Definition of the objective of the project
The CRSES Department set a clear objective for the project based on current institutional needs, which was discussed during the initial meeting The Department aimed to support policymakers by providing reliable information for informed decision-making The thesis's purpose was well-defined, with the project's framework evolution recognized as essential for consistent progress, heavily reliant on the availability of suitable data.
Information procurement
The initial phase of the process involved data mining to assess the environmental conditions, highlighting the necessity for the analysis Key sources of information included scientific papers, publications from related departments, and data from scientific databases.
The following scientific databases were considered to obtain some of the required information:
The Library and Information Service of Stellenbosch University
The Community Research and Development Information Service (CORDIS)
The Austrian Library Composite of the ệsterreichischer Bibliothekenverbund GmbH
The library of the FH Technikum Wien/ the University of Applied Science Vienna
The library of the TU Vienna
The library of Science Direct
The project's expenditure primarily focused on desk and literature research, utilizing a combination of hard copy materials and predominantly online resources Additionally, further information was gathered through interviews, phone calls, emails, and participation in lectures and scientific congresses, which facilitated networking opportunities with other professionals.
Quality assurance
The approach that was adopted to ensure the quality of work is stated as follows:
Mutual control existed between the author, his supervisors, and the project team, with the conclusions, calculations, assumptions, and adoptions being validated
The reliability of the data that were obtained was compared with that of data that were obtained from other sources
In order to ensure the reliability of the data source, information was obtained only from government, institutional and scientific sources.
Implementation of present resources
This section deals with the utilisation of available auxiliary means, such as simulation tools and information-providing applications, which contributed to the completion of the analysis
In addition to developing a specific model, various tools were employed for simulation and verification, demonstrating the link between raw weather data and associated power loads The applications utilized in this process include:
The basic PV model by P Gauché (2011), which is a time series simulation for solar PV issues
The System Advisor Model (SAM), by the National Renewable Energy Laboratory (NREL 2005), for CSP and solar PV applications
The following sources were applied to access the necessary records of data for simulation and/or validation purposes:
The Wind Atlas of South Africa (WASA), for the weather mast (WM) measurements
The GeoModel Solar Ltd (SolarGIS)
The South African Department of Water Affairs (DWA 2013)
All data used in this study underwent thorough validation, ensuring their origin, reliability, and error margins were assessed alongside existing evaluations from other researchers Any irregularities were meticulously identified to facilitate an accurate assessment of the likelihood of error recurrence.
Development of the model
The quasistatic nature of the model was obtained by using basic physical correlations and external time series simulated data records from specific project sites
The model's methodology incorporated various approaches tailored to each modeling technology based on the availability of existing simulation programs Any utilized tool underwent a rigorous validation process, which included comparisons with other simulation methods, alignment with project developers' expectations, and plausibility checks by project team members For uniquely developed methods, like the wind power analysis, similar validation procedures were also implemented to ensure accuracy and reliability.
To address simulation concerns, specific boundary conditions and assumptions were established, with each clearly outlined and technically justified to promote transparency and clarity Additionally, all uncertainties and potential sources of error were explicitly identified.
The simulation approach for every technology (whether wind, solar photovoltaic (PV), CSP, or hydropower) is described in detail in each corresponding chapter
4 Introduction to issues relating to electricity supply and demand
The South African economy heavily relies on fossil fuels, particularly coal, as its main energy source Despite the country's significant potential for renewable energy from hydropower, solar, and wind sources, these alternatives remain largely underutilized.
The present and prospective countrywide power supply is subject to the following three main constraints:
The reliability of a power supply is crucial for maintaining a reserve margin that accommodates both planned and unplanned outages, including maintenance periods In South Africa, this reserve margin significantly decreased from 25% in 2002 to just 10% in 2008, primarily due to strong economic growth and a failure to effectively manage demand, which severely restricted available options for addressing power supply challenges.
In 2007, coal accounted for 92% of electricity generation in South Africa, resulting in significant CO2 equivalent emissions By 2012, Eskom ranked as the second highest CO2-emitting power utility globally South Africa's average CO2 emissions stood at 1.015 tCO2-eqt/MWhel, which is nearly 45% higher than the European average of 0.578 tCO2-eqt/MWhel A comparison of the four highest CO2/MWhel-emitting power companies in 2009 further highlights the sustainability challenges faced by the South African energy sector.
Table 1: The premier CO 2 -emitting power utilities worldwide (Gross, 2012, p 5)
The Southern Co share between CO2 emissions and cumulated energy is almost 27% less than is Eskom’s ratio, which implies an inferior efficiency of resource management, causing higher emissions
Power losses during electricity transmission in South Africa are significant, primarily due to the geographical disparity between major coal deposits in Mpumalanga province and power generation facilities The extensive transmission grid linking the northern and southern regions of the country has led to notable energy losses, which were recorded at 9.5% in 2010 (World Bank, 2013) Experts, including Prof Ernst Uken, suggest that these losses could be as high as 15%, highlighting the need for improved efficiency in the transmission system.
To achieve a viable future scenario with a significant reliance on renewable energy sources, it is essential to enhance supply capacity while simultaneously reducing demand through efficiency improvements and other strategies.
Notice should be taken that the current thesis refers to the commitment of a renewable supply only.
Local renewable energy resource analysis
Solar irradiance
The solar irradiance is mostly represented by the global horizontal irradiance (GHI) and/or the direct normal irradiance (DNI)
The Global Horizontal Irradiance (GHI), measured in kWh/m²/a or W/m², represents the total solar irradiation received on a horizontal surface, combining both direct (beam) and diffuse (scattered) components For solar power applications, the Global Tilted Irradiance (GTI) is utilized to estimate energy production from solar photovoltaic (PV) systems or solar water heaters (SWH) that are installed at a fixed angle.
The DNI value, measured in kWh/m²/a or W/m², indicates the direct solar radiation received perpendicularly on a designated surface, captured through tracking instruments This measurement exclusively accounts for direct sunlight, excluding any diffuse irradiation, and is essential for applications in Concentrated Solar Power (CSP) and Concentrated Photovoltaic (CPV) systems.
In South Africa, the global horizontal irradiance (GHI) can reach up to 2,300 kWh/m²/a, while the direct normal irradiance (DNI) can go as high as 2,900 kWh/m²/a, making it one of the highest in the world Comprehensive solar GHI and DNI data for the country are well-documented and accessible through GeoSUN Africa and SolarGIS For a visual representation of the national variations in DNI and GHI, refer to Figure 2 on page 11.
Wind power
South Africa boasts significant wind potential along its southern and northeastern coastlines A collaborative effort involving the South African National Energy Development Institute (SANEDI), the South African Weather Services (SAWS), the University of Cape Town (UCT), and the Risø Danish Research Institute (DTU), with financial backing from a consortium led by the Department of Energy (DoE), has produced a comprehensive Wind Atlas of South Africa (WASA) This atlas provides detailed wind data across the nation, utilizing a wind model that incorporates measured time series of wind speeds, directions, and terrain topography, including elevation, roughness, and obstacles, to depict the country's wind conditions accurately.
The map in Figure 1 illustrates the generalized annual mean wind speeds over a 30-year period at an elevation of 100 meters above ground level, characterized by flat terrain and a roughness class of 3 cm The numbers 1 to 10 on the map denote the locations of the installed wind measurement stations (WMs).
Figure 1: Wind Atlas of South Africa (WASA 2012, p 4)
Figure 2: South African map for GHI, left and DNI, right (GeoSun Africa 2008)
National energy consumption and allocation
Most of the nationwide primary energy consumption-related publications by governmental sources are both inconsistent and no longer up to date
In 2006, South Africa's total primary energy supply was reported at 5,644,436 TJ by the Energy Department, with coal accounting for 65.9%, crude oil at 21.5%, and renewable sources like biomass contributing 7.6% Conversely, Statistics South Africa estimated a higher primary energy supply of 7,742,673 TJ for the same year, excluding imported energy This highlights the critical role of energy as a key driver of the South African economy.
In the country, the primary final energy consumers are the industrial sector, which accounts for approximately 40% of total consumption, followed by the transportation and residential sectors Detailed insights, including a summary of the national coal and petrol allocation, can be found in Annexure I, Part A, Figure 22, on page 72.
Electricity supply and demand
National electricity supply
Electricity generation in South Africa is heavily reliant on coal, which accounted for 93.2% of the total supply in 2006, significantly higher than the global average of 40% In contrast, nuclear power contributed 4.2% and hydropower only 1.3% to the country's energy mix.
The electricity supply is dominated by the governmental electricity utility, Eskom, a limited range of municipal power purchasers, and some IPPs The Electricity Supply Statistics for
2006 that were released by the National Energy Regulator of South Africa (NERSA) exhibit the results that are shown in Table 2 below
Table 2: List of power capacities (Nersa 2006, p 42)
The location of energy generation capacities is closely linked to the availability of natural resources and the need for security of supply In the northern region, there is a significant concentration of coal resources that aligns with the presence of coal power plants Meanwhile, the southern region hosts several open-cycle gas turbines (OCGTs) that serve peak load demands, along with the Koeberg nuclear power plant, which ensures a stable base load supply This strategic placement is crucial for maintaining a reliable energy supply, especially during transmission grid shortages Additionally, some hydropower plants are located in the Eastern Cape province, contributing to the overall energy mix.
