This article reviews that for the first time in a generation, utilities are starting the regulatory process to build nuclear reactors. There has been a virtual moratorium on new nuclear power plants in the United States during the past generation, and it has many causes. But one significant factor in the industry's decline was the Nuclear Regulatory Commission’s licensing process. There are now dozens of applications being submitted and approved for 20-year license renewals for established nuclear power plants. But before the nuclear power industry truly can be said to be reborn, new reactors must be constructed. The new rules allow for an early site permit and for a separate combined construction and operating license. Although the commission invited the nuclear power industry to test the two new processes when they were first announced, no company volunteered. One of the thorniest technical issues faced by the early applicants so far involves a new way of calculating, for a specific plant site, the ground motion that would result from a seismic event. When older plants were designed and built, the best available technique for these calculations was deterministic.
There's a risk in any business venture, to be sure. But imagine building a multibillion-dollar plant that might not even be allowed to operate. That was what utilities had to face in the 1970s when they constructed nuclear power plants.
The old way of licensing "was that you picked a design and then applied for a construction permit," said Kenneth Hughey, senior manager of new plant licensing and quality assurance at Entergy Nuclear, which owns one of the largest portfolios of nuclear power plants in the United States. "After you built the plant, then you'd apply to operate it. That was when the NRC really looked at the design."
There has been a virtual moratorium on new nuclear power plants in the United States during the past generation, and it has many causes. But one significant factor in the industry's decline was the Nuclear Regulatory Commission's licensing process. Because it didn't allow design approval until after construction, the process caused many costly delays while completed plants waited for operating licenses.
There are now dozens of applications being submitted and approved for 20-year license renewals for established nuclear power plants. But before the nuclear power industry truly can be said to be reborn, new reactors must be constructed. While the NRC has replaced the old licensing process with a new, presumably streamlined procedure, the new regulations haven't been fully tested. As the industry moves toward the first new orders for reactors since the 1970s, the licensing process has proved to be more difficult than anyone expected. Thousands of engineering and NRC staff hours have been poured into the effort.
And that's even before a single spade has touched the ground.
The old regulatory process, established in 1956, had been widely seen as a cumbersome failure. In 1992, the NRC established two new processes to streamline nuclear plant licensing. For instance, under the old 10 CFR 50 or "Part 50" process, regulations could change while a plant was under construction, requiring major retrofitting after completion. Perhaps even worse for plant operators, the operating licensing phase allowed public litigation regarding a plant after construction was all or mostly done. While the litigation and hearings dragged on, fully completed plants could sit idle for months or even years. In the meantime, utilities paid interest on construction loans instead of earning operating revenue. Some utilities couldn't afford the delays and walked away from nearly finished facilities.
The new rules allow for an early site permit and for a separate combined construction and operating license. Although the commission invited the nuclear power industry to test the two new processes when they were first announced, no company volunteered.
By 2003, however, after the September 11 attacks, the California blackouts, the steep rise in the cost of gas powered electricity generation, and the growing awareness that carbon emissions are leading to global warming, nuclear power again became an option. Three nuclear utilities, Exelon Generation, Dominion, and Entergy Nuclear-stepped up to test the early site permit process. In 2003 and 2004, 10 utilities, led by Exelon, formed the NuStart Consortium to test the construction and operating licensing process.
In contrast to the old Part 50 procedure, the new 10 CFR 52 or "Part 52" regulations aim to settle questions about siting, design, and operation before major investments are committed.
"The idea now, starting with the early site permit, is to resolve siting issues well ahead of time and then resolve design and construction and operating licensing issues well before you start construction," said Marvin Smith, director of new reactor projects for Dominion. "With the new process, you design and license, and then build and operate." Under Part 52, public hearings for litigation also take place well ahead of the construction phase (except in extraordinary circumstances), so utilities no longer need worry about losing money daily on a complete, but idle, plant.
