By Professor Emeritus Steve Thomas
This article was published in Business Day Live
Utility cannot square the circle of financing, timing and electricity cost without being someone’s guinea pig
After the latest failure in September of the South African government to issue a request for proposal for new nuclear power plants, Energy Minister Tina Joemat-Pettersson confirmed that Eskom rather than the Department of Energy would procure future nuclear power plants.
Eskom’s group executive for generation, Matshela Koko, has revealed Eskom’s nuclear strategy. This plan is similar to the call for tenders carried out by Eskom in 2008 that ended with the supplier being chosen only to find that the deal could not be financed. So what are the chances that this plan will be more successful than that of 2008?
Eskom expects to buy two reactors with a total capacity of 2,400MW-3,200MW, depending on the output of the design chosen, to be on-line by 2026. It claims it will be able to provide some of the finance required from its balance sheet and it is looking to buy a “standardised design” with no “leading-edge technology”. Koko has stated that the target price was R1/kWh.
Since finance was the sticking point in 2008, it is useful to start there. Eskom may be able to provide some of the finance, but the bulk will have to covered by borrowing. Despite the brave words about Eskom returning to profitability once the long-delayed coal-fired plants are in operation, it is far from clear how much cash from its profits Eskom will have to contribute.
Its credit rating is close to or at subinvestment quality (junk), so it is not credible for it to borrow the money without sovereign loan guarantees. These ensure that if the project goes wrong and the owner cannot repay the loan, taxpayers from the country giving the guarantee will repay the bank. In 2008, the South African government was unwilling to provide such guarantees, leading to the collapse of the tender. Now SA’s credit rating is dangerously close to junk, so even if the government was now prepared to give such guarantees, they might not be credible with banks and financiers. This means that the government of the country hosting the company supplying the reactor would have to support the reactor vendor with loan guarantees or, better still, lend the money itself. This significantly limits the choice of vendor.
While Koko’s claim that “Eskom does not want to be anybody’s guinea pig for new ideas” sound sensible, with today’s market for reactors, it is not clear how this can be avoided. Some have construed his statements as implying Eskom is looking to buy 1980s technology, which would be much cheaper. Such reactors would, for example, not be designed to withstand an impact from a large jet aircraft and would not include a “core-catcher” (something to prevent the core being released into the ground in the event of a core-melt).
Even if we assume this is a sensible strategy, there are at least two problems with it. First, as Koko acknowledges, decisions on what is safe enough are not for Eskom or the government to take. They are the job of the National Nuclear Regulator, a fully independent body. Second, no vendor is offering to sell such technology.
Koko lists the main designs on offer, all of which he claims could satisfy the regulator. This includes the designs tendered in 2008, the French Areva EPR and the Toshiba/ Westinghouse AP1000, plus the Korean APR1400, the Rosatom AES-2006 (Russia), the SNPTC CAP1400 (China) and the Mitsubishi APWR.
If by leading-edge technology Koko means Eskom is looking for designs based on at least some construction experience, that rules out the CAP1400 and the APWR. So while China is the most likely of the vendor governments to be able to offer finance, choosing their design would put SA in the role of guinea pig.
If operating experience is needed, that rules out all, bar the APR1400. The four EPRs under construction are three to nine years late and well over budget. Areva is essentially bankrupt and it is not clear whether attempts by the French government to save it will be successful. Loan guarantees from the French government do not seem likely. The eight AP1000s under construction are doing little better than the EPRs. The vendor, Westinghouse, is based in the US but owned by Toshiba of Japan. Japan has never given loan guarantees for nuclear projects, while the US government would face strong public opposition to risking US citizens’ money for the benefit of Japanese shareholders. Toshiba is doing little better financially than Areva and has been fined heavily for falsifying its accounts for the past seven years.
In the past, there was a lot of publicity suggesting that the nuclear tender was a done deal for Russia, not least because Russia could supply the finance. There are six AES-2006s under construction, all late by three to four years. The first started generating in August but is still in the testing phase. Russia has a huge order book of more than 30 firm reactor export orders that are yet to start construction, all requiring Russian finance and all ahead of SA in the queue for finance. Given the weak state of the Russian economy, particularly its banks and investment funds, there seems little chance that Russia could supply finance.
The first APR1400 only went into service in Korea in January, with three more under construction in Korea and four in the United Arab Emirates (UAE). The orders for the UAE were won in a call for tenders in which Kepco bid significantly less than its competitors. This is seen as a demonstration that reactors can be bought cheaply. However, Kepco acknowledges that this design does not meet European standards and is upgrading it to comply. How much cost advantage the APR1400 would still have with features such as aircraft protection and a core-catcher is a moot point. There were also suspicions that because this was Korea’s first (and only) reactor export order, the price for the UAE was a loss-leader.
This leaves the issue of whether the South African reactors can be on-line by 2026 and whether the cost target can be met. If past experience is a guide, setting up and running the tender is likely to take about two years. The detailed design would then have to be reviewed by the regulator, a process of which it has minimal experience and that in more experienced countries takes at least two years. While this is going on, the site has to be identified and all planning and safety approvals given. The long lead-time components must then be ordered, and site preparations made before construction starts. Realistically, construction will take at best six years but perhaps a lot longer. Even if nothing goes wrong, the 2026 target looks implausible.
As to the cost target, there is a long history of Eskom and the South African government basing their proposed programmes on hopelessly overoptimistic costs, apparently assuming that everyone else is paying far more than they need to for their reactors.
The most realistic guide is the price agreed in the UK for its new reactors of about 10p/kWh in today’s money. But exchange rates are unstable, a serious risk for a South African reactor programme, and this equates to about R1.80-R2.20/kWh. It may be that UK has paid over the odds in this deal, but given that Koko’s desire for proven technology doesn’t appear achievable, is it really plausible that SA can get its nuclear power at half price?
The likely outcome is that, at best, the tender will fail again and another two to three years will be wasted while options that would work are neglected.
• Thomas is emeritus professor of energy policy at the University of Greenwich in London.