WHEN IT STARTED
Westinghouse sold unfinished nuclear plants, leading to major fiscal trouble
Second of two parts
About a decade ago, after years of thinking up the next best thing in nuclear power, Westinghouse Electric Co. sold eight AP1000 plants to kick off the nuclear renaissance.
The deals — two in China, one in Georgia, and one in South Carolina — were worth more than a mere $30 billion.
In the U.S., they would represent the new way of building nuclear plants. They would overcome the history of outrageous delays and cost overruns on every other nuclear power plant built in the country to date — a historythat led to the 30-year hiatus in nuclear construction. The AP1000projects ended the lull.
But when Westinghouse signed the deals in 2008, the Cranberry-based nuclear firm’s design of the power plants wasn’t yet complete.
There are two ways to view that.
One is to note that, historically, all other U.S. nuclear plants had been sold and even moved into construction with the design still being finalized.
Another is to ask: Shouldn’t the industry have learned from the mistakes of the past?
In the pages of Westinghouse’s bankruptcy dockets, in state houses and newspapers, those affected by the company’s failures — regulators, creditors, and ratepayers — are looking for the road map of how a promising multibillion nuclear construction project in South Carolina came to a premature end this summer. More than 6,000 people have lost their jobs at V.C. Summer. Westinghouse is laying off 1,500 employees globally.
“The surprise is that there are no surprises,” Dan Yurman, a nuclear industry blogger, wrote in September. “All of the faults that caused the project in South Carolina to come to an early halt are with failures to follow project management schedule and cost control standard practices that have been known since Admiral Rickover supervised the construction of the first nuclear submarines in the 1950s.”
It has been a long time since a U.S. company commissioned a new nuclear plant, even though nuclear power provides about 20 percent of the nation’s electricity.
In the late 1960s and early 1970s, the nuclear construction boom in the U.S. saw as many as 70 reactors being built at the same time.
“Back in those days, you had designers and [equipment manufacturers] that were doing estimates with only 10 percent of design being done,” recalled Dale Klein, a former chair of the Nuclear Regulatory Commission who now serves as associate vice chancellor for research at the University of Texas’ Cockrell School of Engineering.
At Plant Vogtle in Georgia — where the only two U.S. AP1000 plants are still being built — two existing nuclear power plants went into service in the 1980s.
They were supposed to cost less than $1 billion. The price tag was closer to $10 billion.
The sales pitch behind the AP1000 was that it had evolved past the pitfalls that accompanied engineering on the fly.
It could be built using modules manufactured at factories and assembled on site as a cookie-cutter plant, Westinghouse said. The design relied on gravity to keep the fuel cool in case of an accident, which allowed for a compact footprint, the company said.
That it was a first-of-akind plant, being built in a new nuclear regulatory licensing regime, wasn’t lost on Westinghouse. The company braced for some challenges but not the ones that came.
An earthquake and tsunami devastated the Fukushima Daiichi plant in Japan; natural gas became the cheap fuel of choice for the U.S.; and it became clear that decisive action to mitigate climate change wasn’t a near-term prospect.
“[The] contracts were signed in the context of nuclear renaissance,” Jose Emeterio Gutierrez, Westinghouse’s CEO, said in May at a Nuclear Energy Institute conference. “We know well that that never happened.”
How does that account for the current predicament?
A number of experts and Westinghouse insiders suggested that Westinghouse wasn’t so naive as to expect everything to go smoothly. It anticipated it would likely lose money on the first AP1000 plants.
That would be the price of learning lessons, building its supply chain, and training its engineers. The money wouldbe more than made up for in the volume of future orders, the company thought.
“It’s not that Westinghouse was just reckless,” said Sola Talabi, a former risk manager for the company. “We thought there was going be at least one more plant to distribute these costs over.”
Duke Energy dragged its feet and then canceled plans to build two AP1000 plants in Florida. A follow-up order for eight AP1000 plants in China hasn’t been finalized.
Plans to bring the reactor to India have appeared on the brink for years with no contract in hand. With the cancellation of the project in South Carolina, the project pipeline has actually shrunk.
Where is the schedule?
Against a chorus of optimism and glossy photos of giant steel structures moving around expanses of dirt at Plant Vogtle in Georgia, Bill Jacobs has been a skeptic.
A nuclear expert who monitors activity at the site and reports his findings to the Georgia Public Service Commission each month, Mr. Jacobs has been warning for years that the completion dates that Westinghouse and the utilities were using were unrealistic.
The delays were evident — a part that should have been delivered wasn’t, or its flaws required extensive rework. Instead of tacking those delays onto the schedule, Westinghouse and its construction partner, Stone & Webster, collapsed the timeline, assuming they would be able to make up the slack later.
Mr. Jacobs told regulators this summer that an honest project schedule — one that integrated engineering, procurement and construction costs into one picture — could have avoided the shock that everyone from Westinghouse to utility executives have displayed in recent months at how things have turned out.
“Had an economic analysis been performed at that time,” he said — that time being five years ago — “before hundreds of millions of dollars of sunk costs were incurred, a more accurate picture of the value of the project relative to its alternative would have been developed.”
The lack of a finished design was initially a problem for estimators: It’s hard to know how much steel or concrete you’ll need or how many man hours to schedule if you don’t know exactly what the plant will look like when you start.
But it evolved into a workflow issue since the designs kept changing as components were delivered.
“Often times, the design would change and a component couldn’t be constructed with materials already built. So it had to be purchased [again],” Lonnie Carter, CEO of Santee Cooper, a state-owned utility that owned 45 percent of the nowcanceled V.C. Summer project, told a South Carolina Senate committee last month.