Current electricity supply data is inconsistent and varies among sources, with the last comprehensive statistics released by NERSA in 2006 The most reliable information comes from SSA, which publishes monthly reports on generated and distributable electricity In 2012, SSA reported a total distributed energy of 234 TWh Notably, the period from 2007 to 2009 saw a significant decline in energy supply, largely due to the global economic crisis Trends in electricity supply since 2001 are detailed in Annexure I, Part B, Figure.
The load factor of a power supply system reflects the ratio of average load to maximum load over a specified period, currently at 68.9% A higher load factor is essential for optimizing capacity utilization in base-load power plants For Eskom's coal power generation, the average load factor stands at 73.3%, but it experiences significant fluctuations of +15% to -57%, indicating a high fault rate attributed to maintenance issues, defects, and other related factors (Nersa 2006, p 45).
Sector-specific electricity demand
In 2006, South Africa's total electricity demand reached 205 TWh, predominantly driven by the primary and secondary sectors, with industry accounting for 60% of this demand, including manufacturing, mining, and agriculture The distribution losses, detailed in Chapter 4.4, arose from the disparity between the available electricity for distribution at 233 TWh and the actual energy sold for end use at 205 TWh.
In 2006, domestic customers represented about 94% of the electricity market, yet their actual consumption accounted for only 19% The average retail price for electricity for domestic users was 37.5R/kWh, compared to a significantly lower rate of 16.9R/kWh for the mining industry (Nersa 2006, p 60) For a comprehensive breakdown of sector-specific electricity demand, please refer to Annexure I, Part B, Figure 26, p 78.
Present lack of supply
South Africa's power system faces a heightened risk of outages due to a steadily decreasing reserve margin, which has declined significantly over the past decade Since 1987, power demand has surged by 230%, while supply has only risen by 190% This imbalance has led to a dramatic increase in low-frequency incidents, rising from two occurrences in 2002 to 15 in 2006, alongside a notable escalation in transmission system interruptions lasting over a minute (Nersa 2006, p 37).
From 2006 to 2009, Eskom experienced a significant increase in outage duration, leading to a decline in the energy availability factor (EAF) During this period, over 8000MW of capacity was rendered unavailable for approximately 700 hours annually This situation was primarily caused by poor coal quality, which necessitated higher maintenance rates.
The various causes of the incipient energy crisis are characterised in Annexure 1, Part B
The Department of Energy (DoE) initiated the Medium Term Risk Mitigation Project (MTRM) as part of the Integrated Resource Plan (IRP) 2010 to tackle electricity supply constraints in South Africa While the IRP outlines the long-term generation mix, the MTRM specifically aims to identify short-term supply and demand solutions to mitigate the risk of outages from 2011 onwards.
The 2016 report predicts a significant risk of energy shortages until 2015, primarily due to the delayed commissioning of new coal power plants, Medupi and Kusile, which are expected to add 9.5GW of capacity between 2015 and 2018 The supply-demand balance is anticipated to be particularly strained from 2011 to 2012, with a projected shortfall of 15TWh The Medium-Term Risk Mitigation (MTRM) plan involves collaboration among various stakeholders, including the government, business, labor, civil society, and Eskom, and outlines strategies to address the energy gap, detailed in Annexure 1, Part B.
Prospective development
In 2010, a potential electricity supply scenario for South Africa was calculated based on the IRP 2010 assessment procedure, which established key boundary conditions for future actions (DoE 2010) To ensure accuracy, two independent forecasts were created, detailed in Annexure 1, Part B.
The forecast’s trends drift apart until 2034 is caused by many uncertain assumptions that still essentially had to be set at the time of publication The simulation was done between
In the aftermath of the economic crisis between 2008 and 2010, there was a significant decline in demand, as illustrated in Figure 25 on page 78 This unexpected deviation contributed to the notable disparity between actual demand and forecasts observed from 2010 to 2012.
The projected total energy demand for 2015 is estimated to range from 275 TWh to 315 TWh, as illustrated in Annexure 1, Part B, Figure 28 on page 81 Based on Eskom's contribution, the anticipated power demand for that year is approximately 47 GW (Eskom 2013).
Power distribution
The power grid is divided into transmission and distribution systems, with urban areas primarily benefiting from distribution, while rural regions focus on power transmission As depicted in Annexure I, Part C, Figure 29, p 82, the South African high-voltage power grid extends from the northern to the southern parts of the country, showcasing its extensive integration.
In 2006, electricity generation reached 233 TWh, while the end consumption was 205 TWh, resulting in a system loss of 10.9% when considering both imports and exports The energy flow from generation to consumption is illustrated in the Sankey diagram below, showcasing the figures in TWh.
Figure 3: Transmission and distribution network; units in TWh (Nersa 2006, p 54)
Almost 97% of generation and 100% of electricity transmission were achieved by Eskom, with the distribution being partly managed by the municipalities and the private distributors
The South African electricity grid is challenged by the need to maintain a reliable power supply amid rising consumption in the southern regions As demand increases while supply remains centrally managed, longer distribution distances and power supply imbalances may lead to potential power failures For more insights on future transmission expectations through distribution lines and the power grid, please refer to Annexure I, Part C.
Chapter summary
This introduction has provided crucial background information on electricity supply and demand, emphasizing the need for greater adoption of renewable energy sources The chapter summarizes the key impacts discussed earlier, underscoring the importance of transitioning to sustainable energy solutions.
Although the potential for sustainable energy generation countrywide is vast, it is almost unexploited
South Africa's electricity generation relies heavily on fossil fuels, particularly coal Despite national depletion of this resource, it remains competitively available in the international market; however, its costs have significantly risen over the past decade.
The specific CO2 emissions for electricity generation in South Africa are of the highest worldwide, which implies a lack of efficiency and the likelihood of much environmental pollution
Electricity distribution in the country is managed centrally, leading to power being transmitted over long distances from the north to the south This system results in significant transmission and distribution losses, as well as increased vulnerability to failures.
The demand for electricity has surged at a rate that outpaces generation capacity, necessitating the development of new energy sources to meet this growing need and enhance reserve margins Eskom, the sole national electricity utility, cannot fulfill these requirements independently.
In recent years, South Africa has experienced a significant rise in outages, and scientists predict ongoing capacity shortages for the next decade To address this pressing issue, it is crucial to implement preventive measures aimed at enhancing energy capacity.
In response to the identified challenges, policy frameworks were established to address the existing issues effectively The subsequent chapter will explore the most impactful frameworks concerning renewable energy generation.
5 Policy guidelines and legal framework
In the past decade, the South African government has established legal frameworks to facilitate the large-scale implementation of grid-connected renewable energy sources, addressing the growing electricity demand that exceeds generation capacity Recognizing the importance of private sector involvement in ensuring energy security, the government plans to procure renewable energy from private entities to alleviate current energy constraints This strategy includes long-term guidelines like the IRP 2010-2030 and short-term policies such as the REIPPPP, with key stakeholders including the Department of Energy, the National Energy Regulator (NERSA), Eskom, and independent project developers.
The policy guidelines mandate a specific amount of energy generation from renewable resources This article examines a forecast that illustrates the outcomes of decisions made in this context, highlighting the resulting energy production.
IRP for electricity – IRP 2010
IRP 2010 – content
The IRP 2010 is a dynamic plan that must be updated every two years by the Department of Energy (DoE) to adapt to evolving conditions Public input was integrated into a multi-criteria decision-making process through government working groups, ensuring diverse stakeholder representation The initial phase led to the Revised Balanced Scenario (RBS), highlighting a short-term capacity backlog until 2013 A subsequent public participation round underscored the importance of lowering carbon emissions by enhancing renewable energy use and implementing efficiency measures.
The final adjustment of the IRP 2010 involved a comprehensive re-evaluation of renewable energy sources, incorporating learning rates and breaking down previous renewable categories into specific technologies like wind, concentrated solar power (CSP), and solar photovoltaic (PV) This detailed approach facilitated the development of targeted subsidy mechanisms and led to a notable increase in the number of renewable sources considered, driven by the implications of enhanced competitiveness through learning rates Conversely, nuclear energy costs were raised by 40%, creating a significant disadvantage in comparison to renewable options.
The newly established capacities were advised to be firmly committed for a specified duration—until 2015 for wind and PV, and until 2016 for CSP—to address supply security concerns, highlighting the necessity for the Renewable Energy Bid (REBID) Programme The REBID emerged as a successor to the Renewable Energy Feed-In Tariff (REFIT), as detailed in section 5.2.
Firm commitments have been established for the installation of coal fluidised-bed combustion, nuclear power, and OCGT/CCGT plants The IRP 2010 also tentatively projects final commitments for future IRP iterations aimed at the 2030 unit.
By the end of 2030, the formal results of the policy-adjusted Integrated Resource Plan (IRP) indicate a total energy consumption of 454 TWh, aligning with the modified scenario from the system operator (SO) as illustrated in Figure 28 on page 81 The anticipated distribution of the annual electricity generation for 2030 is projected to reflect these findings.
9% of renewable energies (excluding large-scale hydropower)
5% large-scale hydropower and 1% CCGT
The capacity contribution is allocated differently, taking into account the current fleet and committed power plants, while the IRP 2010 outlines several new build capacity options.
