Ontario’s Darlington nuclear station is on time, on budget and learning from past mistakes

Deep in the books of the Ontario government, there’s an entity called the Ontario Electricity Financial Corporation. It’s little more than a parking spot for roughly $15-billion in debts left over from the 1998 restructuring of Ontario Hydro, formerly the province’s electricity monopoly. It’s a reminder of the legacy of cost overruns during the construction of the Darlington Nuclear Generating Station in Clarington, Ont. More than a generation later, slightly less than half of Ontario Hydro’s debts remain unpaid.

Built in the 1980s and early 1990s, Darlington’s four 880-megawatt reactors supply roughly one-fifth of Ontario’s electricity. They’re in the midst of a $12.8-billion refurbishment, now more than half-complete. In a nutshell, that involves replacing crucial components such as pressure tubes, which have reached the end of their service lives, so that the station can remain in service for another 30 years. It’s the biggest project that Ontario Power Generation, or OPG (a product of the Ontario Hydro breakup, it generates most of the province’s electricity), has ever undertaken.

If you weren’t aware of this megaproject – or had forgotten about it – that’s probably because it’s proceeding very differently from the station’s original construction. “Overall, the four-unit project is on track to be complete on time and on budget,” Ontario’s Ministry of Energy reported earlier this year.

It’s a bland statement, yet one seldom used when referring to nuclear projects. Many go so wrong, in fact, that they’re never completed. Breaking that pattern at Darlington Station stemmed from OPG’s ability to learn from the past – not only its own but also that of other utilities.

Inside a $12.8-billion overhaul

CANDU reactors were designed to run for roughly 30 years. Darlington Station is more than halfway through a major overhaul to refurbish all four reactors by 2026. With reactors offline, other major components such as steam generators and turbines are also being rehabilitated.

Here’s an overview of the process:

Plant schematic

Electrical

grid

Shielding

Steam

turbine

Control room

Transformer

Reactor

Used fuel

Calandria

Condenser

Pressure tubes

Condenser cooling water

Detailed view of pressure tube and related components

Feeder

pipe

Bellows

Fuel channel

Pressure tube

Heavy water

Fuel bundles

End fitting

2: Heavy water removal:

CANDUs use heavy water

(which contains deuterium, a

heavier isotope of hydrogen

than that found in "normal"

water) to sustain nuclear

reactions. It must be drained

from the calandria, stored

and purified for re-use.

1: Isolate and defuel: The

reactor must be sequestered

from the rest of the plant, with

work areas walled off and

much equipment disconnect

ed. Next, irradiated fuel

assemblies (which are the size

of fire logs and contain small

uranium pellets that fuel the

reactor) must be removed.

4. Reassembly and recom

missioning: New calandria

and pressure tubes are

installed, followed by end

fittings and feeder pipes.

Finally new fuel bundles are

loaded and heavy water

returned. Station personnel

take charge of the reactor

and test it for weeks at low

power before returning it

3. Disassembly: Next, pipes that

move water to and from the

reactor (known as feeder pipes)

are cut away and the 480

pressure tubes and calandria

tubes are removed. This radioac-

tive waste is stored in a special

building. With its guts removed,

the reactor's remaining compo-

nents are inspected to ensure

fitness for continued service.

Note: diagrams are not to scale.

matthew mcclearn, john sopinski and murat yükselir/

the globe and mail, Source Canadian Nuclear

Association, OPG

Inside a $12.8-billion overhaul

CANDU reactors were designed to run for roughly 30 years. Darlington Station is more than halfway through a major overhaul to refurbish all four reactors by 2026. With reactors offline, other major components such as steam generators and turbines are also being rehabilitated.

Here’s an overview of the process:

Plant schematic

Electrical

grid

Shielding

Steam

turbine

Control room

Transformer

Reactor

Used fuel

Calandria

Condenser

Pressure tubes

Condenser cooling water

Detailed view of pressure tube and related components

Feeder

pipe

Bellows

Fuel channel

Pressure tube

Heavy water

Fuel bundles

End fitting

2: Heavy water removal:

CANDUs use heavy water

(which contains deuterium, a

heavier isotope of hydrogen

than that found in "normal"

water) to sustain nuclear

reactions. It must be drained

from the calandria, stored

and purified for re-use.

1: Isolate and defuel: The

reactor must be sequestered

from the rest of the plant, with

work areas walled off and

much equipment disconnect

ed. Next, irradiated fuel

assemblies (which are the size

of fire logs and contain small

uranium pellets that fuel the

reactor) must be removed.

