Repurposing South Africa’s Retiring Coal-Fired Power Stations for Renewable Energy Generation: A Techno-Economic Analysis

Abstract

South Africa is one of the most carbon-intensive economies in the world, but it is presently experiencing an energy crisis, as its utility company cannot meet the country’s energy demands. The use of renewable energy sources and retiring of coal-fired power stations are two important ways of alleviating this problem, as well as decarbonizing the grid. Repurposing retiring coal-fired power stations for renewable energy generation (RCP-RES) while maintaining energy sustainability and reliability has rarely been researched. This paper proposes macro- and microelements for repurposing retiring coal-fired power stations for renewable energy generation in Camden with the aim of improving power generation through a low-carbon system. In this model, concentrated solar power (CSP) and solar photovoltaics (SPV), in combination with storage technologies (STs), were employed for RCP-RES, owing to their excellent levels of availability in the retiring fleet regions. The simulation results show that the power densities of CSP and SPV are significantly lower compared with retiring a coal-fired power plant (CFPP). Both are only able to generate 8.4% and 3.84% rated capacity of the retired CFPP, respectively. From an economic perspective, the levelized cost of electricity (LCOE) analysis indicates that CSP is significantly cheaper than coal technology, and even cheaper when considering SPV with a storage system.

1. Introduction

For more than a decade, South Africa has been battling an energy crisis due to failing electricity infrastructure resulting in rolling blackouts, which has negatively impacted the economy [1]. South Africa is one of the most carbon-intensive economies in the world [2] and has some of the highest coal consumption per capita [3]. In order to fulfill the global agreement on climate change, South Africa has committed to significantly reducing all forms of carbon extensive electricity generation stations within the country, thereby compounding its energy demand issues. The transition to sustainable energy systems requires the country to decommission coal-fired power stations, which represent 79% of the country’s electricity generation base. To ameliorate the energy crisis, healthy reserve margins have been considered critical performance metrics in deterring loading shedding events [4,5]. Hence, the South Africa electricity utility company, known as Eskom, has cited a reserve margin estimated at 8% or below as against the international benchmark of 15% [6], resulting in a demand deficit for Eskom. Thus, load shedding is the only control mechanism the utility company can employ to protect the grid against the strain of demand.

In South Africa, most of the coal power plants have been in operation for several decades, and many of the plants have reached their expected end-of-design life span since 2019, as they were designed and constructed for 50 years in operation [7,8]. Eskom proposed a plan for decommissioning the coal power plants that have exceeded their life span. The Integrated Resource Plan (IRP) calls for the decommissioning of 13,410 MW coal power plants from 2019 to 2020, and further decommissioning of 24,100 MW coal plants during the period of 2030 to 2050 [9]. This activity is part of South Africa’s obligations to climate change improvement by reducing greenhouse gas emissions from coal-fired power plants. The IRP establishes that 80% of the country’s greenhouse gases come from the energy sector, while 50% of these emissions come from electricity generation and liquid fuel production [9,10]. Therefore, there is a need to reduce emissions contributed by coal-fired power plants. One of the alternatives is the use of renewable energy sources, which can help reduce emissions and improve energy efficiency [11,12,13,14].

Many studies have discussed the transition from CFPP to natural gas. According to [9], United States Energy started the retirement of coal power plants in 2019 to transition to natural gas. The U.S. Energy Information Administration (EIA) revealed that 121 U.S. coal-fired power stations were repurposed to burn other fuels between 2011 and 2019, while another 103 coal-fired plants were converted to natural-gas-fired power plants. In [15], two main methods were discussed for converting coal-fired plants to natural gas. The first method involves the retirement of a coal-fired plant by replacing it with a new integrated natural gas engine (NGCC), while the second involves the conversion of coal-fired broilers to an alternative fuel by converting the fuel feeding system. From 2011 to 2019, a total of 17 coal-fired power plants in United States were replaced with new NGCC plants. The new NGCC plants have a combined power output of 15.3 GW, 94% more than the 7.9 GW capacity of the power stations that were converted by simply altering the fuel feed system and fuel type of coal-fired power plants [9,10]. The conversion to natural gas has the advantage of dispatchability, but it is still characterized by greenhouse gas emissions [16].