Table 4: Policy-adjusted IRP – intended capacities (IRP 2011, p 7)
New capacity [GW] Committed capacity [GW]
Others (CCGT/OCGT, imported hydropower)
The final Integrated Resource Plan (IRP) recommends replacing nuclear energy generation with renewable capacities if nuclear targets are not achievable, potentially leading to the disconnection of up to 9.6 GW of prospective capacities.
In addition, the IRP 2010 estimates that the committed supply capacities until 2020 will be as follows:
A ‘return to service capacity’ for Eskom: ~1 500MW coal-fired
The DoE’s OCGT programme: 1 020MW
The new coal plants Medupi and Kusile: ~8 700MW
Cogeneration and own build, announced in terms of Eskom’s medium-term power purchase programme (MTPPP): ~390MW
Assumed renewable generation, facilitated by REFIT: 1 025MW
Pump storage: ~1 300MW and Eskom’s Sere wind farm: 100MW
The Integrated Resource Plan (IRP) projects a significant reduction in CO2 emissions from 912g/kWh to 600g/kWh, representing a 34% decrease Within this framework, nuclear energy is classified as emission-free, while the contribution of renewable energy sources, including hydropower, is anticipated to reach 14%.
The Medium-Term Risk Mitigation Plan
The Medium-Term Risk Mitigation (MTRM) Plan for Electricity, published in 2011 as part of the IRP 2010, aims to prevent urgent power outages until 2016 by evaluating risk mitigation strategies Developed by the government, business partners, the National Economic Development and Labour Council, and Eskom, the MTRM highlights the necessity for urgent measures to enhance non-Eskom generation and energy-efficiency initiatives, such as Demand Side Management (DSM), to avoid anticipated rolling blackouts.
The key risks that might lead to a power shortage are summarised below (MTRM 2010, p 2):
A missed EAF of at least 85% by Eskom’s plant fleet
Delays in the new coal power plants Medupi and Kusile
The lack of appropriate procedures related to enabling policy, regulatory instruments, bureaucratic red tape, and other issues
The mitigation plan establishes a legal framework for Independent Power Producers (IPPs), proposing the creation of a neutral body, the Independent System and Market Operator (ISMO), to address the conflicting interests between Eskom and the IPPs Currently, Eskom remains the sole electricity buyer and the exclusive contracting party in the market.
The MTRM evaluation indicates a projected shortfall of 42 GWh from 2011 to 2016 To address these short-term constraints, an additional risk mitigation scheme has been implemented, enabling an increase of 3,500 MW However, a supply gap is still anticipated to persist during 2012 and 2013.
The risk mitigation scheme includes such enterprises as:
Increasing Eskom’s existing generator fleet performance
The Energy Conservation Scheme (ECS) – see subsection 4.3.3
To prevent load-shedding, a mandatory Energy Consumption Standard (ECS) will be implemented, capping the monthly energy usage for consumers Exceeding this limit will result in a penalty rate, ensuring responsible energy consumption.
The MTRM proposed that Independent Power Producers (IPPs) with a renewable capacity of 1,025 MW should begin operations starting in 2012 Consequently, the Multi-Year Price Determination (MYPD Application No 2) could allocate funds for tendering this capacity at REFIT tariffs, as outlined in the MTRM 2010 report.
The Renewable Energy Feed-In Tariffs (REFIT) programme
Based on the Electricity Regulation Act 4 of 2006, which is hereinafter referred to as the
The National Energy Regulator (NERSA) is mandated to set electricity tariffs in accordance with the Electricity Regulation Act (DoE 2011) and the White Paper on Renewable Energies (2003a), as outlined in section 15 of the Electricity Regulation Act (ERA, 2006).
NERSA established the Renewable Energy Feed-In Tariff (REFIT) as a market mechanism to promote the growth of renewable energy generation, aligning with the goal outlined in the 2003 White Paper on Renewable Energies to supply 10,000 GWh.
Since 2013, certain tariffs have been established to ensure electricity prices that cover generation costs, aiming to attract developers to invest in the scheme These tariffs, along with qualifying technologies, have been adjusted annually in line with the development of the Integrated Resource Plan (IRP) 2010, with the goal of achieving an initial capacity of 1,025 MW.
In 2008, a significant rise in appropriation occurred due to the reduced demand from Independent Power Producers (IPPs) for feed-in tariffs, leading to an extension of the contract period from 15 years.
20 years Table 5 below gives insight into the tariff structure (in R/kWh) decided upon
The initial REFIT draft of 2008 did not include Concentrated Solar Power (CSP) and excluded solar photovoltaic (PV) systems; however, the wind power tariff was increased twofold in 2009, while the CSP tariff saw a threefold increase Additionally, the REFIT revisions in 2009 and 2011 expanded funding to encompass solid biomass and biogas projects.
Table 5: Trend in REFIT tariffs (Nersa 2008, 2011)
Until 2011, despite significant investor interest indicated by tariff enhancements, no power purchase agreements (PPAs) were established, leading some to label this period a 'false start' (Kernan A 2013) The delays were attributed to excessive bureaucracy and red tape (Fritz W 2012) Consequently, in 2011, the allocation-based bidding process was implemented as the existing Act did not support a Renewable Energy Feed-In Tariff (REFIT) A media release from the Department of Energy (DoE) on August 31, 2011, stated, “The current legal framework governing the electricity sector in South Africa does not allow REFIT in the guise that had been anticipated; hence a revised procurement process in line with the existing regime had to be developed” (DoE 2011a).
The REBID is an integral component of the REIPPPP, with procurement documents issued by the Department of Energy (DoE) on August 3, 2011, followed by a bidder’s conference in September 2011 The DoE clarified that this approach extends rather than replaces the REFIT process, which is designed to procure small Independent Power Producers (IPPs) and empower local communities to initiate their own power generation initiatives.
The government acknowledged that the initial target of 10,000 GWh could not be achieved by 2013 and extended the deadline to 2015, subsequently increasing the target to a total capacity of 3,725 MW To manage demand during the bidding process, the allocation was capped, reflecting the government's belief that it had exceeded the original energy goal This determination established a capacity of 3,625 MW for large-scale renewable projects and 100 MW for small-scale projects, which range from 1 to 5 MW (DoE 2011g) This total capacity was utilized to solicit tenders in specific bidding rounds.
Table 6 below shows all large-scale technology involved, the allocated total capacities, the maximum permitted capacity for each project, and the past bidding window allocations in Megawatts
Table 6: Overview of the REIPPPP (DoE 2011b, p 2)
Allocation still available Onshore wind
Wind and solar photovoltaic (PV) power have emerged as the leading technologies in the bidding process, accounting for approximately 90% of the available capacity due to their significant reclaimable potential The procurement also includes an allocation of 200MW for concentrated solar power (CSP), representing the latest advancements in solar technology In contrast, other renewable sources such as biomass, biogas, landfill gas, and small hydropower contribute only minimal capacities to the overall energy landscape.
The bidding process aims to set a maximum price for each technology (R/kWh), ensuring that Independent Power Producers (IPPs) do not exceed this ceiling in their bids The procedure begins with the submission of all required documents by the designated deadline, followed by an internal evaluation that identifies preferred bidders This culminates in a financial close, leading to the signing of the Power Purchase Agreements (PPAs) involved.
Table 7 illustrates the bidding strategy for rounds R1 to R3, with R1 and R2 already closed The time frame from the announcement of the preferred bidder to financial closure reduced from 12 months in R1 to 9 months in R3, as the relevant legal departments became more familiar with the process.
Table 7: Bidding approach of the REIPPPP (DoE 2013a)
Bid submission date 4 Nov 2011 5 Mar 2012 19 Aug 2013
Financial close – signature of PPA
The thesis accrual period commenced on May 9, 2013, coinciding with the financial closure of bidding for R2, and concluded with the submission date for R3, as highlighted in Table 7 This date is pivotal for evaluating the renewable energy forecast discussed in Chapter 0 of the thesis Notably, the third bidding round is excluded from this assessment due to the uncertainty surrounding the residual power of 1,176 MW, which would necessitate excessive assumptions.
The Department of Energy (DoE) projected up to five bidding windows to allocate the specified capacity Despite a decline from bidding round R1 to R2, bidding round R3, with a capacity of 1,167 MW, is on track to nearly meet the total target of 3,625 MW The DoE made this determination in collaboration with the National Energy Regulator of South Africa (NERSA) by December.
From 2017 to 2020, the Department of Energy (DoE) aimed to enhance generation capacity, as outlined in the 2012 strategy Under the Energy Regulation Act (ERA) of 2006, an additional 3,200 MW of renewable capacity was earmarked to support energy security and meet the Integrated Resource Plan (IRP) 2010 targets, with 100 MW specifically allocated for small projects The distribution of technologies largely remains consistent with previous allocations, except for Concentrated Solar Power (CSP), which sees its share increase from 5.5% to 12.5%, totaling 400 MW.
The first bidding round achieved the highest capacity and subsidy tariffs due to reduced competition, resulting in fewer project submissions than anticipated The urgency for new capacities linked to MTRM prospects drove the average sales price near the ceiling price, highlighting the initial market launch's distortions This was further evidenced by a significant reduction in subsidy tariffs, with wind decreasing by 21.5% and solar PV by 40% from the first to the second bidding round, indicating an overestimation of subsidy rates (Siepelmeyer T 2013) Table 8 provides insights into the tariff caps and the fully indexed actual subsidy tariffs in R/kWh.
