While the CANDU heavy-water plants that power half of Ontario have proven themselves safe, they're slow and expensive to build (and not cheap to operate, either). If Alberta started building one right now, it would be years before they would see any benefit. Hard to make that work in the current political climate.
Other plant types are, by my understanding, less safe, less proven, or continually five years in the future.
Politicians are punished if proactive and praised (and re-elected) if reactive.
Look at the Carbon tax now.
When the equator is a no-mans-land and tornadoes are the norm, Carbon tax will seem like a "why didn't you do it back then" idea.
We traded nuclear science R&D for a construction worker jobs program a long time ago. Way easier for politicians to sell constant refurbishment of our grandparents reactors than some smarty-pants physicists looking at a computer.
That's not true. The Westinghouse AP1000 got type approval in 2011. The EPR got type approval in the early 2000s. Both are GEN III+ reactors, which are semi-modular and have reduced length and number of pipe-runs and number of pumps, valves and so on. They've got 60 year design lives compared to 30 years for the typical Gen II design. It didn't stop them from being more expensive to build than the prior reactor types.
The EPR2 is currently undergoing certification. It would be a brave utility to roll the dice on a new, untested version of the EPR after the fiascos at Flamanville, Olkiluoto, Taishan and Hinckley.
SMRs to date have been one failure after another. NuScale just cancelled the Idaho project in spite of receiving $4B in government subsidies. X-Energy cancelled plans to go public and laid off 100 staff. Oklo’s Aurora reactor license application was so poor that it was rejected almost immediately by the NRC. Rolls Royce has announced that their £500B SMR program will run out of cash by the end of the year, and so on.
New cost estimates from TerraPower and XEnergy as part of the Department of Energy’s Advanced Reactor Deployment Program are likely to reveal substantially higher cost estimates for the deployment of those new reactor technologies. This would confirm other independent studies on SMR economics.
Four AP1000s have been built in China. The Chinese also have an agreement with Westinghouse to further develop the design, so more are under construction, but those are heavily modified Chinese variants.
Four AP1000s were under construction in the US, two at the Vogtle generating station in Georgia, and two at the V.C Summer plant in South Carolina. The V.C. Summer reactors were cancelled during construction when the initial estimate was revised from $9B to $23B. The Unit 3 at Vogtle has finally completed commissioning and is online, and unit 4 should be completed this year. Costs have exploded from $14B estimate at the beginning of construction to the vicinity of $35B.
There are another 5 reactors planned for Turkey and Poland. I'm not sure where they're at currently.
For the EPR, the first unit to start construction was Olkiluoto-3 in Finland in 2005. It was supposed to complete commissioning in 2010, but finally was completed in 2022 and entered service in 2023. Costs went from €3.3B to €11B
Taishan 1 & 2 started construction in 2009 and were supposed to be completed in 2013. Taishan-1 entered service in 2018, and Taishan-2 in 2019. Though the third EPR project to start construction, these were the first in service. The final $7.5B cost was roughly double the estimate. Since then, unit 1 was offline for a year due to issues with the fueling. There have been some other reliability issues, some causing brief downtime.
Flamanville-3 started construction in 2007, was supposed to be commissioned in 2012, but is currently projected to be in service late this year. Costs bloated from €3B to a projected €20B at completion.
Hinkley Point C started construction of two EPRs in 2017, though a lot of site prep work started well prior to that. It was supposed to be online in 2023. Currently they're projecting 2028. Costs have gone from the initial £16B to a projected £33B.
The average age of a French reactor is 37 years. They get an initial license for thirty years then apply for ten year extensions. They have 56 operational reactors now, and have an ongoing 'grand carénage' refurbishment for mostly the larger units. The estimate for that was in the vicinity of €55B, though has shifted some what. The smaller, older units are being taken out of service.
More often than not, older reactors in the US are taken out of service rather than refurbished due to the economics. Globally there are 407-413 (definitions vary) operational reactors, down from 438 at the peak in 2002.
A closely related problem is what to do with the station’s 2,500 employees. Mr. Attieh said keeping so many people on payroll at an idle plant for a decade or longer could be difficult to justify. But, he added, “They need to keep the staff to maintain the station in operating conditions, and they do not want to lose expertise in running the station.”