The six new nuclear reactors envisaged in South Africa’s energy road map are likely to cost between R322-billion and R712-billion to build and finance, but the total could spiral quickly into trillion-rand territory if they are not built to deadline and budget, according to experts.
Three independent energy specialists participated in a costing exercise facilitated by the Mail & Guardian to clarify the economic implications of the 9 600MW of new nuclear capacity proposed in the Integrated Resource Plan 2010 (IRP2010).
After the M&G pegged the value of a yet-to-be-issued tender at R1-trillion in October last year, Energy Minister Dipuo Peters said “it would be about that or more” but later backtracked and came up with a “cost estimate” of R400-billion. The Democratic Alliance accused her in Parliament of a “thumb suck”.
In fairness to Peters, the experts arrived at widely divergent figures.
Philip Lloyd (R322-billion to R380-billion)
Philip Lloyd, professor at the Cape Peninsula University of Technology’s Energy Institute, based his calculations on estimates provided by the United States-based Electric Power Research Institute, commissioned to do the maths for the integrated resource plan.
The institute provided two cost summaries — one for French pressurised water reactor technology, known as the EPR, and the other for American technology of the same kind, known as the AP1000.
The generating capacity of an EPR reactor is 1 600MW and the AP1000’s is 1 117MW. Six French reactors or eight American ones would therefore meet the 9 600MW target.
Lloyd increased the research institute’s figure for the overnight costs of French and American nuclear technology by 5% to cater for technological adjustments to reactor safety features following the meltdown of the Fukushima reactor in Japan last year.
The overnight cost is a way of calculating the material and labour costs in advance. It removes the uncertainty of variable costs from the equation and calculates material and labour costs as if they have all been paid for simultaneously.
Lloyd used the research institute’s estimates to calculate the cost of financing construction (at 8.6% interest), the construction time (six years) and a cumulative interest formula. Paying for the new units would be done “slowly and out of revenues”.
He said the advantage of building six reactors was that the cost per reactor decreased as the engineering teams improved their skills, purchasing systems and project control procedures.
He warned that South Africa would also need to spend “about R250-billion” to upgrade its electricity distribution network.
Stephen Thomas (R636-billion)
Stephen Thomas, a professor at the University of Greenwich’s business school, dismissed the research institutes cost estimates, saying “it had never built a nuclear power station”.
He also discounted potential bids from China and Korea, because they had yet to demonstrate the cost at which they could export nuclear reactor technology.
He based his ‘very crude calculations” on the ‘real bids” submitted by Areva and Westinghouse in 2008, before Eskom cancelled a previous tender for similar nuclear capacity. “Reactors haven’t got cheaper since then,” he said.
The overnight cost was reportedly $6 000 per kilowatt, according to Nucleonics Week magazine — 62% more than Lloyd’s research institute baseline.
The cost of financing would depend entirely on the terms of contract to sell the plant’s output. “If there was a strong assurance that consumers would pay whatever it took and the South African government explicitly guaranteed the debts, the cost might be low,” he said.
Thomas said the research institute’s 8.6% interest rate was “too low”, although it had adopted this figure after consulting the South African government.
He said that, typically, capital was funded partly by borrowing from the market at a minimum rate of 10% and partly by raising equity — asking shareholders to delay their dividends for a higher return later, usually 15%.
Split 50-50 between borrowing costs and equity costs, Thomas calculated the interest at 12.5%. He admitted, however, that “you can’t say in advance what the interest rate will be — banks will look at the specifics of the contract and decide”.
He applied 12.5% interest to the annual capital expenditure over a six-year construction period to arrive at a figure of R106-billion per EPR reactor. Using this formula, the cost of financing was 46% of the overnight construction cost.
“People fret about construction costs but they totally underestimate the additional cost of financing,” Thomas warned.
But Lloyd criticised Thomas’s interest rate calculations for being “ridiculously high”, saying that 30-year bond rates were currently about 3.4% in the United States. “Bond rates give a good indication of the rates at which you can borrow long term. So even the research institute’s 8.6% is high,” Lloyd said.
‘Flamanville scenario’ (R1.4-trillion)
Thomas also proposed a “Flamanville scenario”, which calculates the cost of building and financing a project over nine years instead of six and at double the amount initially budgeted for.
