reducing CO2 emissions compared to 

BAU, with RE scenarios also demonstrating considerable potential. Notably, the costs of electricity 

production from RE sources remain lower than those of CCS and CT scenarios. Chhay, (2018) places 

a greater emphasis on CO2 emissions mitigation and does not relate the study’s objectives to 

improving Burundi’s electricity access rate. 

 

 

 



28 

 

2.4 Summary of Gaps 

The analyzed studies highlight the importance of having energy plans adapted to countries. 

Depending on the researchers’ objectives, some studies focus on transitioning from fossil fuels to 

renewable energy sources, others on enhancing rural energy access, or reducing carbon emissions. 

The identified gaps are: 

i. Some studies have explored various scenarios that increase RE use, and reduce fossil fuels, 

but failed to contextualize the research to the energy access problem. 

ii. While some tailored energy planning tools to improve energy access, they focused on 

community-based areas rather than at the country level. 

iii. Although some studies developed extensive plans for underdeveloped countries, addressing 

energy access and RE exploration, they only considered the technological aspects, without 

quantifying investment needs or identifying the least-cost expansion scenarios. 

This study addresses the identified gaps by developing an energy planning tool aimed at increasing 

energy access in Burundi. It focuses on utilizing renewable energy instead of fossil fuels and provides 

guidance on the least-cost expansion plan. 

  



29 

 

CHAPTER 3 METHODOLOGY 

3.1 Introduction 

The research methodology comprised four distinct steps: first, an analysis of the current energy 

sector, existing power plants, and the power development plan created by the Ministry of Hydraulics, 

Energy, and Mines (MINHEM); second, an energy demand forecast for 2040; third, a description of 

investment scenarios modeled with the Wien Automatic System Planning (WASP-IV); and fourth, 

the presentation and comparison of results, as illustrated in the Flowchart of the methodological 

approach, Figure 3.1.  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.1 Flowchart of the methodological approach 

Conduct an Energy 

Resource Assessment 

• Existing plants data (current E. sector, 

installed capacities, penetration %) 

• Power development plan (2050 target, PND, 

NDC) 

Forecast Energy Demand 

• Data collection (GDP, pop growth, 

consumption, economic growth…) 

WASP- IV 

Model investments 

scenarios. 

Data Analysis 

Results: Energy Mix, Cost of 

Production 

Comparison between scenarios: 

energy demand and supply 



30 

 

3.2 Data Collection 

The data for this study was acquired from Burundi's Ministry of Mines and Energy, as well as from 

the Water and Electricity Production and Distribution Authority (REGIDESO). Forecasting future 

energy generation requires Gross Domestic Product (GDP), growth rate, population size, historical 

data on electricity generation, and a power development strategy. Furthermore, any power generation 

technique must have maximum availability, fixed O&M costs, efficiency, and a long lifetime. The 

remaining data was obtained from publicly available online sources such as World Bank, IRENA, 

IEA reports on Energy Access, including African Development Bank Group, (2020), Burundi Energy 

Access Project Phase1 (2024), among others. 

3.3 Demand forecasting 

Load forecasting was done using the econometric model, based on key economic and demographic 

indicators, as guided in Pegels, (2010). The methodology follows a step-by-step approach, 

incorporating population growth, GDP per capita trends, and elasticity factors to estimate future 

electricity demand. The electricity demand (Dₜ) was modeled as a function of population (Pₜ), GDP 

per capita (Yₜ), and their respective elasticities. The general demand function is expressed using 

equation 1 as 

D𝑡 = D0 ∗ (
P𝑡

P0
)

𝑒𝑝

∗ (
𝑌𝑡

𝑌0
)

𝑒𝑦

                                              1 

 

Where: 

• D𝑡 = Forecasted electricity demand in year t 

• D0 = Base year electricity demand 

• P𝑡 = Population in year t 

• P0 = Base year population 

• Y𝑡 = GDP per capita in year t 

• Y𝑂 = Base year GDP per capita 

• ey = Population elasticity of electricity demand 

• 𝑒𝑝 = Income (GDP per capita) elasticity of electricity demand 

The population elasticity (ep) is calculated as a function of change in population over the 

corresponding change in electricity demand over a period of time, whereas income elasticity (ey) is 



31 

 

a function of the change in GDP Per Capita over the corresponding change in electricity demand over 

a period of time. 