Within the literature, the retirement of coal-fired plants into renewable energy (RE) technology in order to fulfill the global climate change agreement has been considered. Ref. [17] presents renewable energy technology as a possible energy source for the future when replacing the retired coal-fired plants in Illinois, aiming to improve the health of the residents through emission reductions and saving money. In [18], the authors discussed the possibility of repurposing existing coal-fired power plants into CSPs in Chile. In order for the Chilean power industry to fully decarbonize, the study considered the retrofit of CFPPs with state-of-the-art CSP technology of 250 MWe. Similarly, in [19], a retrofit decarbonization of coal power plants was discussed for Poland. A small modular reactor or geothermal power station was proposed for the replacement of coal-fired plants with the aims of lowering emissions as well as maintaining the existing site with minimum energy generation. Ref. [20] proposes a novel model for retiring coal-fired power plants and the installation of renewable energy sources with a view to expand both the generation and transmission systems. Authors [21] assessed whether future RE generation with the storage system available in Nova Scotia, Canada, will be sufficient for retiring coal-fired power generation.

The transition of energy is an emerging research area that is presently being explored, but limited studies have been conducted on areas that seek to analyze the technical and economic metrics of converting coal-fired power stations to produce baseload noncombustible RE technology, especially in Africa. In this study, we aimed to perform a techno-economic assessment of repurposing coal-fired power plants in the Camden Power Station in South Africa into a baseload renewable energy generation station in order to reduce, and ultimately eliminate, the energy crisis South Africa is experiencing, while also addressing the country’s climate action commitments and avoiding excessive capital outlay by utilizing existing assets. A review of the literature shows that the full potential of RE technologies can be harnessed in South Africa [3,22,23]. This paper presents macro- and microelements for repurposing retiring coal-fired power stations for renewable energy generation in Camden with the aim of improving power generation through low-carbon systems. In this model, concentrated solar power (CSP) and solar photovoltaics (SPV), in combination with storage technologies (ST), were employed for RCP-RES owing to their excellent levels of availability in the retiring fleet regions. A modified IEEE 39 test bus system was simulated, using DigSILENT Power Factory software, as a case study with an RCP-RE configuration to validate the proposed system. A comparative load flow study was conducted with and without an RCP-RE system.

The remainder of this paper is organized as follows: Section 2 discusses the modelling methodology employed for the repurposing of the coal-fired power station into a baseload RE system. In Section 3, the presented simulation results are discussed, and the paper is concluded in Section 4.

2. Methodology

The fundamental data requirements for the RCP-RE analysis were provided using data funneling, as given in Figure 1. The set of refined data obtained were then deployed for the techno-economic analysis for the conversion of coal-fired plants to RE generation.

2.1. Capacity Resizing

The parameters for the coal-fired power plant model presented in

Table 1. Parameters for coal-fired power model.

	Average Heat Rate, kJ/kWh	Net Efficiency, %	CO2 Emission, kg/MWh	SOx Emission, kg/MWh	NOx Emission, kg/MWh	Particulates
Coal (Camden)	9707	37.1	930.2	9.03	1.94	0.13
Coal with Flue Gas Desulphurization	9812	36.7	947.3	0.46	1.94	0.13
Coal Carbon Capture System	14,106	25.5	136.2	0.66	0.42	0.18

show the key indicators in relation to climate change. The renewable energy funneling metric is presented in

Table 2. Renewable energy funneling metric.