It is based on French utility Électricité de France’s problems at Flamanville in northwestern France, where the EPR it is building has hit these time and cost overruns. The only other EPR under construction in the West, Olkiluoto in Finland, is in almost identical trouble.
Thomas said the problem with the EPR was that its design had not been finalised.
In the Flamanville scenario, Thomas envisaged finance costs ballooning to 65.6% of the overnight construction cost and the total spend on six reactors exceeding R1.4-trillion.
“If something goes wrong and you have to start borrowing money on the open market to fund the overruns, it’s going to get very expensive very quickly,” he warned.
But Lloyd said that Flamanville and Olkiluoto were “a disaster from which we’ve all learned”.
He said that the Chinese experience of building EPR technology at Taishan on China’s south coast “has shown they can be built on time and within budget”.
But Thomas said that both Taishan EPRs were still three to four years from completion. “A lot can still go wrong in that time.”
Rod Gurzynski (R599-billion to R712-billion)
Rod Gurzynski, an energy and economics researcher who has studied nuclear energy costs for the Civil Society Energy Caucus and the World Wide Fund for Nature South Africa, said the energy department recognised that the research institute cost estimates were too low and proposed “a possible increase of 40%” in the final gazetted version of the resource plan.
Working from a research institute-plus-40% baseline, and factoring in what he called “owner’s cost” to the overnight construction price, Gurzynski arrived at a significantly higher estimate than Lloyd’s.
Putting his calculations into perspective, Gurzynski said that R600-billion to R700-billion was “close to South Africa’s entire annual tax revenue”. In his budget speech last month, Finance Minister Pravin Gordhan estimated 2011-2012 tax revenues at R739-billion.
“Constructing six [or eight] plants simultaneously would put considerable strain on the country’s resources with no return for 10 to 15 years [the estimated construction period for six reactors],” Gurzynski said.
Operating costs
Once the reactors are commissioned, they begin to pay back their construction and financing costs, although other costs then enter the picture. Lloyd calculated that it would cost R870-million to fuel each EPR -reactor annually, or R620-million for the AP1000.
He also calculated the operation and maintenance costs at R1.14-billion for the AP1000 and R1.23-billion per EPR reactor per year. The combined fuel and operation and maintenance costs for six EPR reactors would therefore be about R12.6-billion a year for 9 600MW of EPR power and R14.1-billion a year for the same AP1000 power.
‘Levelised’ costs
The “levelised” cost for nuclear power is the sum of the construction, finance, fuel and operational and maintenance costs over the reactor’s estimated 60-year lifetime, divided by the estimated output in kWh over that period. In short, it is the cost per unit of the estimated electricity generated.
Lloyd’s levelised cost for nuclear is between R0.70 to R0.81/kWh; and Gurzynski’s is between R1.17 and R1.41/kWh.
According to the research insitute’s figures, on which Lloyd based his calculations, nuclear costs compare favourably with wind technologies (which range between R0.66 and R1.05/kWh) and solar (between R1.49 and R4.01/kWh). Coal-derived electricity is cheaper than nuclear, at between R0.59 and R0.74/kWh.
But Gurzynski said that the cost of renewable energy was likely to come down considerably over the next 10 to 15 years while the nuclear plants were being built. He also said that the research institute’s figures did not include an average mark-up of 40% for transmission and distribution from the plants to the end-users.
Thomas also queried the research institute’s assumption that a nuclear reactor would run at 90% reliability over its lifetime.
“Long-term, only a handful of the 400-plus reactors in the world have averaged a 90% load factor.”
Koeberg’s two reactors have a lifetime load factor of 68.6% and 69.7%, respectively.
Reliability is an important factor: the less reliable a reactor, the higher the cost per unit of electricity it produces.
End-of-the-road costs
The estimated cost of decommissioning a nuclear reactor and disposing of nuclear waste was “a huge guess”, Thomas said.
“Less than a handful of reactors have been decommissioned worldwide. We’re decades away from starting to build the first disposal site for spent fuel so we haven’t a clue what it would cost,” he said.
About $125-million invested today at 3% interest over 100 years would meet a future $2.5-billion decommissioning and waste disposal liability, but Thomas warned that there were no guarantees.
“If this funding fails, the job still has to be done and a future generation of South African taxpayers will have to find $2.5-billion to clean up our mess.”
Experts’ detailed cost estimates