4. Reassembly and recom

missioning: New calandria

and pressure tubes are

installed, followed by end

fittings and feeder pipes.

Finally new fuel bundles are

loaded and heavy water

returned. Station personnel

take charge of the reactor

and test it for weeks at low

power before returning it

3. Disassembly: Next, pipes that

move water to and from the

reactor (known as feeder pipes)

are cut away and the 480

pressure tubes and calandria

tubes are removed. This radioac-

tive waste is stored in a special

building. With its guts removed,

the reactor's remaining compo-

nents are inspected to ensure

fitness for continued service.

Note: diagrams are not to scale.

matthew mcclearn, john sopinski and murat yükselir/

the globe and mail, Source Canadian Nuclear

Association, OPG

Inside a $12.8-billion overhaul

CANDU reactors were designed to run for roughly 30 years. Darlington Station is more than halfway

through a major overhaul to refurbish all four reactors by 2026. With reactors offline, other major com-

ponents such as steam generators and turbines are also being rehabilitated.

Here’s an overview of the process:

Plant schematic

Electrical grid

Shielding

Steam

generator

boiler

Steam

Control room

Water

Steam

turbine

Electrical

generator

Transformer

Reactor

Feeder

pipes

Used fuel

management

Uranium

fuel

Calandria

Condenser

Pressure tubes

Condenser cooling water

Detailed view of pressure tube and related components

Bellows

Feeder

pipe

Fuel channel

Pressure tube

Heavy water

Fuel bundles

End fitting

Note: diagrams are not to scale.

1: Isolate and defuel:

The reactor must be

sequestered from the

rest of the plant, with

work areas walled off

and much equipment

disconnected. Next,

irradiated fuel assem-

blies (which are the size

of fire logs and contain-

small uranium pellets

that fuel the reactor)

must be removed.

2: Heavy water remov

al: CANDUs use heavy

water (which contains

deuterium, a heavier

isotope of hydrogen

than that found in

"normal" water) to

sustain nuclear reac-

tions. It must be

drained from the calan-

dria, stored and puri-

fied for re-use.

3. Disassembly: Next,

pipes that move water to

and from the reactor

(known as feeder pipes)

are cut away and the 480

pressure tubes and caland-

ria tubes are removed. This

radioactive waste is stored

in a special building. With

its guts removed, the

reactor's remaining com-

ponents are inspected to

ensure fitness for contin-

ued service.

4. Reassembly and

recommissioning: New

calandria and pressure

tubes are installed,

followed by end fittings

and feeder pipes. Finally

new fuel bundles are

loaded and heavy water

returned. Station person-

nel take charge of the

reactor and test it for

weeks at low power

before returning it

to service.

matthew mcclearn, john sopinski and murat yükselir/the globe and mail, Source Canadian Nuclear

Association, OPG

Milestones have been knocked off one by one. The first refurbished reactor (Unit 2) was restarted in 2020, four years after the project’s official commencement. On Friday, OPG received regulatory approval to resume normal operations of Unit 3, poising the utility to restart it months ahead of schedule. Disassembly of Unit 1 is complete and rebuilding recently commenced. Work on the final unit is expected to begin soon.

Continued success at Darlington lends confidence to OPG’s ability to manage risky, complex projects at a crucial moment. Last week, Ontario announced that OPG is to begin planning for three small modular reactors (SMRs) at Darlington. (It had already greenlighted one SMR, which OPG promises to build by 2028.)

The government also instructed Bruce Power, Ontario’s other nuclear power utility, to begin planning for new large reactors at the Bruce Station, raising the possibility that OPG, too, might have a shot at building a large new nuclear plant.

Meanwhile, the province awaits a feasibility study from OPG examining the possible refurbishment of four additional reactors at Pickering Nuclear Generating Station, east of Toronto.

Although considerably less risky than building new reactors, refurbishment is nonetheless an enormously complex undertaking. In the decade before Darlington’s commenced, OPG first had to build new offices to house its project staff, plus a 65,000-square-foot training facility and a building to store radioactive guts from four reactors.

Long timelines expose projects to unlikely but consequential events such as financial crises or pandemics, as well as ever-shifting political winds. OPG also must contend with the retirements of a significant contingent of its staff and leadership – not just at OPG but also its numerous contractors. And with Bruce Power simultaneously refurbishing its own reactors, necessary tradespeople are scarce.