	Renewable Energy Resource	Noncombustion	Dispatchability	Resource Availability	Overall Suitability
Bioenergy	✓	X	✓	X	X
Concentrated solar	✓	✓	✓	✓	✓
Hydroelectric	✓	✓	✓	X	X
Geothermal	✓	✓	✓	X	X
Solar Photovoltaic	✓	✓	X	✓	✓
Wind power	✓	✓	X	X	X

and portrays whether each renewable energy source meets the qualitative characteristics required, which are noncombustibility, dispatchability, or ST compatible technology required.

Whether the qualitative requirements of RE technologies are met is determined using Equation (1), which helps ascertain the RCP-RE sizing that can be achieved for the different technologies [13].

$$P{D}_e=PD\times {\eta }_{eff}\times CF\times \partial$$

(1)

where PD_e represents the effective power generation in W/m²; PD represents the function of the preconverted power density in W/m²; ${\eta }_{eff}$ is the conversion efficiency; CF is the capacity factor; and $\partial$ is the infrastructure requirement scale, representing the additional area required for mines, roads, foundation pipelines, etc.

To assess the suitability of ST based on the set-out criteria, mainly being that the ST is suitable for a utility-scale project, we considered a medium to long discharge cycle and negligible self-discharge. Using Equations (2) and (3), we ascertained the space required by such storage facility. The ST energy capacity sizing is expressed as [24]:

$$S{T}_{ec}=D_E\times \frac{1}{DoD}\times \frac{1}{\eta }$$

(2)

where ST_ec represents the storage technology system energy capacity in MWh, and D_E is the energy demand required for a certain number of days.

The land footprint required for the ST system installation is computed as:

$$Space required=\frac{S{T}_{ec}}{S{T}_{ed}}$$

(3)

where ST_ed represents the energy storage system in Wh/L.

In this study, based on the outcome of the feasibility study of RE resources potential at the Camden location, the RE technologies found to be most suitable and feasible for RCP-RE were SPV and CSP, as presented in Table 2, based on the overall suitability. When considering storage technologies, lithium-ion and CAES were considered based on the high energy density required for utility scale deployment. Lithium-ion is favorable for electrochemical STs, whereas CAES is favorable for mechanical ESS systems. The capacity of the existing coal-fired power station in Camden is 1600 MW, as shown in

Table 3. Parameters for modeling CFPP, RCP-RE, and ST models.

Description	Capacity Factor, %	Round Trip Efficiency, %	Life Cycle, Years	Sizing, MW
CAES	-	52	40	60
Coal	85	-	30	1600
Coal (existing plant)	85	-	30	1600
Coal with FDG	85	-	30	1600
Coal with CCS	85	-	30	1600
CSP with storage system	54	-	25	168
Lithium-ion	-	86	20	60
SPV	20	-	25	319

. The baseload of SPV power generation we considered is 60 MW, found at 6.006 MW/km², and the power density is 9.61 MW/km², resulting in a capacity of 168 MW compared with Camden’s power density of 439 MW/km² with reference sizing of 1600 MW. A 60 MW capacity CAE or lithium-ion ST system is required to match the maximum power density of the SPV system, as shown in Table 3.

2.2. Technical Analysis

To analyze the impacts of RCP-RE on the power system, a load flow study on the network is essential. In this study, we used a modified IEEE 39 test bus system as a case study to validate the proposed RCP-RE model with several identified scenarios, as shown in Table 3. The IEEE 39 test bus system has similar characteristics to those of the Camden Power Station, and it represents all elements and nuances found across generation and transmission. The IEEE 39 bus test system comprises of thirty-nine buses, ten generators, thirty-four transmission lines, twelve transformers, and nineteen loads, as shown in the single diagram presented in Figure 2. The three most important case studies considered were:

• Case one: Existing IEEE 39 test bus system.
• Case two: Modified test bus system with 60 MWe Solar PV system and ST cases.
• Case three: Modified test bus system with 168 MWe CSP.