The future population was forecasted using an exponential growth model, based on the historical 

population growth rate (𝑟𝑝 ), as per equation 2 

𝑃𝑡 = 𝑃0 ∗ (1 + 𝑟𝑝)𝑡               2 

Where: 

• 𝑟𝑝  = Historical population growth rate (i.e observed steady 2% from 2010 to 2025) 

• t = Years from the base year 

The GDP per capita was forecasted using a similar exponential growth formula, incorporating the 

GDP growth rate (𝑟𝑦), as indicated in equation 3.  

𝑌𝑡 = 𝑌0 ∗ (1 + 𝑟𝑦)𝑡             3 

Where: 

• 𝑟𝑦 = GDP per capita growth rate (observed steady 2% from 2020 to 2025) 

• t = Years from the base year 

Finally, the peak load was derived using the peak load ratio (PLR₀), which represents the relationship 

between peak demand and total electricity demand. The formula for future PLR is presented in 

equation 4.  

𝑃𝐿𝑅𝑡 = 𝑃𝐿𝑅0 ∗ (
𝐷𝑡

D0
)

𝑒

              4 

  

Where: 

• 𝑃𝐿𝑅0 = Base year peak load ratio 

• 𝐷𝑡 = Forecasted electricity demand 

• 𝐷0 = Base year electricity demand 

• e = Elasticity factor  

This method ensures that peak demand projections align with system planning needs, accounting for 

variability in demand growth. The projections, along with the peak load ratio, were input into the 

WASP software to model the optimal capacity of power plants needed annually. This process 



32 

 

considers peak load capacity and the peak load ratio to determine the required dispatchable energy 

sources for meeting average load demand and assessing the necessary spinning reserves. 

3.4 WASP Model 

WASP-IV is the name given to the updated version of the model, which includes several additional 

features. Its goal is to ascertain, given a user-specified restriction, the economically viable generating 

growth policy for an electric utility system. WASP-IV assesses system characteristics such 

production costs, unseen energy requirements, and dependability using a probabilistic approximation. 

An optimal dispatch policy that satisfies external restrictions on fuel availability, environmental 

pollution, and power generation from particular plants is developed using linear programming. 

Additionally, it compares the costs of several system growth techniques using dynamic optimization. 

The WASP-IV code allows one to identify the best expansion strategy for a power production system 

over a thirty-year period while staying within the planner's limits. The optimal is measured in terms 

of the lowest discounted total costs. Equation 5 represents a cost function/objective function that is 

used to evaluate each potential order of power units integrated into the system that meets the 

requirements. 

𝐵𝑗 = ∑ (𝐼𝑗,𝑡 − 𝑆𝑗,𝑡 + 𝐹𝑗,𝑡 − 𝐿𝑗,𝑡 − 𝑀𝑗,𝑡 − 𝑂𝑗,𝑡)
𝑇

𝑡=1
                             5 

Where, 

𝑩𝒋  is the objective function attached to the expansion plan j 

t  is the time in years (1, 2, ..., T) 

T  is the length of the study period (total number of years) 

I  is the capital investment cost 

S  is the salvage value of investment cost 

F  is the fuel cost 

L  is the fuel inventory cost 

M  is the non-fuel operation and maintenance costs 

O  is the cost of the unserved energy 

It is important to add that the I, S, F, L, M and O values are discounted to a reference date using a 

specific discount rate I. Finally, the least-cost expansion plan among all possible scenarios then 