WATCH: INSIDE THE CONTROL ROOM

On the day The Globe visited the station in late May, tradespeople were inspecting bellows (spring-like components which attach to either end of the reactor’s six-metre-long pressure tubes) on the reactor’s west face, hunting for punctures, dents and debris. On a platform above, more workers installed feeder lines, which carry water to the pressure tubes to cool the reactor’s uranium fuel.

Much of the work unfolds in a radioactive environment. Inside the Refurbishment Control Centre at the station, supervisors read detailed instructions into their headsets. On LCD screens throughout the room, they monitor tradespeople toiling away inside the bowels of a gutted nuclear reactor, clad in white and yellow plastic suits and respirators.

“On this reactor, we finished the full dismantling of the core,” explained Justin Alizadeh, head of the Canatom Power Group, a joint venture between SNC-Lavalin Nuclear and Aecon Construction, which is executing the work inside the reactor.

“We’re in this month-and-a-half phase right now of checking all of the parts of the reactor we’re going to reuse, before we start installing new components.”

Open this photo in gallery:

Subo Sinnathamby, OPG’s senior vice-president of nuclear refurbishment, is the project’s latest steward, but she’s not new to the project; prior to her appointment in 2020, she supervised Unit 2′s defuelling. OPG derived thousands of lessons from that first refurbished reactor that it applied to subsequent units, she said – but simply recognizing those lessons is insufficient.

“A lot of people learn a lot” during complex projects, she said. “But if you don’t implement them, you’re not going to see the benefits.”

Nuclear Disasters

With apologies to Tolstoy, one might say that all happy nuclear projects are alike; each unhappy one is unhappy in its own way.

Bent Flyvbjerg’s 2023 book How Big Things Get Done cites a database of 16,000 projects of all kinds, including notorious budget-busters such as bridges, tunnels, railways and IT projects. “Nuclear power plants are one of the worst-performing project types in my database,” Mr. Flyvbjerg wrote, “with an average cost overrun of 120 per cent in real terms and schedules running 65 per cent longer than planned.” The most egregious examples overshot their budgets by more than 500 per cent.

In its latest World Nuclear Industry Status Report, Mycle Schneider Consulting found that half of all reactors currently under construction globally have suffered delays. Some examples are extreme: Construction of two reactors in Slovakia, Mochovce-3 and -4, began 37 years ago. Iran’s Bushehr-2′s commenced in 1976.

This record represents a considerable competitive disadvantage as the nuclear industry jockeys for position in a widely anticipated energy transition. While hydroelectric projects are similarly notorious, Mr. Flyvbjerg found that solar and wind power projects are among a small handful of project types that are unlikely to go “disastrously” wrong.

Darlington Station itself has an unhappy history. Ontario Hydro approved the station in 1977. By completion in 1993, its cost had ballooned to $14.5-billion, nearly double initial estimates.

OPG’s last major plant overhaul suffered a similar fate. Ontario Hydro mothballed its four Pickering A reactors in 1997, planning to restart all of them at an estimated cost of $780-million. By the time the first fired up in 2003 – three years late – the cost for that unit had almost tripled.

“Management of the project from initial planning to execution was seriously flawed,” concluded a report by an independent review known as the Pickering A Panel. The list of missteps, too lengthy to recount fully here, included failure to recognize the project’s complexity or plan it properly.

“Well-established industry practices and steps for carrying out a project of this size and complexity were not followed.”

Workers arrived at Pickering Station 18 months before engineering design work had been completed. So instead of deftly co-ordinating the work, it was parcelled out haphazardly as plans arrived. Sometimes completed work was later deemed unnecessary, or else had to be redone. And because material specifications arrived late, procurement was on a just-in-time – or, really, a not-in-time – basis.

The project overlapped with the Sept. 11, 2001, attacks, after which security at nuclear facilities tightened considerably. Workers lined up daily in queues, waiting as long as three hours just to gain entry.

The Pickering A review was so scathing that in its aftermath it would have been difficult to justify allowing OPG to refurbish a toaster. Its three top executives and its entire board resigned. In 2005, the new board decided that restarting Units 2 and 3 was commercially unjustifiable (both remain mothballed to this day); five years later, OPG passed on refurbishing the plant’s other four reactors, known as Pickering B.

Lessons Learned

Many working on the Darlington refurbishment were early in their careers, or hadn’t yet completed their education, during the Pickering A debacle. (Ms. Sinnathamby joined OPG in 2001.) Yet the Darlington refurbishment incorporated many relevant lessons.