In case two, we modified the IEEE 39 test bus system with a generator G3 sizing change to match the solar PV and the two ST cases to simulate the integration of the RCP-RE system. As shown in Table 3, the generator G3 parameters were amended from that of a synchronous machine with a 650 MW capacity to represent the SPV RE generation solution coupled with ST that was modeled to provide a baseload power of 60 MW. This case is representative of both the CAES and lithium-ion ST combinations cases. In case three, the IEEE 39 test bus system was changed from that of a synchronous machine to match the 168 MW CSP RE generation solution, as shown in Table 3.

2.2.1. Load Flow Modeling

Here, the load flow of the IEEE 39 test bus system, together with the modified test bus system, was used to analyze the technical impacts of RCP-RE on the CFPP grids. In the load flow calculation, each bus was categorized into one of the following three bus types:

• Swing bus, which is also known at the slack bus. There is one swing bus, and it is also referred to as bus 1. The swing bus is the reference bus for which V has angle δ_k, typically 1.0 p.u.
• Load (PQ) bus: P_k and Q_k are input data. The power-flow program computes V_k and δ_k. Most buses in a typical power-flow program are load buses.
• Voltage controlled (PV) bus: P_k and V_k are input data. The power flow calculation computes Q_k and δ_k. Examples are buses to which generators, switched shunt capacitors, or static var systems are connected.

Therefore, the power delivered to bus k, separated into generator and load, is expressed as:

$$P_k=P_{GK}-P_{LK}$$

(4)

$$Q_k=Q_{GK}-Q_{LK}$$

(5)

where P_k and Q_k are the real power and reactive power supplied to bus k, respectively; P_Gk and Q_Gk are the real and reactive generator bus, respectively; and P_Lk and Q_Lk are the real and reactive load bus, respectively.

The bus admittance matrix Y_bus was constructed from the line and transformer input data. The elements of Y_bus were used to compute the nodal equations for a power system network as follows:

$$I=Y_{bus}V$$

(6)

where I is the vector source of currents injected into each bus, and V is the vector of bus voltages. For bus k, the nodal equation is expressed as:

$$I_k={{\displaystyle \sum }}_{n=1}^N{Y}_{kn}{V}_n$$

(7)

where Y_kn is the N vector of the line admittance at bus k. The complex power $S_k$ delivered to bus k is expressed as:

$$S_k=P_k+j{Q}_k=V_k{I}_k$$

(8)

The power flow solutions by Gauss Seidel are based on nodal equations, where each current source I_k is calculated as:

$$P_k+j{Q}_k=V_k{\left[{{\displaystyle \sum }}_{n=1}^N{Y}_{kn}{V}_n\right]}^{*} k=1,2\dots \dots ,N$$

(9)

$$V_n=\left|V_n\right|e^{j{\delta }_n}$$

(10)

$$Y_{kn}=Y_{kn}{e}^{j{\theta }_{kn}}=G_{kn}+j{B}_{kn} k=1,2\dots \dots ,N$$

(11)

$$P_k+j{Q}_k=V_k{{\displaystyle \sum }}_{n=1}^N{Y}_{kn}{V}_n{e}^{j\left({\delta }_k-{\delta }_n-{\theta }_{kn}\right)}{}^{*}$$

(12)

$$P_k=V_k{{\displaystyle \sum }}_{n=1}^N{Y}_{kn}{V}_n\cos \left({\delta }_k-{\delta }_n-{\theta }_{kn}\right)$$

(13)

$$Q_k=V_k{{\displaystyle \sum }}_{n=1}^N{Y}_{kn}{V}_n\sin \left({\delta }_k-{\delta }_n-{\theta }_{kn}\right)$$

(14)

$$P_k=V_k{{\displaystyle \sum }}_{n=1}^N{V}_n\left[G_{kn}\cos \left({\delta }_k-{\delta }_n\right)+B_{kn}\sin \left({\delta }_k-{\delta }_n\right)\right]$$

(15)

$$P_k=V_k{{\displaystyle \sum }}_{n=1}^N{V}_n\left[G_{kn}\sin \left({\delta }_k-{\delta }_n\right)-B_{kn}\cos \left({\delta }_k-{\delta }_n\right)\right]$$

(16)

where G_kn and B_kn are the N vector line conductance and susceptance at bus k, respectively; ${\delta }_k, {\delta }_n,\mathrm{and} {\theta }_{kn}$ are the voltage bus angles.