33 

 

determines the optimal expansion strategy. It is also important to note that the software operates 

under specific constraints. It assumes that the total installed capacity and generation must always 

meet or exceed peak demand. Furthermore, it allows users to define the available technologies and 

specify their expected commissioning years. As a result, expansion decisions are constrained only by 

predefined resource availability and technology limits. Lastly, power plants must operate within set 

fuel and efficiency parameters, meaning the model does not account for unexpected fuel supply 

disruptions. WASP-IV consists of several interdependent modules, each performing a specific 

function within the optimization process. Table 3.1 describes the 7 modules of WASP- IV, while 

Figure 3.2 represents a simplified flow chart of the model WASP-IV computer code. 

Table 3.41 WASP Modules (adapted from WASP Manual) 

Modules Description 

Module 1: LOADSY 

(Load System 

Description) 

 

This module processes information describing period peak loads and load 

duration curves for the power system over the study period. 

Module 2: FIXSYS  

(Fixed System 

Description) 

Information on the current generation system, any planned additions or 

retirements, and any user-imposed restrictions on fuel supply, environmental 

emissions, or the ability of certain plants to generate energy are all processed 

by FIXSYS.  

Module 3: VARSYS  

(Variable System 

Description) 

This last input module processes data describing the various generating plants 

which are to be considered as candidates for expanding the generation plan. 

Generator forced and unforced outrage, LOLP, reserve margin were studied. 

Key inputs include installed capacity, fuel characteristics, outage rates, 

emissions limits, and hydro reservoir constraints.  

Module 4: CONGEN 

(Configuration 

Generator) 

 

This module computes every feasible combination of expansion candidate 

additions from year to year that meet specific input criteria and that, when 

combined with the fixed system, can meet the loads. Additionally, it determines 

the combined list of FIXSYS and VARSYS plants' basic economic loading 

order.  

Module 5: MERSIM  This module considers all configurations put forward by CONGEN and uses 

probabilistic simulation of system operation to calculate the associated 



34 

 

(Merge and Simulate) 

 

production costs, energy not served and system reliability for each 

configuration. In the process, any limitations imposed on some groups of plants 

for their environmental emissions, fuel availability or electricity generation are 

also considered. The dispatching of plants is determined in such a way that plant 

availability, maintenance requirement, spinning reserve FIXSYS and VARSYS 

plants. 

Module 6: DYNPRO  

(Dynamic Programming 

Optimization) 

 

Based on previously calculated running expenses, input data on capital costs, 

energy not served costs, economic characteristics, and reliability standards, it 

determines the best growth strategy. 

Module 7: REPROBAT  

(Report Writer of WASP 

in a Batched 

Environment)  

 

For set expansion schedules and the optimal or nearly optimal power system 

expansion plan, it generates a report that summarizes the entire or partial 

outcomes. The file that can be utilized for a graphical representation of the 

WASP results also contains some of the outcomes of the computations carried 

out by REPROBAT.  

 



35 

 

 

Figure 3.2 User interaction in running the WASP code (adapted from ORNL 73-7759 RI) 

Since 2023 provided the most recent complete hourly load duration curve, it was selected as the base 

year. However, while 2023 serves as the reference point, this study assumes that the modeled 



36 

 

planning begins in 2026 to develop a more practical and achievable expansion strategy. The data 

entered into Fixsys and Varsys was sourced from REGIDESO as part of the planning process for 

Burundi's integration into the Central African Power Pool (CAPP). 

In the model, solar, peat-fired, and municipal waste power plants were classified as thermal plants 

since WASP-IV is designed to accommodate only hydro and thermal plant types. Additionally, the 

software enables generation and demand data to be divided into up to 12 periods within a year. For 

this study, three periods were considered, corresponding to Burundi’s three main seasonal weather 

variations. This approach is particularly relevant because the country's primary energy source is 

hydro, which is significantly affected by seasonal fluctuations in water availability. 