Table 8: Tariff cap and recent subsidy tariffs (DoE 2012a; Greyling A 2012, p 14)
Approved facilities prior to 2011
Existing wind resources
As of now, three on-grid wind farms have been established, but there are no solar PV, hydropower, biomass, or biogas power plants in operation Eskom has built two of these wind facilities and manages the transmission grid, which limits the number of Power Purchase Agreement (PPA) participants The Darling Wind Farm stands out as the first on-grid wind energy project developed and operated by an Independent Power Producer (IPP), despite the absence of timely policy guidelines Additionally, the Klipheuwel Wind Energy Facility was built in 2003 under the guise of a research initiative.
The following table gives an overview of the three already committed wind farms
Table 9: Facilities committed prior to 2011
Developer Capacity per turbine [MW]
Sere Wind Farm Eskom 2 100 Oct 2013
REIPPPP-approved projects
The REIPPPP bidding rounds R1 and R2 resulted in the approval of 47 Independent Power Producer (IPP) projects, with commissioning dates set for the end of 2013 for R1 and the end of 2014 for R2 These timelines align with the thesis objectives to simulate the annual performance for 2015 using existing data from the approved developers in both rounds The subsequent tables offer a comprehensive overview of the wind farms, solar photovoltaic (PV) systems, concentrated solar power (CSP) plants, and hydropower projects associated with R1 and R2, with some projects renamed for clarity All projects were submitted by international IPPs.
Table 10: Wind facilities approved for R1 and R2
Capacity per unit [MW] Rated capacity [MW]
MetroWind Van Stadens Wind Farm 3 26.2
Red Cap Kouga Wind Farm 2.5 77.6
Table 11: Solar PV facilities approved for R1 and R2
Solar PV facilities – project designation Rated capacity [MW]
Mulilo Renewable Energy Solar PV De Aar 9.7
Mulilo Renewable Energy Solar PV Prieska 19.9
SA Mainstream Renewable Power Droogfontein 48.3
Solar Capital De Aar (Pty) Ltd 75.0
Table 12: CSP and hydropower facilities approved for R1 and R2
CSP & hydropower – project designation Technology Rated capacity
R1 Central receiver – heat storage 50.0 CSP – KaXu Solar One Parabolic trough – heat storage 100.0
Hydro Run-of-river power plant 4.3
Hydropower Run-of-river power plant 10.0
Table 6 on page 24 summarizes the cumulative capacities of all listed tables, with the registered capacities for R1 and R2 sourced from the Department of Energy's 'REIPP Announcement' and the 'Window Two Preferred Bidders' announcement (DoE 2012b, DoE 2011f).
Developers often publish nominal capacities that exceed the actual registered capacities for renewable energy facilities The registered capacity for wind, concentrated solar power (CSP), and hydropower typically reflects the total power output at the grid connection point under full load, after accounting for losses In contrast, the peak power for solar photovoltaic (PV) systems is measured under standard test conditions (STC).
Figure 6 illustrates the spatial distribution of all approved Independent Power Producer (IPP) projects, clearly showing that their locations align with the resource potentials identified in the local resource analysis detailed in section 4.1.
Figure 6: Location of IPP projects countrywide, featured by way of Google Maps
7 Modelling the prospective load contribution
This chapter introduces a model that encompasses four distinct approaches utilizing the mentioned technologies It outlines the model's objectives, data assessment, methodology, and various related aspects, providing a detailed description of each specific procedure involved.
Introduction
The government's future energy supply goals are closely tied to specific technology classifications and their individual capacities, particularly with volatile renewable sources like wind and solar This volatility introduces significant uncertainty regarding peak load supply and the unpredictable annual energy output of each technology To address this, the developed model offers a method for analyzing load curves, allowing for the observation of load behavior across various approved renewable technologies and the assessment of minimal to maximal fluctuations throughout different seasons The findings yield insights into nationwide weather patterns and the electricity generated from renewable sources following the implementation of initiatives like the Renewable Energy Independent Power Producer Procurement Programme (REIPPPP).
The model aims to establish a method with default input parameters that produces reproducible and reliable results It incorporates physical laws and mathematical correlations to achieve this goal To assess any deviations, reference values are collected, which include predicted values from independent developers and calculated values derived from other understandable methods The model utilizes annual input data for its analysis.
2010, and approved renewable capacities from bidding R1 and R2 to process an exemplary load course for the year 2015
For time series simulation purposes in every specific approach, various boundary conditions/ assumptions have had to be met The assumptions will be scientifically justified and clearly specified
The simulation requires such necessary tools as:
Microsoft Excel by Windows Microsoft
The solar PV model developed by the Solar Thermal Energy Research Institute
The SAM developed by the NREL – US DoE
The Wind Atlas Analysis and Application Programme (WAsP), developed by the Technical University of Denmark
Input parameters
GeoModel Solar Ltd provided the default input parameters for the simulation, utilizing its solar GIS database The simulation produced hourly averaged time series records, totaling 8,760 entries, which represent the annual period from January 1 to December 31, 2010.
2010) The solar GIS data specification document (Solar GIS 2013) provides detailed information about the data acquisition, as well as about the related method and its occurrence
The four data records used for research purposes are discussed below
Solar radiation parameters, including Global Horizontal Irradiance (GHI) and Direct Normal Irradiance (DNI), are generated using advanced, scientifically validated models that integrate satellite data and atmospheric model outputs These models consider various input parameters, such as cloud index, water vapor data, atmospheric optical depth, elevation, and horizontal profiles The spatial resolution for GHI and DNI is set at a 3 arc-second raster, equivalent to approximately 90 meters at the Equator, with resolution decreasing towards the poles.
According to solar GIS, the quality assessment in South Africa shows a low bias within a range of ± 2.5% and an hourly root mean square error (RMSE) between 16 and 22%
The spatial resolution of simulated air temperature is 1km, at an elevation of 2 meters above surface
According to Solar GIS, wind speed data is intended to be used as ancillary parameters, with wind speeds, directions, and humidity levels derived from numerical weather model outputs However, the spatial resolution of this data is lower than that of solar resource data, which may not accurately reflect site-specific conditions It is essential to verify wind velocity data before utilizing it for further calculations to account for potential deviations The spatial resolution for wind speeds is 900 km² (30 × 30 km) at a height of 10 meters above the surface.
Wind simulation
Data verification
To enhance the accuracy of GeoModel wind speed measurements at a height of 10 meters, the data is cross-referenced with publicly available wind measurement records This process is part of the Wind Atlas of South Africa (WASA), which, as detailed in subsection 4.1.2, involves the strategic placement of ten wind measurement stations across the country to support model development and validation.
The WM's records are specific to individual locations, while the GeoModel simulation offers an average spatial resolution of 900km², potentially leading to more damped data and higher maximum wind speed amplitudes However, comparing both data sets is essential for verification and future adjustments.
The WM sites were not placed in areas that were expected to be windy The placement was done in line with the following criteria (Otto A 2013):
Spaced out evenly across the project area (respecting the numerical wind atlas)
At a distance from complex terrain
In areas uniform in terms of roughness and topography
Within such different climatological regions as coastal and inland low-/high-lying
The WMs provide wind speed values at various elevations (10m, 20m, 40m, 60m, and 62m), along with accurate wind directions, temperatures, barometric pressures, and relative humidity Data recording by WASA began in August 2010, limiting the comparison period to just five months, which means seasonal deviations are not considered To assess wind potential in W/m², the cubed average hourly values from all WASA records are calculated, allowing for effective comparison between data sets.
An evaluation of the standard deviation (SD) at five selected sites confirmed the initial expectations, revealing that the WM's average SD is 24% higher than that of the GeoModel, with values ranging from 2.1 to 2.8 m/s.
Various methods were employed to compare the data records, utilizing formulas such as mean error (ME), root mean squared error (RMSE), and mean absolute error (MAE), among others detailed in Annexure III.
The table below presents a comparison of average wind speed values alongside the relative influence of wind power potential, calculated as (Σvi 3)/1000 Notably, measurements from WM 09 and 10 were excluded due to their implausible deviations.
Table 13: Comparison of GeoModel data at WM sites and WM measurements
The wind measurements from WM consistently exceeded those of the GeoModel, with discrepancies in mean wind speeds ranging from 116% to 169% Additionally, the deviation in summarized single-cubed values, which indicate theoretical wind energy potential, varied significantly between 148% and 475% This analysis focused solely on accurate geographical data.
The following five WMs were the nearest to the related 17 wind farms
Table 14: Error evaluation between GeoModel data and WM measurements
Site ME [m/s] RMSE [m/s] MAE [m/s] MAPE [%]
The error evaluation revealed a substantial deviation, with the mean absolute percentage error (MAPE) displaying a notable gap A further breakdown of the monthly mean absolute percentage error (MMAPE) showed slight variations between months, although limited comparable data spanning only five months restricted a more in-depth analysis.
As illustrated in Figure 7, the averaged distribution curve reveals notable trends in wind speed patterns Notably, the lower GeoModel simulation demonstrated a higher distribution in the 0.5-6m/s range, while the higher wind speeds above 7m/s significantly impacted the apparent gap, ultimately influencing the overall distribution curve.
Figure 7: Mean wind distribution curve – wind mast measurement and GeoModel simulation
The data records exhibited a consistent trend, despite variations in absolute values Relative values, particularly trends, can be utilized for further calculations, although differences in standard deviation highlight the mean relative error.
