BusinessDay, Energy Research Centre, UCT, 12 December, 2017.
The economic benefits of a nuclear fleet are no better than a flexible build plan, even in a future where we assumed nuclear is cheap
Energy Minister David Mahlobo reportedly wants to finalise quickly the latest iteration of our electricity plan in support of new nuclear power. The minister claims that “there’s no discussion about the need, the need is there” for nuclear power.
Yet research that we have undertaken at the Energy Research Centre supports neither a need for, nor benefits of, forcing a large nuclear fleet into our electricity system.
Modelling of all available electricity generation options continues to show that nuclear power is not the least-cost solution. Nor does the country have the ability to finance the investments required for a 9.6GW fleet of large reactors. SA currently faces an excess of capacity and will not need this power in the short to medium term.
The latest modelling shows nuclear only coming into the mix around 2040. This is a finding consistent with earlier work the centre undertook for the National Planning Commission in 2013. Current research together with economic modellers also suggests a wait-and-see approach. The rush to complete the Integrated Resource Plan (IRP) and increase the share of nuclear is suboptimal for the electricity system and for the South African economy. There is no urgency about the decision around nuclear.
When would we need nuclear power? Nuclear plants take 10 years to build and will run for decades after, but it is virtually impossible to predict demand half a century into the future. Electricity demand projections have consistently been higher than actual growth, when evaluated ex post. Smaller nuclear reactors could in future track demand more closely than those being considered for the fleet.
In the past few years, electricity demand has flattened and is even declining. The global financial crisis reduced economic demand, which is a key driver assumed in modelling electricity demand. The period of load-shedding that followed in SA further kept electricity demand low. So SA has time to carefully consider future investment needs as no new generation is needed before the late 2020s.
Does SA “need” nuclear when it is not the lowest-cost option for the country? Good policy should be informed by sound evidence. Here’s an explanation on how we cost nuclear power, in research terms.
Much of the public debate centres on “overnight capital costs”, which are the costs of construction, excluding inflation or interest. There are divergent figures on the “overnight costs” of nuclear, dependent on certain assumptions, technology choices and country of construction.
The IRP 2013 used a range of about $5,000/kW–$7,000/kW. This range was found to be consistent with literature for the types of plants SA would be considering, and was used in studies by the centre on nuclear power and bounding uncertainty, including those on costs. A more recent review by three research groups of overnight costs suggests that the upper range could be as high as $8,500/kW.
The “overnight cost” is not a very good basis for comparing the costs of electricity plant since it excludes other key components — fuel and operating costs, aggregate availability, lifetime, interest during construction, borrowing rates, system integration aspects and risk. Another measure of cost is the “levelised cost of energy”. This cost is expressed in cents per kWh, and takes into account the overnight costs and the other aspects listed above except for the system integration aspects. Risk is taken into account to a certain extent through the discount rate, but this does not fully account for the risk of over-build.
In SA, renewable energy prices have fallen rapidly, echoing global cost reduction trends. Actual average tariffs from solar photovoltaic (PV) and wind electricity generation decreased from R3.65/kWh and R1.51/kWh in 2011 respectively to R0.62/kWh in 2015, making it cheaper than electricity produced from a new-build coal-fired power plant (R1.03/kWh) as well as nuclear (R1.09/kWh), the latter figures being those published by the Department of Energy in 2016.
The measure of levelised costs can be useful for comparing the overall observed and expected energy cost from different technologies, but can be misleading when comparing technologies with very different characteristics. For example, non-dispatchable solar PV and wind do not provide the same value to the system as dispatchable generators. The actual value (and costs) to the energy system of any technology is a complex and dynamic combination of all prospective new and existing capacity and their overall ability to meet demand. Both demand and supply options change over time — over a day, week, month, year — as the structure of the overall power system evolves.
It is important to the operation of the system when supply and demand-side options produce and whether this is at the same time as demand. To fully understand the implications of the advances in energy technologies on future electricity generation in SA, a fully integrated energy systems assessment is required. An energy system model is also useful to compare different scenarios.
Our research has compared the economic effects of a nuclear fleet against a flexible, least-cost build plan. We found that the economic benefits of a nuclear fleet are no better than a flexible build plan, even in a future where we assumed nuclear is cheap. Given that the result depends on many inputs, the centre’s researchers further analysed many variants of these two scenarios and found that nuclear is not the least-cost option. A forced nuclear scenario results in electricity prices that are higher and this “would have negative impacts on growth, employment and welfare in SA”. In plain language, one has to cherry-pick a future in which nuclear power is affordable.
In a world where there is uncertainty about future demand, future technology costs and capabilities, future grids with distributed generation and storage, committing ourselves to a large investment far in advance is not prudent.
So nuclear power is not the most affordable option, by overnight costs, levelised costs or by running an energy system model. But there are factors other than cost to consider. SA would do well to invest in technologies that deliver what we really need, especially employment.
The localisation and respective job-creation potential of a nuclear fleet is low compared with other technologies, as most of the local jobs will be temporary construction jobs and a couple of thousand permanent jobs in operations and maintenance, depending on the number of nuclear plants being built. Pushing up the local content requirements for the nuclear programme is another way of increasing the cost to levels even unknown to the industry.
Over- and under-supply are both costly to the economy, and we have a poor track record in avoiding either. The “fleet” approach taken to nuclear in IRP 2010 makes the investment particularly large. A 9.6GW fleet has been estimated to cost between R322bn and R1.4-trillion. These estimates do not include cost overruns, which are common on mega-projects. Many studies do not include interest during construction, which due to long lead times of nuclear and depending on interest rates, can increase the capital cost of projects by 40%-50%.
The government is already committed to providing a R350bn debt guarantee to Eskom, and we have an unaffordable debt-to-GDP ratio (currently at 51.7%). Another R1.4-trillion in guarantees or sovereign debt would more than double our national debt, which is currently about R870bn. The Treasury is seeking to reduce debt to keep the interest paid on our national debt under control. Increasing that debt in the current economic climate seems unwise.
• Caetano, Merven and Winkler work at the University of Cape Town’s Energy Research Centre. They write in their personal capacities.