Within the Parameters
Standardization of every aspect of an application's form and content is also a crucial feature of Part 52. With this approach, the NRC will have to review and approve a specific design only once. After that, it need look only at areas where new applications differ, such as the environmental characteristics of a specific site.
"This has immediate payback in regulatory efficiency, but the bigger payback is operational," said Marilyn Kray, vice president of project development for Exelon Nuclear and president of NuStart Energy Development. "If you have the same design, you can share parts, operating experience, training, and procedures."
Issuance of a Part 52 early site permit, or ESP, depends on whether or not a specific location is suitable for a nuclear power plant. The site permit doesn't allow for actual construction. According to Hughey, the site permit considers issues such as flooding, wind and snow loads, hurricanes, tsunamis, or seismic activity. "It also looks at public safety issues from an emergency-planning point of view," Hughey said.
In September and October 2003, Exelon, Dominion, and Entergy (through its subsidiary, System Energy Resources Inc.) submitted applications for ESPs. None of the three was ready to specify a reactor design. "We decided to include a group of designs and build a set of criteria that would bound all of them," Kray said. "Then we would evaluate our sites against this plant parameter envelope."
Although the plant parameter envelope approach sounds clever, the NRC was used to reviewing applications based on a specific design. The commission kept asking for design details.
"We told the NRC that we weren't going to provide more detail than the PPE," Hughey said. "The NRC didn't have to review and approve the PPE; it just had to assume those parameters and decide whether or not it would approve the site characteristics."
The Nuclear Energy Institute, an industry trade group, coordinated the three companies' generic approach. With the NEI serving as facilitator between the utilities and the commission, the NRC ultimately accepted the plant parameter envelope technique.
The push for standardization went beyond developing a generic, plain vanilla· nuclear plant model. The parties involved in testing the new licensing process also wanted to establish a standard format and content for early site permit applications. For example, the plant parameter envelope approach defined 70 required plant parameters. Future applicants can use the generic approach, but if they select a specific reactor technology and plant design, they need only provide the information for those 70 parameters. "Having the parameters specified is a gift," Kray said.
According to Adrian Heymer, senior director of new plant deployment at NE!, "Later applicants will benefit from the 70 percent to 80 percent standardization in design and construction, and the NRC will benefit in terms of review speed as well." Heymer estimates that standardization can cut 24 to 30 months from the period between deciding to build and becoming operational.
While continuing to work with the NRC on their early site permit applications, Exelon, Entergy, and Dominion also all decided in 2003 and 2004 to be first in line for testing the construction and operating licensing process. Entergy joined the NuStart consortium with Exelon, but Dominion chose to work independently. According to Kray, the utilities and nuclear equipment suppliers saw the need for regulatory certainty and to firm up plant designs. Also, they felt it was necessary for the industry to take a unified approach in dealing with issues and to achieve standardization in the construction and operating licensing process.
Kray recalls that the idea in 2003 and 2004 was that NuStart would prepare two construction and operating licensing applications, one for each of the latest-generation reactor technologies, the AP-l000 and the ESBWR. "At the time, it was thought that NuStart would collectively prepare these," Kray said. "Once they were approved by the NRC, individual companies would go forth using these reference applications as the basis for . their own projects." Instead, a number of time-limited financial incentives have spurred several companies, both in and out of the consortium, to simultaneously start working on their own applications.
Although there are independent applications in the works, the consortium helps its members share the costs and risks of working through the new process for the first time. In addition, companies throughout the nuclear power industry-even utilities that aren't NuStart members are using the consortium's work as the basis of their Own efforts, with NuStart's blessing. Future applicants will do the same.
Investing in Nuclear
Nuclear power plants are enormous investments. And the financiers who provide the money for building new reactors are somewhat divided over the near-term future of nukes.
