- PyPSA-ZA: Investment and operation co-optimization of integrating wind and solar in South Africa (1.87 MB)
A new, independently produced techno-economic model of South Africa’s cost-optimal power generation mix in 2040 outlines a system where 69% of the electrical energy is produced from onshore wind and solar photovoltaic (PV) generators, supported by batteries and gas-fired generators.
Conducted by the Frankfurt Institute for Advanced Studies (FIAS), the study concludes that there will be no need for the addition of new coal or nuclear power stations beyond what is already installed.
In fact, FIAS’s Jonas Hörsch tells Engineering News Online that the results highlight South Africa’s “fortunate position” of having access to high-quality wind and solar resources, which will result in wind and solar PV energy production costs well below those that will be possible from future thermal-generation installations.
Constructed using open-source energy modelling software, known as Python for Power System Analysis (PyPSA), the study has drawn the bulk of its assumptions from the draft Integrated Resource Plan (IRP) base case, published by the Department of Energy (DoE) at the tail end of 2016. Some of the information was also sourced from South Africa’s Council for Scientific and Industrial Research (CSIR).
The DoE document makes pricing assumptions for various generation technologies and assumes, optimistically, that total yearly electricity demand will grow to 428 TWh by 2040, from around 250 TWh currently.
However, the study’s parameters deviate from the draft IRP by aligning wind and solar PV costs with the R0.62/kWh achieved during the most recent bid window of the Renewable Energy Independent Power Producer Procurement Programme.
‘Moderate’ learning rates have been assumed that halve the cost reductions forecasted by Bloomberg New Energy Finance. The analysis by FIAS assumes that, by 2040, the overnight costs for onshore wind will reduce by one-quarter and for solar PV by one-third, while energy storage costs will drop by half. However, Hörsch notes that even if the analysis were to be based on today's costs, the results would not change qualitatively.
Another deviation is that the assumptions have been modelled using the PyPSA software tool, rather than the sophisticated Plexos system used by Eskom and the DoE to produce South Africa’s IRP. However, Hörsch stresses that the models are mathematically identical.
The outcome is a least-cost configuration in 2040 comprising 53% wind, 16% solar PV and 25% coal, with the balance of the energy derived from nuclear, gas, hydro and batteries.
The system will still comprise 19 GW of residual coal-fired generation and 1.8 GW of nuclear in the mix, despite the closure of several coal plants between 2020 and 2040. There will also be 2.9 GW of pumped storage, while Cahora Bassa is expected to contribute 1.5 GW of capacity.
Increasingly, though, the variable wind and solar generators will need to be backed up by 11 GW battery units and 12 GW closed cycle gas turbines.
The least-cost mix has an average system cost of R510/MWh in 2040; a “base case” that is nearly 20% cheaper than a scenario where at least 10 GW apiece of new coal and new nuclear is incorporated into mix.
Hörsch adds that the integration of renewables generators is also relatively cheap and easy, owing to the fact that the expansion of the South African transmission system is less land-constrained than is the case in many other countries, particularly those in Europe.
“Only a moderate expansion of transmission corridors will be needed, comprising only 10%, or R10/MWh, to the yearly system cost in 2040,” he explains.
The results are more or less in line with a recently updated techno-economic study conducted by the CSIR Energy Centre, which pointed to a least-cost electricity mix, by 2050, in which solar PV and onshore wind contribute nearly 80% of the country’s electrical energy.
However, the FIAS study has a higher proportion of wind compared with solar PV than is the case with the CSIR study, which outlines almost even contributions from the two renewables technologies.
Hörsch attributes this deviation in results to the fact that the PyPSA analysis forecasts greater “smoothing” in the variability associated with wind generation as a result of the geographical distances between generators assumed in the model.
Secondly, the FIAS study has included higher costs for battery energy storage than is the case in the most recent CSIR analysis, released in November. Lower battery costs would result in a higher penetration of solar PV, with batteries materially reducing the variability associated with the technology and extending operating hours into periods when the sun is no longer shining.
The latest CSIR study, also includes a moderate amount of battery electric vehicles and flexibility arising from demand side management, which favours a higher penetration of solar PV.
Both studies concur, though, that the cheapest future mix will be renewables-led and that the introduction of additional new coal and new nuclear will only raise overall system costs.
“Overall, our investigation shows that renewable energy in South Africa is incredibly cheap and easily integrated into the system,” Hörsch states, adding that the main conclusion is that South Africa should immediately turn its attention to building a future mix based on solar PV and wind.