It is advisable to avoid using GeoModel data records without a corresponding projection linked to the aforementioned results Any approximation should be clear, based on a single value, and evenly distributed to align with the trends observed in the records A potential method for achieving this could involve multiplying each value by a factor of 1.4, which represents the mean average deviation.
Height-related extrapolation
To ensure accurate operational elevation for wind turbines, GeoModel data collected at a height of 10 meters above ground needs to be extrapolated Commercial wind turbines are designed to operate at optimal heights where wind speeds are more consistent and generally increase with altitude The potential energy of the wind is significantly influenced by its velocity, as it varies with the third power of the wind speed.
Key factors influencing wind turbine performance include wind velocity, which correlates with height, and the diameter of the wind stream The hub height of a wind turbine is typically determined by surface roughness and vertical wind distribution For the sake of simplified simulations, a standard hub height of 100 meters is commonly utilized.
The Hellmann exponential law and the logarithmic wind profile law are empirical relationships that describe the correlation between wind velocity and hub height, both of which have been simplified and empirically validated However, the more accurate Monin-Obukhov relation could not be utilized due to insufficient input parameters, including temperature differences and friction lengths (Banuelos, Camacho 2011).
The following formulas represent the Hellmann exponential law (2) and the logarithmic wind profile law (3)
In this context, v denotes the velocity at height H, while v0 represents the velocity at height H0 The Hellmann exponent, also known as the friction coefficient, is indicated by α, and z0 signifies the roughness length in meters The values for both the Hellmann exponent and roughness length correspond to specific terrain classes, as illustrated in the accompanying table.
Table 15: Friction coefficient and roughness lengths (Tong 2010; Patel 2006)
Landscape type Friction coefficient α Roughness length [m]
Both laws were validated using randomly selected data samples from the WASA WMs at heights of 10m and 60m The coefficients α and z0 were calculated using transformed equations for each dataset and averaged through mean and median values These homogenized coefficients were then applied to extrapolate new wind speed samples at 60m, allowing for a comparison with the corresponding true wind speeds.
Graphs were created using the mean and median coefficients α and z0 at a sample wind speed of 8 m/s The comparison revealed consistent results, provided that the variation in the individual calculated coefficients remained minimal.
As a result, the Hellman exponential law was chosen for extrapolation purposes
Figure 8: Comparison of the data according to the Hellmann exponential law and the logarithmic wind profile law
A homogenised α coefficient was calculated using all WM data to extrapolate wind records consistently across the project site, employing a unified method as previously described.
The calculated α coefficient of 0.126 indicates that the sites varied from hard ground to grassland/open areas A lower exponential coefficient signifies a smaller relative increase in the extrapolated value This method assumes that wind farms are typically situated on flat terrain, similar to the locations of wind measurement stations, to minimize the need for excessively tall wind turbine masts.
The projection was made up to a generalised hub height of 100m, although the coefficient was calculated between 10 and 60m The increase was expected to match an approximately realistic result.
Power conversion
The power conversion from an extrapolated wind speed at 100 meters was estimated using the specific power curves of selected wind turbines Two turbines from different performance categories were analyzed to represent the range of turbines utilized across all approved wind farms.
The load behavior of various turbines within the same performance category exhibits notable similarities, likely due to fundamental physical constraints such as Betz's law and component efficiencies.
The analysis utilized a logarithmic conversion for wind speeds between 6 and 10 m/s, while applying linear relationships for speeds outside this range This study focused on two Enercon Ltd wind turbines with capacities of 2.35 MW and 3 MW The minimum operational wind speed was established at 2 m/s, and power output was capped above 13 m/s due to wing stall.
The graph in Figure 9 below illustrates the mathematical approximation to the real power curves involved
Figure 9: Approximation of the wind power curve
At wind speeds up to 9 m/s, power curves for different turbine sizes are nearly identical In a wind farm with a 50 MW capacity located in an area with an average wind speed below 9 m/s, smaller turbines can yield a higher technically exploitable energy output due to the increased number of units installed at a constant hub height However, the choice of turbine size is primarily influenced by economic viability, particularly the specific costs associated with each turbine Therefore, selecting turbine size for individual projects is guided by the estimated load curve.
Method 1 – assorted approach
The approach adopted in the simulation can be summarised as follows:
GeoModel 10m data projection to compensate for WM 10m deviation
The simulation approach evaluated five wind farms, where developers provided annual expected energy outputs to validate the methodology The results, measured in GWh, indicate a significant overvaluation for four of the five projects using the developed methodology.
An extrapolation without a GM 10m data projection surprisingly meets the expected output Table 16: Wind speed extrapolation – verification Units in GWh
Wind farms Expected energy output
Red Cap Kouga Wind Farm 290 426 283
The initial discovery and subsequent validation of the capacity factors (CF) for all projects revealed that the developed method lacked the reliability needed for straightforward application The CF values varied significantly from the expected realistic range.
A further step was contained an assorted approach for different groups of projects regarding to available IPP information The proceeding is described below:
The calculation without a ‘GM to WM – 10m projection’ led to an approximate result compared to the IPPs’ expectations, or fell within a plausible CF bandwidth (in the
An IPP expectation measured in GWh or an estimated Capacity Factor (CF) was provided for five projects To meet the anticipated outcomes, an iterative adjustment of the 'GM to WM – 10m projection' factor was implemented.
No specific project information was available (in the case of six projects): o The resulting CF fell below the expectancy range after calculation without a
‘GM to WM – 10m projection’ o Relative adjustment into a plausible CF range resulted in a subsequent energy output o Iterative adjustment of the ‘GM to WM – 10m projection’ factor achieved the expected output
The plausible CF range was established based on the results obtained and validated through literature review For this study, the CF range is defined as being between 27% and 42%, as referenced by Siepelmeyer (2013), Soni (2012, p 34), and Stanley.
2009) Although the CF bandwidth contained the highest assumption uncertainty, the fluctuation margin in the high wind records allowed for no other possibility than to implement such an approach.
Method 2 – single approach
In order to accomplish a single and comprehensible approach for all cases, and to simulate consistently, the following approach was derived in terms of the methodology discussed above:
1 GeoModel 10m records were used directly for height extrapolation from 10 to 100m
2 The CF’s of the IPP projects were adjusted into the defined bandwidth
3 The ‘GM to WM – 10m projection’ factor was iteratively adjusted to achieve the cumulated energy output (which was derived by the adjustment of the CF, in accordance with the above-mentioned step)
4 A power conversion and cumulative display was performed
Figure 10 illustrates the CF adjustments within a 27-42% bandwidth, with the blue spots representing the original CFs obtained through simulation, and the red spots indicating the adjusted CFs following assimilation.
Figure 10: Capacity factors of wind facilities – model and assimilation
Validation – assorted and single approach
The grouped approach yielded an output of 3,598 GWh, while the single approach produced 3,685 GWh The mean deviation was 13 MW, representing 1% of the installed capacity, with a maximum deviation reaching 60 MW.
Adopting an assorted approach can yield more accurate results tailored to specific project information, unlike the single approach, which overlooks such details However, the assorted approach lacks standardization, hindering replication and relying on unverified references Given that the hourly trends and annual yields of both methodologies were similar, the single approach (Method 2) was selected for further study.
It must be borne in mind that the assorted approach serves as a verification model for the single approach.
Results
To effectively evaluate existing uncertainties before analyzing any results, it is crucial to clearly define the primary boundary conditions that influence the simulation The key assumptions established for this process are summarized below.
A CF bandwidth of between 27 and 42%
The usage of 10m GeoModel records, even though the deviation to measured values records was high
A general turbine hub height of 100m
The generalisation of the exponential coefficient α = 0.126 (determination taken from a WM 10m and 60m in height)
Load curves approximated by Enercon Ltd
The following influences could not be fully taken into account, based on the lack of information, and, to some degree, the limited extent of the thesis:
The seasonal evaluation of WM and GeoModel data
The wind farm’s specific simultaneities
Probable deviation from a standard wind year
The lack of available data in a range that was less than hourly based
The simulation results indicate a total annual energy yield of 3,685 GWh, with a maximum available power of 1,302 MW, closely aligning with the total installed capacity However, the minimum available power recorded was just 8.3 MW, highlighting significant temporal fluctuations in the wind resource While wind farms can operate at peak capacity during certain hours, the firm capacity was notably low at approximately 0.6%, raising concerns about the reliability of wind power as a base load energy source The annual mean power output was 421 MW, but considerable seasonal variations were observed, as detailed in Table 17.
Examining the contribution of wind power to meet winter peak demand is essential, as political frameworks like the IRP 2010 – MTRM necessitate significant input from each generation unit for supply stability Wind energy typically peaks between 19:00 and 22:00, accounting for about 15% of its distribution, which aligns closely with the expected time distribution of 16.7% The firm capacity during this period was 23.5 MW, with an annual firm capacity of 26 MW Further analysis reveals that 40.4% of wind distribution occurs between 08:00 and 15:00, exceeding the expected time distribution Thus, in 2010, wind power contributed modestly to morning peak hours and nearly matched its time distribution-related energy output.
The article includes an annual load duration curve and a seasonally separated duration curve, detailed in Annexure IV Notably, autumn is highlighted as a period characterized by abundant wind resources.