3.5 Scenario Description 

This section of the methodology involves modelling different scenarios, each sourced from the 

Ministry of Hydraulics, Energy and Mines. This includes examining current REGIDESO’s plan 

towards increasing Burundi’s electrification rate, existing and revised Nationally Determined 

Contributions (NDCs), and finally compute the optimal expansion plan. These strategies form the 

basis of the scenarios. In total, three distinct scenarios were established as outlined in Table 3.2.  

Table 3.5 Scenarios Modeled 

Scenario Description 

  

Business As Usual 

(BAU) 

This scenario consists of modelling the results of the forecasted energy when 

all variables and parameters of the Burundian energy system follow 

historical trends.  

Nationally Determined 

Contributions (NDCs)  

This scenario focuses on enhancing the production capacity of hydro and 

solar PV plants. 

Optimal Expansion Plan 

This scenario integrates the existing plants outlined in both the BAU and 

NDC scenarios and involved analyzing DYNPRO outputs before 

conducting new CONGEN iterations. Each new iteration was generated by 

increasing the minimum number of plants marked with (+) by one while 

reducing those marked with (-) by one in CONGEN, ultimately identifying 

the optimal expansion plan. 

 



37 

 

3.6 Ethical Considerations  

This work shall be conducted in accordance with the ethical research provisions of the University. 

This will include protecting the respondents, by observing anonymity and by giving them the option 

to opt out of the research whenever they feel like. Consent shall also be sought from the respondents. 

Full disclosure on the uses of the data will be useful in enhancing the ethicality of the process. The 

respondents will be informed that the data is useful for the research work and will be exposed to 

examiners and classmates during seminars.  

Any requested documents will be submitted to the Strathmore University Institutional Scientific and 

Ethics Review Committee (SU-IERC). The committee holds the mandate of reviewing all proposed 

research projects, evaluating them, and deciding whether the research should be permitted to proceed 

based on the ethical and scientific merits of the project. Upon submission of requested documents, 

feedback on a proposal will take a minimum of 14 working days, and if successful, the research 

permit will be granted Ethical Approval.  

  



38 

 

CHAPTER 4 RESULTS AND DISCUSSION 

4.1 Introduction 

This chapter presents the findings of the assessment on sustainable power generation scenarios in 

Burundi using a multi-criteria approach. It is structured into three main sections. The first section 

provides an overview of the country’s energy resources, highlighting their availability and potential 

for electricity generation. The second section presents the results of the demand forecast over a 15-

year period. Finally, the third section covers the simulation of various investment scenarios using the 

WASP model, evaluating their feasibility in terms of cost, sustainability, and energy security.  

4.2 Energy resource assessment 

Burundi possesses a diverse range of energy resources that have the potential to contribute to the 

country’s electricity generation mix. Table 4.1 provides an assessment of the key energy resources 

available in Burundi, including hydropower, solar energy, biomass, wind energy, and fossil fuels. 

Table 4.2 Energy Resources in Burundi 

 Energy source Potential Location 

Hydro 300 MW  
Ruzizi III & IV, Kagunuzi, Mulembwe, 

Kaburantwa, Kagera River 

Solar 4.5 - 5.5 kWh/m²/day 
Northern and Eastern Burundi (Gitega, Muyinga, 

Kirundo) 

Biomass 
~7.5 million tons of 

wood/year 

Rural areas, coffee & sugarcane plantations (e.g., 

Ngozi, Kayanza, Ruyigi) 

Wind Wind speeds ~3-5 m/s  
Highlands & Lake Tanganyika shores (e.g., 

Bururi, Rutana) 

Municipal waste   Bujumbura 

Peat 47-58M tons   

 

Hydropower offers the greatest potential at approximately 300 MW economically viable. It is 

followed by solar energy which holds significant potential, as indicated by the Global Solar Atlas, 

which shows irradiation levels ranging from 4.5 to 5.5 kWh/m²/day, particularly in the northern and 

eastern regions. In addition, biomass remains a widely used energy source, primarily for cooking. 