2.2.2. Economic Analysis

This section describes the economic impacts of repurposing retiring coal-fired power stations for renewable energy generation in the grid. The levelized cost of energy (LCOE) is considered fundamental for the economic evaluation of the RCP-RE, and this allows for the comparison of cost across differing generation costs. LCOE allows for comparisons between multisized power generation technologies, lifecycles, complete project costs, and build timelines. LCOE is a function of the total discounted cost over the electricity generated, and it is mathematically expressed as [14]:

$$\mathrm{LCOE}=\frac{\mathrm{Total}\mathrm{Discounted}\mathrm{Expenses}}{\mathrm{Total}\mathrm{Discounted}\mathrm{Power}}$$

(17)

The LCOE is a function of many factors, such as generation technology’s cost of capital, fuel, operations, and maintenance; it is expressed as:

$$\mathrm{LCOE}=\frac{{{\displaystyle \sum }}_{t=1}^n\frac{\left(I_t+M_t+F_t\right)}{{\left(1+r\right)}^t}}{{{\displaystyle \sum }}_{t=1}^n\frac{\left(E_t\right)}{{\left(1+r\right)}^t}}$$

(18)

where I_t represents the capital outlay at a specific point in year t; M_t represents the operations and maintenance costs at a specific point in year t; F_t represents input energy/fuel cost at a specific point in year t; E_t is the generated electricity at a specific point in year t; r is the weighted cost of capital/discount/interest rate; and n represents the life span of the power plant.

Table 4. Summary of LCOE input parameters.

Description	Capital Cost, ZAR/kW	Fixed O&M Cost ZAR/kW	Variable O&M Cost ZAR/kW	Fuel Cost, ZAR/GJ	Capacity Factor, %	Round Trip Efficiency, %	Life Cycle, Years
CAES	27,454.05	274.7	3.45	63.9	-	52	40
Coal	34,557	789	0.07	22.3	85	-	30
Coal Plant (Camden)	2809.98	789	0.07	22.3	85	-	30
Coal with FDG	43,062	1136.5	0.09	22.3	85	-	30
Coal with CCS	81,617	1912.5	0.17	22.3	85	-	30
CSP with storage system	96,623.78	-	0.2	-	54	-	25
Lithium-ion	30,859.07	164.45	-	-	-	86	20
Solar PV	21,729.65	301.02	-	-	20	-	25

presents a summary of the commercial and performance LCOE input parameters of each reviewed technology and option. All costs presented are in South African Rands.

The LCOE allows for a split of each of the cost components to distinguish between capital, maintenance, and fuel costs. The concept of time value for money is considered in LCOE to ensure that the costs are reduced, and this is referred to as discount [14,26]:

$${\left(1+r\right)}^t$$

(19)

where r represents the discount, and t is the year.

3. Results and Discussion

In this section, the performance of the RCP-RE systems is evaluated on the three different case studies, and the simulation results are discussed based on the technical and economic impacts on the power system network. The simulation was carried out for the three cases using DigSILENT Power Factory software (Johannesburg, South Africa) for the load flow analysis.

In case one, which involved the existing IEEE 39 test bus system, the power flow simulation was solved on the fourth iteration using the Newton Raphson algorithm. Figure 3 presents the voltage angle at each bus in degrees, while Figure 4 shows both the real and reactive power outputs of the generators. The simulation results obtained for case one indicated that the bus voltages are well-maintained, and all lines are loaded within their limits. Additionally, they revealed that all the generators are operating within their operating limits, and the error reference sources were not found in the simulation.

