US reactors prepare to review seismic hazards23 April 2012
In March 2012, the Nuclear Regulatory Commission asked all 104 nuclear power plants in the United States to re-evaluate seismic hazards. As part of the process, plants in the Central and Eastern US will use a new seismic model that includes data from earthquakes over the last 440 years. By Caroline Peachey
The safety of nuclear power plants against earthquakes and other natural phenomena has been thrust into the spotlight recently following two significant earthquakes in 2011. The first was the 11 March 2011 earthquake and resulting tsunami that crippled the Fukushima Daiichi nuclear plant in Japan. The second occurred in August 2011 in Virginia, USA, and forced three reactors at Dominion’s North Anna nuclear power station offline for over two montahs.
The NRC Fukushima Near Term Task Force (NTTF) has recommended that the re-evaluation of seismic and flooding hazards for each nuclear power plant in the United States (recommendation 2.1) be carried out ‘without unnecessary delay.’ (It is among eight Tier 1 or top priority recommendations resulting from the USA’s post-Fukushima safety review.) However it is important to note that a review of seismic hazards at US nuclear power plants has been on the agenda for over a decade.
This re-evaluation has now officially begun. The NRC has issued a formal request (docket number ML1203A188 on the USNRC’s ADAMS document database) asking all US nuclear power plants to provide information about the current hazard and potential risk posed by seismic events. This process will require plants to conduct a probabilistic seismic hazard analysis (PSHA). PSHA is an analytical method that estimates the likelihood that various levels of earthquake-caused ground motions will be exceeded at a given location in a future time period.
As part of this PSHA, NRC is expecting plants in the central and eastern United States to use the new Central and Eastern United States Seismic Source Characterization for Nuclear Facilities model (CEUS-SSC). This model contains the most up-to-date seismic source information for the CEUS, replacing the previous regional models that date back to the 1980s and 1990s. Plants in the Western US, a seismically active region, will use the latest site-specific earthquake models and data.
While the findings themselves will be site-specific, calculations made with the new model are expected to lead to an increase in seismic hazard, according to the Electric Power Research Institute.
“Calculations with the new model are expected to result in a higher likelihood of a given ground motion compared to calculations done using previous models,” the NRC said in a statement.
“These calculations, however, are not equivalent to a nuclear power plant’s overall risk,” it continued, adding that nuclear power plant operators must combine the information from the new seismic source characterization model with a plant’s design and safety features to determine site-specific risks.
This re-evaluation could potentially be very significant for US nuclear power plants, leading to requirements for upgrades to the design bases of structures, systems and components (SSCs) important to nuclear safety to protect against updated seismic hazards. However the NRC says that it is premature to try to assess how many, if any, nuclear power plants might need additional work. It is expected to be several years before the seismic hazard re-revaluations are completed and assessed. Following that, there will be a further period of several years during which the NRC will consider the regulatory implications of those findings.
Motivation for a new seismic model
In the early 2000s when utilities started using the new reactor licensing process, they began taking advantage of new technology, using a probabilistic approach to seismic hazard assessment, rather than the deterministic approach that had been originally used to license the 104 US reactors in operation today.
While the deterministic approach focused on a single earthquake event (the strongest earthquake that occurred during the previous 100,000 years) to determine the finite probability of damage occurring, the PSHA attempt to quantify the probability of exceeding various ground-motion levels at a site given all possible earthquakes. The differences between these approaches prompted the regulator to compare the two ways of doing things.
In support of the early site permits (ESPs) and combined license applications (COLs) for new reactors, the NRC staff reviewed and compared the seismic source and ground motion models provided by applicants to the previous deterministic models. These reviews identified higher seismic hazard estimates than previously assumed, and NRC decided that further study was warranted.
In 2005, the NRC started Generic Issue 199 to take a broader look at the issue. Staff conducted a screening analysis on all 96 plants in the CEUS using a probabilistic approach. In that analysis there were 27 instances where the magnitude of the change in the seismic hazard estimate compared with the original deterministic method was sufficient to warrant additional investigation. In order to conduct a more detailed analysis, however, the industry required a recent model for seismic sources in the Central and Eastern United States. (The seismic source model used for the screening analysis was the latest US Geological Survey, but it focused on the more frequent events rather than the less frequent events (10-5 to 10-7) that are important for critical infrastructure such as nuclear power plants.