Table 17: Seasonal characteristics of wind generation
The cumulative wind power shows concurrent increases and decreases within a few hours
As a result, the spatial distribution does not imply an upper simultaneity, as is illustrated by the exemplary wind power course that is pictured in Figure 11 below
Figure 11: Exemplary wind power course in January
Figure 32 (Annexure IV) verifies the rapid wind speed changes, with locally relevant WM records taken at a height of 60m.
Solar PV simulation
Data verification
The data provided by GeoModel consists of 2010 annual GHI and DNI, ambient
Day 1 Day 2 Day 3 Day 4 records are defined in section 7.2 The temperature and the wind speeds are intended to be used as ancillary parameters The STERG and its director, Paul Gauché, have already published papers regarding the mentioned GeoModel data In terms of said data, the quality of the irradiance values is stated as being reliable (Gauché, Pfenninger 2012; Gauché, Heller 2012) The cooperation with GeoSun Africa © , which is a spin-off company of CRSES, and with the representatives of GeoModel in South Africa verifies the reliability of the data obtained Based on such quality references, GeoModel data will be used for further processing.
Methodology
The simulation method utilized a modified Microsoft Excel tool created by Paul Gauché, designed to model a central receiver (CR) concentrated solar power (CSP) plant Additionally, a solar photovoltaic (PV) model was developed as part of this approach (Gauché 2011).
The solar calculation tool provides a comprehensive analysis of the sun's hourly position, incorporating various parameters such as time, altitude, and azimuth, as well as universally applicable coefficients By taking into account mutual module shading, the model delivers accurate results that can be tailored to suit diverse solar PV applications, including fixed tilt, periodic adjustment, azimuth tracking, and full tracking systems.
For solar PV purposes, the tool processes the following input parameter:
Site’s coordinates – longitude, latitude [deg]
GHI, DNI, diffuse horizontal irradiance [W/m²]
Inverter efficiency [%] o Coefficients o Temperature efficiency [% per °C above 25°C] o Irradiance efficiency [% per W/m² below 1000W/m²] o Temperature rise coefficient [°C per W/m²] and computes, among others, the following values:
Maximum actual power output [Wmax]
Time series load behaviour [Wh]
Annual amount of energy [Wh]
The results are validated through randomly conducted single projects utilizing the System Advisor Model (SAM) from the US National Renewable Energy Laboratory (NREL), ensuring the reliability of the model.
The making of assumptions
The solar PV simulation requires the following modifications/assumptions in terms of a consistent approach:
Out of 27 solar PV systems, 25 were installed in a fixed position on a north-facing rack GeoModel has established an optimal tilt angle, ranging from 24 to 32°, tailored to specific locations to maximize annual energy yield, which has been incorporated into simulation methods (Suri, Cebecauer 2012, p 5).
A conversion factor was applied to translate peak power into a specific aperture area, as the model necessitates a PV size while only peak power data was provided by the developers Five suitable modules from various manufacturers, ranging from 240 to 250Wp, were selected, leading to a calculated average specific peak capacity of 167Wp/m² (refer to Annexure V).
The module efficiency was generalised to 15.1%, corresponding to an average value for the five chosen modules described above
The panels did not cast shadows on each other
Since the required DHI was not available from GeoModel, it was derived by means of the following formula Theta represents the zenith angle in the following equation:
At the start of the day, certain DHI values were negative but were later adjusted to zero, an issue attributed to inaccuracies in DNI/GHI simulations To validate this observation, solar irradiance data from the measurement station at Stellenbosch University was analyzed, revealing a similar occurrence.
The following coefficients were derived by Gauché (2011) and received from Stine and Geyer (2001) o Temperature efficiency −0.5% per °C o Irradiance efficiency 0.0125% per W/m² o Temperature rise coefficient 0.03°C per W/m²
For the 36MW CPV plant, a simplified and reproducible methodology was
The concentration lenses focused only on the DNI, which was multiplied by the efficiency and the net module size, as the following formula describes:
The efficiency was derived by means of a concentrator triple-junction solar cell, type 3C40, made by Azur Space Solar Power Ltd
Table 18: Solar CPV properties, type 3C40, Azur Space Solar Power Ltd, under STC
Based on the lack of temperature rise coefficients that could have designated the cell temperatures, the cell efficiency was reduced by means of a mean alternation between 25 and 80°C, and yielded 35.3%
Based on the above assumptions, the simulation was done for each project.
Results
The annual cumulative energy output from 25 fixed-tilt solar PV plants, one one-axis tracker plant, and one fully tracked CPV plant reached a total of 1,906 GWh The cumulative maximum power generated was nearly 900 MW, reflecting a 14.2% shortfall compared to the registered capacity of 1,049 MWp This 149 MW difference is attributed to the peak power measurements based on Standard Test Conditions (STC), which do not accurately represent daily variations in irradiance and cell temperature.
The capacity factors (CF) for fixed tilt solar plants ranged from 18% to 22%, indicating stable performance, while one-axis trackers achieved nearly 25% and Concentrated Photovoltaics (CPV) reached 28.5% Eleven independent power producer (IPP) forecasts for annual energy generation and a SAM simulation for a 75MWp plant at Kalkbult Solar PV corroborated these findings, showing minimal deviation (refer to Annexure V).
Spring generated the highest energy output, providing nearly 26% more than autumn The average daily energy production was 5.9 GWh in spring compared to 4.7 GWh in autumn.
Table 19: Seasonal characteristics of solar PV generation
The cumulative solar PV energy generation closely aligns with solar irradiance, which is affected by local weather conditions Without energy storage, solar PV primarily meets daytime demand but cannot guarantee a consistent base-load capacity In winter, limited irradiance leads to reduced output, exacerbating the inability to support evening peak loads Figure 12 illustrates the cumulative generation over two days in January.
Figure 12: Exemplary cumulative solar PV generation in January
Concentrated solar power simulation
Methodology and assumptions
The SAM provided a range of input parameters for project implementation, with default values assigned to the residual unknown parameters Business considerations were overlooked, and in addition to the specific weather data for each project, Table 20 below outlines several modified input parameters.
Table 20: CSP SAM – input parameters (CSP World 2013)
KaXu Solar One Bokpoort CSP Khi Solar One Technology Parabolic trough Parabolic trough CR
Storage 2.5h/ molten salt 9h/ molten salt 3h/ saturated steam
Annexure VI reports the remaining boundary conditions Some further parameters were plausibly adapted to obtain the hypothesised input values
The thermal storage dispatch control was consistently defined as follows (Gilman 2012), with no specified approach being required by the Single Buyer Office (SBO):
The turbine was operated at nameplate capacity, as long as sufficient energy was available from the solar field, or from thermal energy storage (TES)
The plant-generated electricity operated at nameplate capacity, using solar field energy with TES to cover low sunlight conditions
If there were no sunlight, the TES would dispatch energy, as long as there was some thermal energy in storage
The backup boiler does not operate, except for in response to thermal oil freezing protection issues
The model required ambient air temperature (TA), dew bulb temperature (Tdb), relative humidity (RH), and wind speeds for accurate analysis However, the available GeoModel record only included data on ambient air temperatures and wind speeds, lacking dew bulb and relative humidity information To address this gap, the closest WM data (WM02) were utilized for all three projects Although the WM's mean ambient temperature was lower than the site temperatures, it was assumed that TA, Tdb, and RH primarily affected the re-cooling process, resulting in manageable changes to the outcome Details on the dew point temperature calculation can be found in Annexure VI.
Upon finalizing the input parameters, the output for each plant was determined using the SAM tool, which provided a variety of metrics, including the essential net sent-out load.
Results
The annual energy output met the developers' expectations, with a relative deviation ranging from -0.8% to +4.1% The capacity factor (CF) was significantly higher compared to that of solar PV systems, attributed to available storage capacities and a solar multiple exceeding 1 A solar multiple of 1 indicates the necessary aperture area to provide adequate thermal energy for the power cycle to operate at its nameplate capacity under design conditions Results are detailed in Table 21 below.
Table 21: Verification of CSP simulation results
Energy in GWh KaXu Solar One Bokpoort CSP Khi Solar One
In the past year, a total of 752 GWh of energy was delivered, with a peak power output reaching 217 MW The energy storage dispatch varied significantly based on the storage capacity and seasonal irradiance levels Notably, KaXu and Khi Solar One, which possess a nameplate storage capacity of 2.5 to 3 hours, infrequently support the winter evening peak, despite the potential for generation to be adjusted for evening use.
Khi Solar One experienced a consistent decline in annual energy share after 18:00, with the number of days lacking energy generation rising significantly from 7% between 16:00 and 17:00 to 34% from 19:00 to 20:00 During the 19:00 to 20:00 timeframe, the plant could deliver a minimum power output of 18.5MW, despite its 50MW nameplate capacity While Khi Solar One may not guarantee a specific amount of firm capacity at all times, its contribution probabilities highlight its reliability compared to other fluctuating technologies Seasonal energy share trends are illustrated in Figure 13.
Figure 13: Khi Solar One – seasonal course with 3h storage
In summer, the Bokpoort CSP plant can deliver energy until 16:00 with a high probability of 93%, which decreases to 82% between 19:00 and 20:00, provided that the capacities exceed 45MW However, during winter, the plant's increased storage capacity is less effective due to insufficient solar input, preventing the storage from being fully charged while operating at its nameplate capacity.