On the one hand, the current work of testing the Nuclear Regulatory Commission's new permitting and licensing procedures for new plants is roundly applauded. "Most Wall Street analysts would give it a resounding 'yes: " said Gary Hovis, vice president and senior utility analyst for New York City-based Argus Research Group
But the prospect of actually funding new nuclear construction has received a cooler reception. "Once nuclear plants are built, Wall Street analysts love them because they're safe and they work," said Kenneth Hughey, senior manager for new plant licensing and quality assurance at Entergy. When it comes to building new plants, however, analysts are "scared that nuclear is the only technology that must go through this Nuclear Regulatory Commission licensing process," Hughey said.
The old Byzantine licensing process created delays that could double or triple a plant's ultimate cost. The fear on Wall Street is that similar cost overruns will dog new nuclear plants. Hovis cites one New York plant where costs ballooned from $2.3 billion in 1981, when construction started, to $6 billion in 1990 when operations began.
"The utilities involved had to write off about $3.7 billion," Hovis said. "That reduced shareholder equity, increased the debt-to-total-market-capitalization ratio, reduced the utilities' credit ratings, and forced them to pay higher interest rates." That's bad for investors, power companies, and customers.
NRC's new streamlined licensing procedures are intended to significantly reduce delays and the consequent cost overruns. One or two successful applications still won't be enough to stimulate investment capital, though. Before that happens, Hughey said, "Wall Street wants to see a number of examples of a consistent, reliable process."
According to Kray, consortium members depend on NuStart to prepare the final safety analysis and the environmental report, which account for much of the application. Kray estimates that about 75 percent of the safety analysis for each reactor design is generic and that "the rest is site-specific."
Because the environmental section of an application is very site-specific, future applicants won't duplicate that part of NuStart's construction and operating licensing documents. They will, however, use the same approach and format, unlike in the past when many applications started from scratch.
One of the thorniest technical issues faced by the early applicants so far involves a new way of calculating, for a specific plant site, the ground motion that would result from a seismic event. When older plants were designed and built, the best available technique for these calculations was deterministic. And the only available data for the calculations came from seismic events in the western U.S. Now, both the NRC and industry want utilities to use a probabilistic method and, where appropriate, new data from seismic events in the central and eastern U.S. Not everyone has been happy with that decision.
The deterministic method had three steps, according to Nilesh Chokshi, deputy director of the site and environmental review division in the NRC's office of new reactor licensing. First, a utility would gather the best available scientific data about the site-its seismology, its geology, and the history of seismic events in the area. The area, or "seismic zone," doesn't just mean next door to the plant site, either; events as far away as 300 miles away must be considered.
Next, engineers calculated the peak ground acceleration at the site. Finding that single number might seem sufficient for a plant design, but real seismic events produce a whole spectrum of ground acceleration values. A design must account for them all. So in the last step, designers essentially match their site's peak ground acceleration to a "generic" full spectrum curve derived from western U.S. earthquake data. Thus, a site's full ground-motion spectrum can be determined and used in the plant design.
The new method takes new earthquake data into account and uses a probabilistic way of calculating the ground response spectra. "In the last 15 to 20 years, we have obtained new data from the eastern U.S.," Chokshi said, "and we've learned new approaches."
What those data show is that, for a particular site, an eastern or central U.S. earthquake could generate more high-frequency (greater than 10 Hz) ground motion, whereas a western U.S. earthquake of the same magnitude and distance from the site would generate more low-frequency (less than 10Hz) ground motion. So using only western seismic data to assess the response of a plant site in the eastern or central U.S. won't fully account for high frequency motion.
Although the deterministic method wasn't wrong, it didn't address uncertainties in the actual size of a seismic zone, the size of the maximum earthquake in that zone, or how the ground motion of such an event would attenuate over distance. For example, the deterministic model used the biggest earthquake that could affect a site operato find the site's ground response characteristics and peak ground acceleration. "The probabilistic model came about to address the uncertainties of the deterministic model," Chokshi says. To find some uncertain quantities, the probabilistic method can examine different rates and magnitudes of seismic events over varying distances and much more.
New Designs for the U.S.