April 2008 saw the launch of a $7 million project to update earthquake and geological data for the CEUS region. The Central and Eastern United States Seismic Source Characterization for Nuclear Facilities model was released in January 2012. It is the result of a four-year project, sponsored jointly by EPRI, the US Department of Energy and the US Nuclear Regulatory Commission, which gathered and analyzed new historical earthquake and geological data for the CEUS region.
The joint sponsorship of the study by both public and private sector representatives is unique for regional seismic hazard assessments in the United States. A primary objective of the collaborative effort was to develop a “consistent and stable model for performing PSHAs for future reactor license applications that will have greater longevity and reduce the time and cost required to complete a commercial nuclear site’s Early Site Permit or COL licensing application.”
Technical experts from DOE, NRC, U.S. Geological Survey (USGS), Defense Nuclear Facility Safety Board (DNFSB), nuclear power industry and academia participated in the project as part of the technical integration team or as members of a peer review panel.
The work, which was both theoretical and empirical, involved updating of data, models, and methods for characterizing seismic sources in the Central and Eastern United States, according to Jeff Hamel, programme manager of advanced nuclear technology at EPRI, who was part of the management team for the CEUS-SCC project.
“As part of the project we developed a new earthquake catalogue of historical events in the region from 1568 to 2008,” said Hamel.
This comprised four main steps: catalogue compilation, assessment of a uniform size measure to apply to each earthquake (to establish a consistent basis of comparison), identification of dependent earthquakes, that is, foreshocks and aftershocks (also known as catalogue declustering), and assessment of the completeness of the catalogue as a function of location, time, and earthquake size.
The final project catalogue contains 3298 individual earthquakes of uniform moment magnitude. Most of these earthquakes (2642) are also contained in the 2008 USGS seismic hazard-mapping catalogue. Ten regional catalogues were used to obtain additional information on the size measures for earthquakes and to identify the magnitude types reported for each earthquake.
The project team gathered the latest data through three public workshops where experts from inside and outside the nuclear seismic community presented work that had been conducted over the last 20-25 years–since the last CEUS seismic characterization model was produced.
In addition to compiling the database, considerable effort was devoted to the development and incorporation of new methods for assessing key seismic source characterization parameters.
For example, the CEUS SSC model is based largely on the assumption, typical in probabilistic seismic hazard analysis (PSHA) studies, that spatial stationarity of seismicity is expected to persist for a period of 50 years. Spatial stationarity is a model in which the locations of future earthquakes are assumed to follow the spatial distribution of past earthquakes. The improved method for spatial smoothing used in this project provides maps that express the future spatial pattern of recurrence rates.
When asked to comment on the length of time since development of the previous model, Hamel said that from a scientific perspective such models are only updated ‘once significant new science data is accumulated.’
The NRC is in the process of considering another Fukushima task force recommendation 2.2: to re-evaluate seismic and flooding hazards every ten years. However according to Cliff Munson, of the NRC it is likely to be two to three years before any rulemaking on this.
As part of the CEUS-SCC project, the new seismic model was compared to previous models by calculating seismic hazards at seven test sites.
These demonstration hazard calculations were carried out to gain insights into the uncertainties in the model.
Akes said that the sites were specifically chosen not be at any nuclear power plant or DOE facility. He said that one site chosen was relatively close to Charleston, SC, the site of a major earthquake in 1885, while others were mid-continent with no seismic sources nearby.
The sample calculations indicated that the largest predicted ground motions could occur in the vicinity of repeated large magnitude earthquake sources, such as New Madrid, Missouri and Charleston, South Carolina. Other RLME sources are Charlevoix (lower St. Lawrence Quebec, Canada), Cheraw Fault (High Plains in southeastern Colorado), Meers Fault (southwestern Oklahoma), Reelfoot Rift – Marianna (Marianna, Arkansas; 75 km southwest of Memphis, Tennessee), Reelfoot Rift – Commerce Fault Zone (Tamms, Illinois to Qulin, Missouri) and the Wabash Valley (Indiana and Illinois).