As soon as temporal referred tariffs are implemented, the dispatchability of CSP plants with large storage facilities will have to be re-evaluated
The seasonal performance of Bokpoort CSP, illustrated in Figure 14, highlights significant energy contributions during winter and autumn between 10:00 and 17:00 This increase in energy output is primarily due to the inability to fully charge the storage system, leading to higher proportions of energy generation during these peak hours.
Figure 14: Bokpoort CSP – seasonal course with 3h storage
A sample power course is illustrated in Annexure VI, Figure 35.
Hydropower simulation
Methodology and assumptions
Both plants utilized a similar methodology for their approach The power calculations were derived from detailed project data and daily flow rate measurements in cubic meters per second, specifically from monitoring stations D7H014 and C8H036 of the DWA.
The following formula was used for the power calculation, with ρH2O standing for density, g for gravity, h for height difference, and ηsystem for the total system efficiency:
The required input parameters are shown below The numbers given in bold were not available, but were approximated or assumed
Max./Min flow rate [m³/s] through turbine
The maximum flow rate of Neusberg Hydropower was found to be inconsistent, as it was calculated using formula (7) based on the registered capacity Additionally, Neusberg Hydropower adopted the minimum flow rate and system efficiency parameters from Stortemelk Hydropower.
Results
The validation of the expected annual output by the developer demonstrated reliable results, with Neusberg showing a deviation of -0.5% and Stortemelk Hydropower at +8% The capacity factors were 82% for Neusberg and 76% for Stortemelk, yielding outputs of 71.5 GWh and 27.2 GWh, respectively Stortemelk Hydropower maintained a consistent annual output, consistently exceeding its design flow rate In contrast, Neusberg's flow rate on the Orange River experienced seasonal variations, with surplus flow during summer and autumn, including flooding, and reduced discharge in winter and spring, as illustrated in Figure 15.
Figure 15: Neusberg hydropower − annual course
The two approved hydropower plants demonstrated a maximum capacity of 14.1 MW and a minimum capacity of 5.4 MW, providing a superior base-load contribution compared to solar PV, wind, and CSP technologies Notably, the capacity factor (CF) for these hydropower plants was significantly higher, with full load hours surpassing those of alternative energy sources.
The study evaluates the cumulative output load behavior to forecast future electricity contributions in South Africa, while also assessing the role of each technology in ensuring supply security This evaluation aims to identify the strengths and weaknesses of the various technologies employed in the energy sector.
The findings are derived from the weather data collected in 2010, which includes anomalies caused by individual weather events Therefore, the simulation results do not reflect typical annual conditions over an extended timeframe.
Cumulated output
Overview of general results
The model predicts an annual energy output of 6,442 GWh, which includes 319 GWh from the Sere, Klipheuwel, and Darling wind farms, although these are not included in the REIPPPP framework.
A maximum occurring power of 2 302MW (27/03, 13:00–14:00), which constitutes 95% of the maximum possible capacity of 2 433MW
The firm capacity is defined as the minimum power output of 27.2 MW, occurring between 15:00 and 16:00 on October 20, which accounts for just 1.1% of the maximum potential capacity It is important to note that this firm capacity is a singular value, necessitating further analysis to evaluate the system's overall quality of minimum contribution.
The IRP 2010 prediction analysis forecasts a demand of 275TWh to 315TWh for 2015, with annual renewable energy distribution projected between 2.05% and 2.34% Additionally, the installed capacity is expected to reach 4.9% of the anticipated maximum demand load of 47GW for that year.
A breakdown of the different technologies is shown in Table 23 below
Table 23: Technology-specific annual energy yield
Wind power Solar PV CSP Hydropower Delivered energy [GWh] 3 685 1 906 752 99
Share of total occurring output 57% 30% 12% 1.5%
Share of max possible capacity 54% 37% 9% 0.6%
Approximately 57% of the annual energy output is generated through wind power, although the ratio of maximum output to total installed capacity is relatively low at 54% This indicates that wind power benefits from higher full load hours compared to the system's average In contrast, the dispatchability of Concentrated Solar Power (CSP) and the reliability of hydropower significantly enhance their annual contributions, with CSP achieving a yield of up to 133% and hydropower reaching as high as 250% of its capacity share.
The 2011 target outlined in the 2003 White Paper on Renewable Energy, which aimed for over 10,000 GWh by 2015, has not been met The government's estimates encompassed all tendered capacities, and a linear extrapolation indicates an annual output of 9,124 GWh, assuming biomass and biogas applications achieve full load hours comparable to hydropower It is expected that the announced capacity from bidding in R3 will be operational by 2016.
A duration curve was calculated to assess the frequency of system load occurrences, sorting 8,760 values by size and illustrating them over time Figure 16 showcases the total system load and highlights the contributions from each technology.
The total duration curve does not simply represent the sum of individual curves due to differing timings of maximum values Among the various energy sources, hydropower and wind power exhibit the most consistent performance, demonstrating the highest average power contribution over the year.
As derived from the table above, a classification into time-related quartiles (Qi) was done The interquartile range covered 50% of the distribution (4 380h) and was between 325 and
1 103MW The results are illustrated below
Table 24: System duration curve – classification into quartiles
The duration curve exhibits a high-power bandwidth regarding the fluctuation of the wind power and the solar PV The power was less than 632MW for 6 months each year
The seasonal distribution of energy sources was relatively balanced, with winter contributing 24% and summer 25.6% to the overall yield Notably, autumn was the peak season for wind energy, accounting for 27% of the annual output, while solar photovoltaic (PV) energy saw its highest contribution in spring, reaching 28%.
Contribution to winter demand peak
During winter, significant challenges related to supply security were observed around 20:00 (refer to Annexure 1, Part B, Figure 27) Consequently, an assessment of availability was conducted for the period between 19:00 and 24:00 to analyze these results.
The supply system's firm capacity exhibited seasonal variations, peaking in summer due to higher wind speeds, increased hydropower output, and the dispatchability of Concentrated Solar Power (CSP) Notably, Solar Photovoltaics (PV) contributed only during the hour from 19:00 to 20:00 in summer.
Figure 17: Winter and summer firm capacity – 19:00 to 01:00
The firm capacity represents a single value only Although it can be used to evaluate the merest temporal contribution, it does not provide information about the probability of occurrence
To effectively compare results with other fluctuating power systems, a frequency scale for specific hours, as illustrated in Table 27, is beneficial The frequency range shows a notable difference between winter and summer, with the gap narrowing from 19:00 to 24:00 due to a minimal contribution from Concentrated Solar Power (CSP) During summer, only 3% of power loads fell below 350MW from 19:00 to 20:00, compared to nearly 40% in winter This disparity highlights the seasonal gap while also indicating a lack of performance from wintertime systems.
The gap between summertime and wintertime decreases temporally until both graphs are almost equal (supplementary information in Annexure VII)
Figure 18: Frequency distribution during summer and winter (A)
Figure 19 illustrates the seasonal contribution of various energy sources, highlighting that wind power accounts for the majority of the contributed load The reduction in Concentrated Solar Power (CSP) usage is attributed to limited storage capabilities, with only one of the three plants equipped with storage exceeding three hours of nameplate capacity.
Figure 19: Mean power distribution during winter and summer – 19:00 to 24:00
Between 19:00 and 24:00, energy delivery declined from 235 GWh in summer to 187 GWh in winter, highlighting seasonal variations in energy production Additionally, the proportion of energy generated by wind power compared to concentrated solar power (CSP) shifted, while solar photovoltaic (PV) systems contributed minimally, as indicated by the yellow band.
Fluctuation characteristics
Due to the inherent volatility of renewable energy sources, power output can vary significantly over short time frames This chapter explores the susceptibility to hourly fluctuations, providing insights into the characteristics of generation volatility.
Fluctuation refers to the rate of change in power over time, with 8,760 alterations recorded annually The peak variations reached +960MW and -1,073MW Seasonal analysis using standard deviation revealed minor differences, with winter showing a standard deviation of 199MW and spring exhibiting a deviation of 182MW.
A repeated classification into quartiles revealed significant findings, highlighting the presence of extreme values Notably, the ratio of the lower quartile power (Q0.25) to the maximum power was only 18%, indicating substantial fluctuations in the data.
25 below) The median was proportionally low (10%) in comparison to the ceiling power change, which confirmed that the majority of the values fell within a limited bandwidth
Figure 20 depicts the systems absolute- and a real fluctuation duration curve
Table 25: Fluctuation duration curve – classification into quartiles
An exemplary power course is illustrated in Figure 16 below
Figure 21: Exemplary single and cumulative power course – 01/01/2010 – 11/01/2010
Conclusion
This thesis evaluates annual renewable energy output, analyzes load behavior, assesses contributions to winter peak demand, and examines fluctuations Through this comprehensive approach, the scientific question posed is effectively addressed.
This study explores various technology-driven methods utilized to calculate the total annual load for 2015, using data from 2010 It simulates all approved Independent Power Producer (IPP) projects up to the financial closure of the Renewable Energy Independent Power Producer Procurement Programme (REIPPPP) bidding round 2, which concluded on May 9, 2013, along with three additional authorized renewable energy initiatives.