The last nuclear reactor built in the United States was at the Watts Bar plant in Tennessee and, following a recent decision by the Tennessee Valley Authority, it looks like the next reactor to be built in the U.S. will be at the same place. Both use a tried-and-true technology. Other future reactors, however, will likely be in one of two designs never before used in this country.
One of them, the AP-l000 power plant designed by Westinghouse, is a 1.1-gigawatt pressurized water reactor based on a previous (and never-built) 600-megawatt version. According to Westinghouse, the design features two steam generators connected to the reactor pressure vessel and an array of passive safety systems that don't require ac power or cooling water to be operational. The design also requires halt the valves and about a sixth of the piping of a conventional plant with similar output. Westinghouse claims it can be constructed in as few as 36 months.
The other plant is a General Electric design called the Economic Simplified Boiling Water Reactor. It is billed as an evolutionary step in boiling water reactors. The recirculation and safety system pumps have been replaced by natural, gravity-driven water circulation and passive safety systems. Indeed, GE claims that the ESBWR has more than 72 hours of passive running capability. The simplified systems should enable cuts in construction and operation costs, according to the company.
The Next Reactor
It looks like the next nuclear reactor to be built in the United States will be in Tennessee, right beside the last one to come on line in the country. The Tennessee Valley Authority is taking construction bids to complete Reactor 2 at its Watts Bar power station and expects to start construction this fall.
Watts Bar 1 began commercial operation in May 1996, and since then no other reactor has come on line in the U.S.
According to the TVA, the board's decision to complete the second reactor was based on the results of four studies that examined future power needs, cost and schedule, environmental impact, and financing and risks.
Construction is expected to begin this fall and continue for approximately five years. The site has been cleared, since a reactor is operating there already.
A spokesman said the TVA has a construction license valid until 2010 for the second reactor. The authority will need to extend it because construction will not be complete until sometime in 2013. Reactor 2 was partly built when construction stopped in the mid-1990s. It took the TVA more than 20 years to build Reactor 1.
The TVA said its board unanimously approved completing Watts Bar Unit 2 at an estimated cost of $2.49 billion. When completed by 2013, the nuclear unit will provide about 1,180 megawatts of electricity, or enough to supply about 650,000 homes a day. Reactor 1 has about the same capacity. Both designs are Westinghouse pressurized-water reactors.
The Watts Bar plant is on 1,700 acres beside Chickamauga Reservoir in Spring City, Tenn. It is named for a bar at Watts Island that blocked navigation in the Tennessee River channel before it was flooded by the reservoir.
Not everyone sees the new approach as an improvement. According to Jack Bailey, vice president of nuclear generation development at Tennessee Valley Authority, a NuStart member, when potential plant sites experience seismic events once in a hundred years (or even more rarely), the absence of current, reliable data from such events means there is a high level of uncertainty in estimating the seismic response. Subsequently, the probabilistic method can exaggerate the resulting ground motion. "It's been said that Diablo Canyon can satisfy the new requirements better than a site farther away from a fault," Bailey said. That's because the Diablo Canyon facility in California has real data that reduces the uncertainty of calculating a site's seismic response.
The youngest plant: Reactor 1 at Watts Bar in Tennessee was the last to come online in the U.S. TVA has voted to complete construction of Reactor 2, so it will likely be the first in a new round of development.
Another complaint is that low-frequency ground motion is known to cause the lion's share of damage to buildings, so why should it be necessary to evaluate the effects of high-frequency shaking? Some plant components, however, such as electrical relays, are sensitive to high-frequency ground motion, Chokshi said, although he acknowledges that plants that experience such motion aren't expected to incur significant structural damage. Nevertheless, he said, "An applicant must show that the design they'll use at a specific site can withstand any site-specific ground motion."
The seismic issue is just one of many that the first construction and operating license applicants are addressing. The NRC's invitation to the industry to test the new Part 52 processes was specifically intended to discover and resolve just that sort of thorny issue in order to produce standardized solutions, form, and content.
"All this has worked as intended," Heymer said. "It would have been nice to have done it 10 years ago, but we're doing it now."