In case two, which involved the modification of the IEEE 39 test bus system with a 60 MWe solar PV system and ST cases, the simulation considered the integration of the dispatchable RE system in the grid. In this instance, generator G3 parameters were changed to represent an SPV-ST model system providing baseload power at 60 MW. This case was a representation of both the CAES and lithium-ion storage system options. Figure 5 presents the voltage angle at each bus in degrees, while Figure 6 shows both the real and reactive power outputs of the generators of this case study. The simulation results obtained showed that due to changing the generator G3 parameters, the grid lost stability, as generators G2, G3, and the slack generator overloaded and operated outside of their limits. Generator G2 reached an active power of 1109.17 MW and reactive power of 360.42 MVAr, resulting in 166.61% loading, while generator G3 operated at its set dispatch active power level of 60 MW but exceeded its reactive power rating with a value of 179.82 MVAr. This resulted in generator G3 having a loading of 315.95%. Similarly, the transmission line 21–22 operated at above 95% loading with no reserve capacity.

In case three, the IEEE 39 test bus system was modified with a 168 MWe CSP system. In this instance, the dispatchable RE system was a CSP, and the generator G3 parameters were modified to represent a CSP model providing baseload power at 168 MW. Figure 7 presents the voltage angle at each bus in degrees, while Figure 8 shows both the real and reactive power outputs of the generators in this case study. The obtained results showed that upon changing the parameters of generator G3 to a dispatch power of 168 MW, the system lost stability. The slack generator, generators G2, and G3 overloaded and operated outside their limits. Generator G2 reached an active power of 1001 MW and a reactive power of 316.59 MVAr at 149.98% loading, while generator G3 operated at its set dispatch active power of 168 MW but exceeded its reactive power rating at a value of 171.932 MVAr. The transmission lines 5–6 and 21–22 operated at above 90% loading with no reserve capacity.

From the economic perspective in the simulation,

Table 5. LCOE costs and capacity metric.

Description	Charging Cost, ZAR/MWh	Fuel Cost, ZAR/MWh	O&M Cost, ZAR/MWh	Capital Cost, ZAR/MWh
CAES	3458.06	-	47.04	587.04
Coal	-	215.51	291.42	615.49
Coal Plant (Camden)	-	215.51	291.42	572.72
Coal with FDG	-	218.09	415.15	766.97
Coal with CCS	-	311.98	713.26	1453.67
CSP with storage system	-	-	197.39	2665.76
Lithium-ion	-	-	137.04	3680.88
Solar PV	1798.19	-	174.42	1623.77

presents an overview of the LCOE and capacity metric achieved for the CFPP, RPC-RE, and ST models. The obtained results showed that the cost of using clean coal technology CCS is higher than the cost of generating electricity from a greenfield CFPP. The generation cost from SPV is higher than that of a CFPP but significantly cheaper than that of clean coal technology with an integrated FGD or CCS. However, when pairing SPV with ST using the CAES or lithium-ion technology options, the cost of the SPV-ST system is reduced.

4. Conclusions

This paper proposes macro- and microelements for repurposing retiring coal-fired power stations for renewable energy generation in Camden with the aim of improving power generation through low-carbon systems with a major focus on retiring Camden’s coal-fired station. In order to analyze the technical and economic implications of RCP-RE models, a load flow analysis was carried out using DigSILENT Power Factory software, and the LCOE was considered for the economic impacts. The models were validated using the IEEE 39 test bus system. The simulation results showed that the reduction in capacity of generator G3, due to the low energy density of RPC-RE solutions, leads to the instability of the grid. The modified grid with a nearby generator needs to increase its output to compensate for the drop in generator G2’s power output. The LCOE evaluation revealed that the CSP is significantly cheaper than coal technology and even cheaper when considering SPV with a storage system. Finally, the research showed that there are no technical or economic constraints to achieving the retiring coal power station conversion into renewable energy generation. In addition, further expenditure will be required to reach steady-state stability, due to network constraints created by the reduced generator sizing resulting in an increase in output of nearby generators and increased line load capability.