The country's biomass potential was estimated at 8.5 tC/ha/yr, significantly higher than the global 

average Net Primary Production of 3-4 tC/ha/yr.  This is primarily from agricultural residues and 



39 

 

wood remains widely used for cooking but has potential for electricity generation through biogas and 

biomass power plants. Additionally, peat and municipal waste, concentrated in Bujumbura, represent 

further potential energy sources.  

Meanwhile, wind energy assessments indicate moderate wind speeds (3-5 m/s) in highland and 

coastal areas, but further feasibility studies are needed to determine its viability for large-scale 

deployment. Finally, fossil fuels in Burundi are fully imported and primarily used for backup power. 

Given the high cost of imports and the country’s commitment to sustainable development, fossil fuels 

are an increasingly less favorable option for future energy expansion. 

4.3 Peak Load Demand Forecasting  

This section presents the load forecasting results of the future peak load demand of Burundi over the 

study period (2025–2040). This was computed based on equations 1, 2, 3 and 4. The results are 

presented in table 4.2.  

Table 4.3 Forecasted Peak Load Demand 

Year 
Population 

Forecast (Million 
inhabitants), Pt 

GDP Per Capita 
Forecast (US$), Yt 

Demand Forecast 
(MW), Dt Peak Load Ratio 

2023 13,689,450 193.01 82.67 0.61 
2024 13,963,239 196.75 90.50 0.63 
2025 14,242,504 200.57 99.08 0.67 
2026 14,527,354 204.46 108.47 0.70 
2027 14,817,901 208.43 118.75 0.73 
2028 15,114,259 212.47 130.00 0.77 
2029 15,416,544 216.59 142.32 0.81 
2030 15,724,875 220.79 155.81 0.85 
2031 16,039,373 225.08 170.57 0.89 
2032 16,360,160 229.44 186.73 0.93 
2033 16,687,363 233.89 204.43 0.98 
2034 17,021,110 238.43 223.80 1.03 
2035 17,361,533 243.06 245.01 1.08 
2036 17,708,763 247.77 268.23 1.13 
2037 18,062,939 252.58 293.65 1.18 
2038 18,424,197 257.48 321.47 1.24 
2039 18,792,681 262.47 351.94 1.30 
2040 19,168,535 267.57 385.29 1.37 

 

The load forecasting results indicate a significant percentage rise in electricity demand in Burundi, 

increasing by over 366% over the study period. This surge in demand is attributed to a confluence of 



40 

 

factors, including a 40% growth in population and a 39% rise in GDP per capita. These figures reflect 

both demographic expansion and economic development, which are typically strong drivers of 

electricity consumption as industrialization, urbanization, and household energy access improve. 

Notably, the projected demand of 108.47 MW in 2026 closely aligns with the African Development 

Bank's (AfDB) forecast of 100 MW for the same year in the Burundi Energy Access Project Phase1 

(2021). 

4.4 Optimal Expansion Plan Simulation on WASP  

4.4.1 Input Modules  

The demand projections over the study period were input into Loadsys. The first three modules were 

run one after the other as illustrated in Figure 3.2. The values in Table 4.3 and Table 4.4 summarize 

the existing, candidate and committed power plants used as inputs in the Fixsys and Varsys modules 

respectively. These represent the total currently installed capacities power plants, including the hydro 

power stations and thermal plants. 

