I find this a weirdly emotional subject for me, which it shouldn't be, heck i don't even work in the industry! but our very Canadian CANDU's are something I hold dear and so to make some sense of this I need to lay out WHY this is such an issue, despite the fact I doubt the government is going to do anything to change the course.
Back in the early 1960's after numerous experiments at Chalk River, the first commercial CANDU, built explicitly for the purpose of power generation, was constructed at what is now known as the Bruce site, home to the world's largest nuclear power plant at 6,420MWe. This single unit, dubbed Douglas Point, was a small (220MWe) proof of concept unit that operated for ~20 years before being shut down. The success of this unit set the stage for the construction of the first major large scale nuclear power plant in Canada: Pickering.
At the time, in the early 1960's, Ontario Hydro was looking to build another large (~4,000MW) coal plant. They were approached by AECL (Atomic Energy Canada Limited, a Crown corporation, wholly owned and operated by the Federal government) with respect to building a nuclear plant instead. The promise was that the nuke would be cheaper to operate and due to the ridiculously low cost of fuelling it (natural uranium), that it would pay for itself within 15 years. This was the deal struck with the government of Ontario at the time and the commitment to build Pickering was on, based on some pretty heavy documentation proving AECL's claims and funding split between the entities. Built adjacent to the largest consumer of electricity in Ontario, the GTA, Pickering's purpose was to directly feed the sprawling metro Toronto area at close proximity.
Pickering, like all CANDU's post Douglas Point in Ontario, was built as a pair of 4-packs. Construction started in 1966 and by 1971 the first A unit was online. All 4x units were grid connected by 1973. Due to the roaring success of this project, as it was coming online, construction of its much larger sibling, Bruce, commenced adjacent to the Douglas Point site. Pickering B units started construction shortly after all the A units were online and were themselves grid connected started in 1983.
Early in the 1980's, there were some issues discovered in Pickering A, which had now been producing power for a bit over 10 years. The annulus spacers, which keep the pressure tubes separate from the calandria tubes were migrating to the ends due to flow and allowing the pressure tubes to sag. This resulted in some localized hardening and resulted in a pressure tube failure in I believe 1983. This resulted in a full re-tube of all Pickering A units and the annulus spacers were revised to provide a tension-fit and became an inspection item. The pressure tube design (materials) was also revised (improved).
Because Pickering was very much an FOAK design, there were many lessons learned from it. One annoyance with Pickering is that during the summer months, due to its shore-level inlet and outlet configuration, it tends to breed algae, which can accumulate and actually result in a shutdown until it is cleared up. Typically, it is only a maximum of 4 units that are impacted by this (either A or B, never both), but this is reasonably rare, and does not happen every summer.
CANDU's are designed to have a mid-life service at what was originally intended to be around the 30 year mark, the acronym for which varies depending on who is running the site. OPG calls it a refurbishment, Bruce Power calls it an MCR (Major Component Replacement). While this was always part of the design criteria, actually executing it was a whole other matter. The first major refurbishment was performed on Pickering A units 1 and 4. This resulted in all the pressure tubes, end fittings..etc being replaced, but due to political volatility at the time, the two other A units, 2 and 3, were NOT refurbished and were instead shuttered. This resulted in the conversion of Pickering into a 6-unit plant, which fortunately, could be facilitated because Pickering A and B share a vacuum building and their turbine halls are directly beside each other. Such a feat would not be possible at Bruce, where there is significant physical distance between the A and B plants.
O/T: "Mid-life" in this context is also a bit of a misnomer, as technically, there's nothing preventing you from doing another MCR/refurbishment 35-40 years down the road, but that's not overly relevant to this rant.
The next refurbishment was Bruce units 1 and 2 at the Bruce facility, whose operation had been outsourced to private operator Bruce Power. The entire Bruce A site had been laid up after these two units had been damaged by contractors and when Bruce Power took over the site, only the 4x B units were operational. This (the refurbishment) was a learning experience for them, it went slightly over budget, but the results have been positive and the 2x refurbished units have exhibited an operating record as good or better than the units pre-refurb. The other two A-units were returned to service with some slight tweaks and all A units are producing significantly more than rated power.
There are different ways to approach the refurbishment process, depending on the intentions of the operator. At Darlington, Unit 2, in addition to pressure tubes, end fittings and the like, it and its siblings get new control systems, new turbines, new generators...etc. At Bruce, the intention is not only new equipment, but to extend the lifespan of existing equipment in the interim (maximize ROI, which they call Life Extension) and more importantly, increase output (uprates) which in turn, makes the site more profitable.
Pickering B, now 37 years old, has never had its mid-life refurbishment. A decision that was made by the previous administration, to retire the site by 2024. Though that's been slightly extended by the current administration, and may be extended further, depending on how things go with the CNSC, things are, at present, expected to start winding down as early as 2023.
The reason, despite the refurbishment of the two A units, appears to centre around the lower output, and thus longer payback period, of the 380 fuel channel Pickering units at ~515MWe (NET) nameplate, versus the 480 fuel channel 800+MWe of the Bruce A units, or 880MWe Darlington ones. However, a recent maintenance outage at Pickering B resulted in A5 coming back at ~545MWe, which points to there being uprate potential, which would help with the payback.
Over the last decade or so, operation procedures and maintenance practices have improved dramatically. This has resulted in all sites having shorter maintenance outages, higher outputs, and generally better operating records. Pickering in particular, is now operating better than it ever has. It was sold on a 60% capacity factor to break even, but in 2019, it produced 26.4TWh, putting it at 88% capacity factor, the best of its operating life, not bad for units using pressure tubes that are up to 7 years past their original intended end of life
Right now, Ontario has enviably low emissions. Today, 88% of Ontario's electricity was being provided by the three nuke plants, despite us being down to 13 units, because spring and fall are when maintenance outages are performed so that we have all hands on deck for the summer peaks. Ontario had the lowest emissions in the world today at 20gCO2/kWh, thanks to our nuke fleet. When Pickering is gone, emissions will increase dramatically and in a time where climate policy is a major subset of any political platform, shuttering 3,100MW of ultra-low emissions power because there's some high CAPEX that needs to be eaten to keep it going for the next 40 years isn't going to be looked upon favourably once emissions spike when it is gone.
Pickering deserves its mid-life refurbishment. Its lower density fuel channel configuration, when compared to the other two sites, is the reason it has been able, and allowed, to operate into its 40's on the original ones. The new pressure tubes are an improved design using better materials, to get 50 years out of them isn't out of the question. We've got Bruce until at least 2064, we could have Pickering that long too if we were willing to put forward the CAPEX to facilitate it, OPEX is, and has always been a wash, the CANDU' are cheap to run, it's the construction and refurb's that are expensive. The A units could be replaced with SMR's once they are EOL, to maintain the viability of the site, something that should be well sorted by virtue of the SMR Roadmap by the time that rolls around.