This time, OPG appears to have grasped the refurbishment’s complexity from the outset. Early planning began a full decade before the project officially started, and engineering work was completed up front.

OPG also built a full-scale reactor mock-up replicating every hallway, door, pipe and light fixture to precise measurements; if a piece of equipment won’t fit into the mock-up, it won’t fit into an actual reactor vault. Decked out in the same plastic suits and respirators worn in the actual vault, workers practised using the same tools and procedures, without the risk of radiation exposure. Ms. Sinnathamby said the mock-up paid for itself many times over because trades showed up ready to work.

“They’re not seeing the tool set for the first time, because they’ve seen it in training and they’ve built the proficiency in that environment,” she said.

Parts were acquired for all units at once, Ms. Sinnathamby said – a move that paid unexpected dividends when the pandemic roiled global supply chains.

As many as 2,000 OPG employees and contractors work on the refurbishment, roughly doubling the station’s regular operating staff. But the two groups have been physically separated: Refurbishment workers have their own office building and parking lot.

Open this photo in gallery:

Open this photo in gallery:

Amid a construction boom, and with Bruce Power starting its own refurbishment in 2020, OPG knew boilermakers, millwrights, scaffolders, welders, insulators and other trades would be scarce. Ms. Sinnathamby said the two companies co-ordinated as best they could.

“If they’re doing a series that’s very similar to ours, when theirs is done, we would transition a number of trades over to our project, and vice-versa,” she said.

Not everything went according to plan. The COVID-19 pandemic set the project back several months: An OPG report published late last year estimated $175-million of additional costs. The current plan is to have all four reactors returned to service by the end of 2026, compared with the original target of February that year – a significant but hardly catastrophic setback for a decade-long project.

In 2018, several workers in the retube waste storage building were found to have been exposed to radioactive alpha particles. A similar problem had occurred during an earlier refurbishment at Bruce Station. “And so when it showed up on the First Darlington unit, it was like, ‘Whoa, we should have learned from that one,’ ” said Rumina Velshi, president of the Canadian Nuclear Safety Commission.

But Ms. Velshi praised OPG’s response, which she described as “absolutely right.”

“They shut down, they regrouped, they addressed it.”

Ontario Auditor-General Bonnie Lysyk reviewed the refurbishment in 2018 and followed up in 2020. She found the most significant cost overrun occurred during construction of a storage facility for heavy water. That ran about $400-million over budget before OPG terminated the contract. But Ms. Lysyk found OPG “subsequently established time and cost estimates for the Project based on reliable information and reasonable assumptions,” and also applied lessons from previous refurbishments.

Learning from past mistakes might seem obvious, yet that’s precisely where many projects fall down. In his book, Mr. Flyvbjerg identified what he described as “uniqueness bias.” Planners tend to view their projects as special undertakings unfolding in exceptional circumstances and with highly specific objectives. Therefore, they find “little or nothing to learn from earlier projects,” he wrote.

Tap to view the before and after

Darlington station's Unit 3 in October 2014, and again in May 2023 amid the ongoing project to refurbish the four reactors at Canada’s second-largest nuclear station. Frank Gunn/The Canadian Press, Carlos Osorio/The Globe and Mail

OPG continued finding better ways of doing things. In Unit 2, it removed pressure tubes first, and then the calandria tubes that housed them. This didn’t go well: Fittings that separate the two tubes, known as spacers, sometimes got knocked off in the process, creating a radiological hazard. For subsequent units, OPG changed its tools and processes to remove both tubes simultaneously.

It’s one reason Unit 2′s refurbishment took 44 months, but the final one is expected to take just 37 months. Though risks remain, there’s no reason to suppose remaining units will be any more challenging to refurbish than the previous ones.

But there remains a considerable risk the refurbished reactors’ performance might not live up to expectations.

In 2012, New Brunswick’s Point Lepreau Nuclear Generating Station restarted after a botched refurbishment. In recent years its lone reactor has been out of service for long stretches, in part because scheduled maintenance outages have consistently taken longer than NB Power projected. Its poor performance is one reason NB Power remains mired in debts.

Early performance at Darlington Station, though, has been encouraging: After its first restart, Unit 2 ran for 529 consecutive days.

“None of the other CANDU plants that have come out of refurbishment have had that sort of a run,” Ms. Sinnathamby said.