Table 4.4 Existing Power Plants 

Plant Name Technology 
Installed 
Capacity 

Status 
Commissioning 

Year 

CHE Rwegura Hydro 18.00 Existing 1986 

CHE Ruzizi II  Hydro 12.00 Existing 1958 

CHE Ruzizi I  Hydro 3.50 Existing 1989 

CHE Ruzibazi Hydro 15.00 Existing - 

Rusumo Hydro 26.67 Existing - 

KABU Hydro 20.00 Existing - 

CHE Mugere Hydro 8.00 Existing 1982 

CHE Nyemanga Hydro 2.80 Existing 1988 

CHE Ruvyironza Hydro 1.50 Existing 1984 

CHE Gikonge Hydro 1.00 Existing 1982 

CHE Kayenzi Hydro 0.80 Existing 1984 

CHE Marangara Hydro 0.28 Existing 1986 

CHE Buhiga Hydro 0.47 Existing 1984 

CHE Nyamyotsi Hydro 0.30 Existing - 

CTH. en location (Interpetrol) Diesel Plant 30.00 Existing 2017 

CTH. de la REGIDESO Diesel Plant 5.00 Existing 1996 

Mubuga Solar Plant Solar PV 7.50 Existing 2021 



41 

 

TOTAL EXISTING INSTALLED CAPACITY, MW 152.82     

Table 4.4 Candidate and Committed Power Plants 

Plant Name Technology 
Installed 
Capacity 

Status 
Commissioning 

Year 

Kirasa-Karonke Hydro 16 Committed  2027 

Mpanda Hydro 10.4 Committed  2028 

Ruvyi Mule Hydro 10.65 Committed 2026 

Jiji Hydro 32.5 Committed 2025 

Mulembwe Hydro 17 Committed 2025 

Kabu 23 Hydro 15.5 Committed 2027 

Ruzizi III Hydro 49 Candidate 2027 

Siguvyaye Hydro 6.6 Candidate 2028 

Ruvubu Hydro 50 Candidate 2028 

Mini Hydropower Hydro 11 Candidate 2028 

Solar Project Solar PV 17 Committed 2027 

TOTAL CANDIDATE INSTALLED CAPACITY, 
MW 

235.65     

 

  



42 

 

4.4.2 Output Modules Results 

4.4.2.1 Business as Usual Scenario (BAU) 

CONGEN-MERSIM-DYNPRO sequential runs were undertaken without changing any input in the 

other modules. Based on predefined parameters in Loadsys, Fixsys and Varsys, WASP applied 

Equation 5 to execute the simulation. Figure 4.1 presents the number of units required for each year.  

 

Figure 4.1 BAU DYNPRO Results on the Recommended Number of Units for an Optimal 

Expansion Strategy 

The (+) and (-) signs signify that this expansion plan is not optimal and indicate the need for either 

adding or reducing a unit, as shown above. The model advises against incorporating peat-fired or 

municipal waste power plants before 2034, while recommending the expansion of hydropower 

capacity to meet growing demand. 

4.4.2.2 Nationally Determined Contributions 

The NDC submitted scenario focused on an increase of the hydro power plants and integration of the 

solar plants, while reducing them by one, as suggested in the BAU scenario. Figure 4.2 presents the 

simulation results. 



43 

 

 

Figure 4.2  DYNPRO Results on the Recommended Number of Units for an Optimal Expansion 

Strategy 

As shown in Figure 4.2, this combination also indicates that the expansion plan is not optimal. It 

recommends keeping all four hydropower units added in Varsys from 2028 while introducing the 

first solar unit in 2030, followed by at least three additional units in 2035. This is likely due to the 

absence of diesel power plants in the plan, making it difficult to ensure sufficient dispatchable 

capacity given Solar’s intermittency. Furthermore, after 2034, more than two units of both peat-fired 

and municipal waste power plants are required to meet demand. 

  



44 

 

4.4.2.3 Optimal Expansion Plan 

After several iterations of combinations in CONGEN, the optimal expansion plan was determined. 

Figure 4.3 presents the required number of optimal units for each year.  

Figure 2.1

 

Figure 4.3 Optimal Units Required for Each Year 

Figure 4.3 indicates that no additions or reductions are suggested in the plan, confirming that this 

represents the optimal combination. Unlike the BAU and NDC scenarios, the plan proposes a 

significant increase in peat-fired, municipal waste, solar, and import/diesel power plant units starting 

in 2037. 

Figure 4.4 illustrates the corresponding