Conclusions drawn from the work
The conclusion of the thesis is split into the following aspects that summarise the most important findings of the work:
In 2015, the annual delivered energy reached 6,442 GWh, accounting for only 2.05% to 2.34% of the anticipated electricity demand A linear extrapolation from the bidding results of R1 and R2 to R3 suggests a potential output of 9,124 GWh Given these figures, the government's goal to exceed the 10,000 GWh threshold appears unattainable.
Wind farms across the country exhibit consistent wind speeds spatially, yet experience significant temporal fluctuations Despite these variations, less than 1% of firm capacities can be assured, indicating a need for further analysis in this area.
Solar PV systems do not significantly address evening peak demand, with their actual capacity being 14% lower than the registered peak capacity as per STC specifications However, retail prices for solar PV and wind power generation are the most competitive, highlighting their strong economic viability.
Concentrated Solar Power (CSP) provides significant flexibility in ensuring a secure energy supply, particularly during peak evening demand periods With capacity factors nearly 2.5 times greater than those of solar photovoltaic systems, CSP can enhance the reliability of renewable energy generation By integrating more advanced technical solutions into the energy supply mix, the overall availability of electricity generated from renewable sources can be increased.
The analysis revealed a significant difference in contributions between summer and winter, particularly in the frequency distribution from 19:00 to 24:00 Notably, winter contributions were 20% lower compared to those in summer.
The volatility of sources feeding into the public grid increases the demand on power hubs and overall system burden When a regulatory strategic electricity plan mandates a higher share of peaking distribution, it becomes essential to consider the integration of storage capabilities.
The shock rate of the system that expresses fluctuation in behaviour demonstrated high, but temporally rare, amplitudes The median was 10 times less, which implies a smoothed course
The methodology employed in this study involved several uncertainties and assumptions that significantly impacted the outcomes Each assumption associated with the various approaches has been explicitly outlined and scientifically justified The primary objective of each method was to create a single, applicable approach, utilizing external, reliable information sources, such as developer expectations, solely for validation purposes.
Throughout the development process, various new perspectives and challenges emerged that could not be addressed within the rigid structure of the thesis To explore these unresolved issues, additional research is necessary and has already been planned for the upcoming period by the CRSES.
The thesis introduces a pioneering forecasting method that analyzes tendered capacities on an annual basis, highlighting the strengths and weaknesses of the evolving renewable energy mix The findings offer valuable insights for policymakers, as they now have access to detailed annual projections that can inform future energy decisions.
This thesis serves as a foundational step in predicting the renewable energy supply for South Africa's industrial sector It is part of a continuous effort to maintain current results, incorporate additional guidelines, and include future project opportunities.
Research in the following fields is required:
A focus on the contribution that is made by each technology, to gain a more detailed forecast and to take such technologies as landfill gas, biomass, and others into account
An improved method of evaluating the simulated results for every single approach
The susceptibility of models to aberrations is influenced by uncertain assumptions, and assessing each met boundary value allows for error minimization through simultaneous adjustments Each aberration can be quantified by its corresponding error bandwidths.
A closer collaboration than at present with the weather records, enabling GeoModel
An additional examination of the weather data in terms of long period records would reduce the dependency on unforeseen weather phenomenon, and provide an ordinary annual output
A prospective, ongoing evaluation of the model’s results, according to real values from 2015 onwards
Wind Speed Data at Different Heights and Its Impact in the Wind Energy Resource Assessment in a Region
Wind Farm - Technical Regulations, Potential Estimation and Siting Assessment, 97 – 114 Orlando Suvire
BP Statistical Review, 2012 British Petrol Statistical Review of World Energy, June 2012 United Kingdom
Available from: http://www.bp.com/content/dam/bp/pdf/Statistical-Review- 2012/statistical_review_of_world_energy_2012.pdf [Accessed 11/07/2013]
Centre for Renewable and Sustainable Energy Studies (CRSES), 2013 AcrMap 2010 – electricity grid image Stellenbosch, South Africa
Chamber of Mines of South Africa, 2007 Annual Report 2006 – 2007 Mining together Available from: http://www.bullion.org.za/documents/text%20compressed.pdf [Accessed 11/07/2013]
In March 2013, Vodacom unveiled Africa's largest solar rooftop, showcasing its commitment to renewable energy and sustainability This significant development highlights Vodacom's leadership in the telecommunications industry and its efforts to reduce carbon emissions The solar installation is part of a broader initiative to promote eco-friendly practices within the company For more information, visit Engineering News Online.
CSP World – World News about Concentrated Solar Power, 2013 Servicios Avanzados de Comunicación y Marketing SL (SACM) Spain Available from: http://www.csp-world.com [Accessed 20/06/2013]
Department of Energy (DoE), 2009 Digest of South African Energy Statistics 2009 Directorate Pretoria, South
Africa Available from: http://www.energy.gov.za/files/media/explained/2009%20Digest%20PDF%20version.pdf [Accessed 15/04/2013]
Department of Energy (DoE), 2010 Annual Demand forecast 2010 Forecast - IRP 2010 Parameter, Overview sheet Pretoria, South Africa Available from: http://www.energy.gov.za/IRP/factsheet.html [Accessed 29/04/2013]
Department of Energy (DoE), 2013 Formal homepage, Available from: http://www.energy.gov.za/files/petroleum_frame.html [Accessed 22/04/2013]
Department of Energy (DoE), 2013a Renewable Energy Independent Power Producer Procurement Programme, formal homepage 2013 South Africa Available from: http://www.ipprenewables.co.za/#site/index [Accessed 13/05/2013]
Department of Energy (DoE), Aug 2011a Question and Answer for IPP Briefing session 31 Media &
Publications., South Africa Available from: http://www.energy.gov.za/files/media/pr/media_pressreleases2011.html, [Accessed 10/05/2013]
In August 2011, the Department of Energy (DoE) of South Africa released a media statement regarding the Renewable Energy Independent Power Producer Programme This initiative aims to promote renewable energy sources in the country For more information, visit the official DoE website at http://www.energy.gov.za/files/media/pr/media_pressreleases2011.html, accessed on December 5, 2013.
Department of Energy (DoE), August 2011c Request for qualification and proposals for new generation capacity under the IPP procurement programme, Part A – general requirements, rules and provisions Tender
Department of Energy (DoE), August 2011d Request for qualification and proposals for new generation capacity under the IPP Procurement Programme, Part B – Qualification Criteria Tender No.:
Department of Energy (DoE), August 2011e Request for Qualification and Proposals for New Generation Capacity under the IPP Procurement Programme, Volume 5: Economic Development Requirements Tender
Department of Energy (DoE), December 2011f Renewable Energy Independent Power Producer Announcement 2011 − Summary of Evaluation Findings Renewable Energy Independent Power Producer
Department of Energy (DoE), December 2012 IPP Procurement Programme 2012 Determination under
Section 34(!) of the Electricity Regulation Act 4 of 2006 No 1074 South Africa
Department of Energy (DoE), May 2011 Electricity Regulation Act (ERA) No.4 of 2006, Electricity Regulations on New Generation Capacity Government Notice Pretoria, South Africa
Department of Energy (DoE), May 2012a Renewable Energy IPP Procurement Programme, Window Two Preferred Bidders Announcement 21 May 2012 South Africa
Department of Energy (DoE), May 2012b Window Two Preferred Bidders’ Announcement 2012, Renewable
Energy Independent Power Producer Procurement Programme South Africa
Department of Energy (DoE), September 2009a Integrated Resource Plan for Electricity 2009 (IRP 2009) Preliminary Report South Africa
Department of Minerals and Energy (DME), December 1998 White Paper on the Energy Policy of the Republic of South Africa ISBN: 0-9584235-8-X Pretoria, South Africa
Department of Minerals and Energy (DME), July 2003 Free Basic Electricity Policy Electricity Basic Services
Support Tariff Government Gazette No 25088, Pretoria, South Africa
Department of Minerals and Energy (DME), March 2003a Integrated Energy Plan for the Republic of South Africa 2003 (IEP), South Africa
Department of Minerals and Energy (DME), May 2002 Energy Outlook for South Africa: 2002, Eskom Holding, Energy Research Institute – University of Cape Town, South Africa
Department of Minerals and Energy (DME), November 2003b White Paper on Renewable Energy 2003 South Africa
Eskom, 2012 Annual demand, hourly based – featured by MS Excel Provided by Paul Gauché Source:
Eskom, 2013 Transmission Ten Year Development Plan 2013 – 2022 Eskom Transmission Division, Eskom
Holding, Johannesburg, South Africa Available from: www.eskom.co.za[Accessed 29/04/2013]
Fritz W., October 2012 Renewable Energy Feed in Tariffs, REBID, SASGI and the Smart Grid Cape
Peninsula University of Technology Journal: Energize, p 80 Stellenbosch, South Africa
Gauché, P., 2011 [Microsoft Excel tool] Stellenbosch South Africa, Conn.: Solar Thermal Energy Research Group
Gauchộ, P., Pfenninger, S., Meyer, A.J., Von Backstrửm, T.W and Brent, A.C., 2012 Modeling Dispatchability Potential of CSP in South Africa South African Solar Energy Conference (SASEC) Stellenbosch, South Africa
GeoSun Africa, 2008 South Africa Maps, Spin-off company of the Centre for Renewable and Sustainable Energy Studies Stellenbosch, South Africa, Available from: http://geosun.co.za/solar-maps/ [Accessed 12/04/2013]