The National Nuclear Regulator ruled last year that Koeberg’s first reactor, Unit 1, could continue operations for another 20 years. The regulator will make a decision about the second reactor in early November. Yet Eskom’s application for the long-term operation of Unit 2 is an impressive mix of shady logic and grand leaps of faith.
The main issue is Unit 2’s containment structure: the concrete dome and cylindrical walls that you see from the street or across the bay. In the case of a serious accident — a full or partial meltdown — the containment structure is designed to prevent the release of radiation. So, no small matter. Unless, of course, you rather like leukaemia, thyroid cancer and Geiger counter tourism.
At this stage, after all the upgrades and repairs to get Koeberg ready for another 20 years of operation, you’d think both Eskom and the regulator would know what the actual status of Unit 2’s containment structure is. But they don’t know. They can’t know, because a vital monitoring system broke down and a crucial test keeps on being delayed. The regulator and Eskom are not only flying blind, they’ve cast aside both a critical regulatory code and a global convention on nuclear safety.
The containment monitoring system consists of invar wires, pendulums, strain gauges, thermocouples and dynamometers. These devices are supposed to provide a multidecadal data set, starting in 1985 when Unit 2 was commissioned, on how the containment structure changes over time. A multidecadal data set is necessary to assess the current health of the containment building and to project structural integrity into the future.
The total collapse of the monitoring system is extremely problematic. In its 2024 licence decision for Unit 1, the regulator castigated Eskom for a “complete absence of dome data for Unit 2”. Correspondence between Eskom and the regulator in 2024 exposed that there was only one working strain gauge in the dome.
In a July 9 2025 engineering report on Unit 2’s containment structure Eskom states that “degradation of the invar wires led to some invar wires breaking and the datum for the measurements were lost. As such, long-term monitoring results since commissioning are not available for the invar wires or not useful ... The results are therefore not considered as part of this report.”
Defective invar wire stations were first discovered in 2004.
In the same report, Eskom states that “the pendulums at Koeberg have degraded with time and the lifetime measurements and/or the documentation of the pendulum results had been poor and the long-term monitoring results since commissioning is not available or not useful.” While the dynamometers were refurbished in 2023, their previously erratic readings resulted in, as Eskom puts it, “gaps in the data”.
The International Atomic Energy Agency strongly recommended in 2022 that the monitoring system be restored to full functionality. In 2024, the agency repeated its recommendation.
Eskom told the regulator, in the aforementioned engineering report, that a pre-feasibility study into restoring the monitoring system’s functionality was under way. It describes the planned modification as “complex” and, for Unit 2, requires the installation of 34 strain gauges and a selection of associated thermocouples. Crack monitoring devices are needed at least four locations along the dome. Installation date is somewhere around 2029.
Business Day asked both Eskom and the regulator if the containment monitoring system was presently functional, a simple yes or no question. Neither has responded. Excessive secrecy within the nuclear industry globally has been, and remains, a terrifying curse.
Regardless of mysterious and perhaps mythical functionality, there isn’t a multidecadal data set for Unit 2 and that’s not going to change. So Eskom pulled a bit of a move and took data about Unit 1’s containment structure, extrapolated the data set to Unit 2, applied an “engineering judgment” and informed the regulator that containment integrity for Unit 2 was proven for the next 20 years.
This desperate approach is akin to a mechanic saying that since the brakes work on your 1985 Mercedes-Benz 380 at high speed, they will work on your other Merc of the same year and model. No need to look at the brakes on both cars, one car will do and you can clock the other in sublime confidence that you’ll be able make a sudden emergency stop.
However, there is actually a test that could provide some real data — an integrated leak rate test (ILRT). The test replicates the conditions of a partial or complete meltdown. A 2022 Eskom report discloses that engineers told management that an ILRT was required before the life extension, ideally in 2024. Eskom management first delayed the ILRT to this year and is now talking about some vague point in 2026.
Eskom and the National Nuclear Regulator follow the US Nuclear Regulatory Commission’s codes, and for a good reason. The first American commercial nuclear power station opened in 1957 and today there are 94 working reactors. In its application for Unit 2’s life extension Eskom cites regulatory code 1.90 as a normative document. The US code, excitingly titled “Inservice inspection of prestressed concrete containment structures With grouted tendons”, lays out the rules for a containment building.
Unit 2’s last ILRT was in October 2015 and the code says they should be carried out every 10 years. So, Eskom should be conducting an ILRT right about now. If the monitoring system becomes dysfunctional, Regulatory Code 1.90 explicitly states that an ILRT must happen every five years. A regulator committed to international standards would have forced Eskom to do an ILRT in 2020 and then again in 2025. How long does it take to conduct an ILRT? About two-and-a-half days.
Parliament ratified the Convention on Nuclear Safety in 1996 and the National Nuclear Regulator is the national representative body. On behalf of the country, the regulator made a written commitment at a 2022 meeting of the convention regarding the containment building, which is cracking, delaminating and corroding. The latter is most worrying. If corrosion spreads to the tendons — huge cables that are the backbone of the containment building — Koeberg would have to shut down permanently.
There is a fix though. Impressed cathodic current protection (ICCP) halts the rusting. Eskom has known since 2015 that Koeberg urgently requires ICCP. At the 2022 convention meeting, the NNR promised that ICCP would be installed no later than 2024, but then allowed Eskom to reschedule for 2026. Don’t hold your breath for next year though. A contractor hasn’t been sourced and the regulator still hasn’t approved the ICCP’s design. If we're lucky, the corrosion won’t spread.
The slackness, the delays and lack of regulatory enforcement all derive from a belief that Koeberg won’t have a major accident: “Hey dude, ICCP and an ILRT can be done, like, whenever. Chill.” Yet not taking the possibility of a runaway chain reaction seriously — the messianic belief of impossibility — was a major causal factor at Three Mile Island, Chernobyl and Fukushima.
While nuclear power is a contested topic, the point of agreement between the pro and anti-nuclear factions is that nuclear plants must be as safe as possible; everyone is on the same page. Everyone, that is, except the National Nuclear Regulator and Eskom.
• Taylor is a research fellow in environmental ethics at Stellenbosch University.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.