Meanwhile, the original design had to go back to Westinghouse’s Cranberry office to be reworked, launching a domino of recalculations through various engineering departments that could take months.
All of this was cutting into the engineer’s task of finishing the whole design, thereby further pushing the schedule and the materials and the workforce.
As delays piled up, Westinghouse switched from providing utilities with electronic monthly reports to physical printouts — hundreds of 11-inch-by-17-inch pages of site activities that were harder to digest.
Westinghouse also had another mouth to feed. The four AP1000 plants being built in China were supposed to be as much as 18 months ahead of the U.S. projects.
But there, too, the lack of a final design and other factors had caused delays.
Construction, which was being handled by Chinese firms and not the Westinghouse-led consortium, had progressed to the point of erecting structures and still the design hadn’t been completed.
So, in 2013 virtually all of Westinghouse’s engineering bandwidth got shifted to China with the idea that once design was completed there, it would translate to the U.S. projects. The design for the Chinese power plants was finally done in September 2014. Then the U.S. push began. In an upbeat article in Modern Power Systems magazine in January, three Westinghouse project management directors wrote that the company’s experience in China with nuclear modules led to “more than 14,000 module-related improvements to be realized by the U.S. AP1000 plant projects.”
Read another way, there were 14,000 changes to the procurement, fabrication and assembly of the 714 submodules that make up an AP1000 power plant.
The improvements saved between 30 percent and 70 percent on the cost, the Westinghouse officials wrote.
In a paper titled “AP1000 worldwide deployment, learning through delivery,” Westinghouse authors wrote that by 2012, the first AP1000 unit at Plant Vogtle had 4,800 design changes and 300 “design deviations.”
Yet as late as September 2014, Westinghouse was still showing slides at international nuclear conferences promising a four-year construction schedule from the first concrete pour.
At Vogtle, the concrete milestone was reached in 2013 and its current projected completiondate is 2021.
In China, where concrete was poured in 2009, the first plant is expected to have fuel loaded into it late this year.
Difficult shipping logistics
When Mr. Talabi, the former risk manager at Westinghouse, was tasked with assessing the risk of building the first AP1000 plants, he used NASA projects as a reference. The complexity involved, the regulatory regime, the lack of an established supply all skewed toward something outside the realm of a typical large infrastructure job, he said.
The biggest risk identified in the beginning, he said, was the supply chain.
Over the decades when nuclear construction was at a standstill in the U.S., this country lost the ability to produce certain large components. Mr. Talabi knew that many critical parts would have to be forged abroad. Even getting them to a site was a logistical labyrinth.
The steam generators weighed more than 1.3 million pounds each and spanned more than 80 feet. Westinghouse had to modify the nation’s largest train car and cut down trees along the route to transport them from a port in Savannah to the site.
Given these baked-in challenges, the wrenches thrown in at every step weighed heavily on the project’s timeline.
When modules coming out of a factory in Louisiana arrived at Vogtle and V.C. Summer, quality control inspectors crawled over them to document deficiencies. Workers onsite then had to take them apart, fix errors or refabricate components — negating the promise of modular construction.
When the Nuclear Regulatory Commission mandated that the shield building that holds the reactor must be able to withstand an airplane crash, that delayed the final design certification until late 2011 — nearly four years after Westinghouse sold the plants.
Slower, more expensive
When workers who weren’t yet steeped in nuclear culture would lose documentation for certain components or when their product tags weathered beyond recognition as they awaited installation, mountains of metal were put out to pasture in large laydown yards.
A ding in the chain of custody of a nuclear component renders it useless.
Productivity on both sites reached a low in early 2016.
Workers at Vogtle were taking more than twice as long as the budget projected during the last six months before Westinghouse’s bankruptcy, according to Kenrich GroupLLC, a consulting firm hiredby Southern Co.
These productivity lags dominoed across the supply chain, Kenrich said.
Each 10 percent decrease in productivity, the consultant said, would add $203 million to the price tag.
In August 2016, Westinghouse hired a consultant to collect data on how workers were spending their time. The results were never publicly disclosed, but the consultant made more than 100 recommendations to boost productivity and cited “idle time, early quits and late starts” as a concern. Mark vice Rauckhorst,at presidentVogtle, toldof executive constructionthe Georgia Public Service Commission this month that over the past year, Westinghouse’s financial situation was to blame for the productivity deficits. “Contracts with key suppliers were delayed or suspended; the workforce was reduced; and we began to hear from various subcontractors that needed material was not going to be delivered to the site because Westinghouse had stopped paying them,” he said in written testimony. Southern wasn’t told that this was happening until after Westinghouse’s bankruptcy, Mr. Rauckhorst claimed. “Throughout the process, Westinghouse attempted to maintain transparency in its communications with the owners,” Westinghouse spokesperson Sarah Cassella said. Project happenings were communicated in meetings, through change orders, “president’s meetings and many other communications,” she said. Current and former employees at Vogtle questioned whether Southern, which had some 200 people devoted to the project, could have been as caught off guard as its leadership has been claiming. the Since Vogtle Southern project tookthis summer,over Mr. Rauckhorst said, it has tightened the grip over financials, restricted design changes, and improved productivity.
Mr. Gutierrez, the Westinghouse Westinghouse CEO, said last month that work is “progressing well.”
“In the past month, we have been working well with Southern and things are moving,” he said. “Nothing bad to report and everything positive.”