The main advantage of the new model from a regulatory perspective is that it will give a “common starting point for more analysis. This is essential for ensuring the regulatory approach is consistent for all operating plants,” NRC spokesman Scott Burnell told NEI.
For new reactor applications (ESPs and COLs) licencees used a seismic source characterization models that had been originally developed by EPRI in the late 1980s. On top of this they also had to include updates for various sources. The CEUS-SSC model was developed using the NRC’s Senior Seismic Hazard Analysis Committee (SSHAC) guidelines, which ensure compliance with seismic regulations and properly quantifies uncertainties in seismic design basis for nuclear facilities.
The CEUS-SCC model is ready to be used by all 96 operating reactors in the CEUS region. (There are also 22 potential new nuclear sites and five Department of Energy nuclear facilities in the CEUS region that could benefit from this model.)
How the CEUS SSC model be used
The NRC’s post-Fukushima recommendation 2.1 for seismic hazards will be implemented in two phases.
The first step in the process aims to identify potential plant vulnerabilities. It will comprise a re-evaluation of the seismic hazards for each nuclear power plant site using current NRC requirements and guidance. Each plant will use the latest seismic source and ground motion models to conduct a Probabilistic Seismic Hazard Analysis (PSHA). The result of the PSHA will be a ground motion response spectrum (GMRS) for each site.
This GMRS will be compared with the safe shutdown earthquake ground motion (SSE) curves that represent the seismic design basis ground motion for the plant. The SSE varies from plant to plant; however the minimum mandated SSE for a US nuclear power plant is 0.1g. Plants in California have the highest SSE, which go up to around 0.6g.
Depending on the results of this comparison either no further action would be necessary, or plants would be required to conduct one of two more detailed analyses (either a seismic margin assessment, SMA, or a seismic probabilistic risk assessment, SPRA) in phase 2.
The plant operators in the CEUS will have a year and a half for the first phase of the project, according to the order published in March. Plants in the Western US (Columbia, Diablo Canyon, Palo Verde, and San Onofre), which will need to use site-specific models, will have three years to complete this initial phase.
If it is necessary, phase 2, will determine whether additional regulatory actions are necessary (e.g., update the design basis and SSCs important to safety) to protect against the updated hazards.
No further action would need to be taken if the GMRS is less than SSE or if GRMS exceeds SSE at frequencies only above 10Hz but not at all at anchor point. The frequency range of 1-10 Hz is typically of greatest interest, because this is where structural damage can occur due to seismic energy. Higher frequencies have minimal structural impact, but can affect more sensitive equipment such as instrumentation and electrical relays, says Hamel.
Plants where the reevaluated hazard exceeds the current design basis may decide to carry out a comprehensive seismic probabilistic risk assessment will need to be conducted. Seismic PRA is a complete model of the plant and its robustness to handle earthquake ground motion.
In cases where there is less deviation from the design basis, the NRC said it would provide guidance to utilities to determine whether a SPRA or a Seismic Margin Analysis (SMA) will be needed. Although a plant’s design basis references a certain ground motion, when reactors were built the designers tended to add in some contingency onto this design basis. A voluntary programme carried out in the 1990s, the individual plant examinations of external events (IPEEEs) showed that plants were in fact capable of withstanding earthquakes beyond this initial design basis.
The margins assessment will look at two initiators (station blackout and small break loss of coolant accident) and how these would affect SSC.
Plants that are identified as higher priority that is, those that need to conduct seismic PRA, will have three years to complete this stage. Plants that need to conduct the less-detailed SMA will have up to four years, according to the schedule indicated in the order.
In addition to the new CEUS-SSC model, an update to the EPRI (2004, 2006) ground motion prediction equations (GMPEs) are being considered to support the resolution of recommendation 2.1. The update would consider additional and new information obtained since the model was last updated (2006). The assessment process and the evaluations that will be made are such that no predictions of results can be made at this time, EPRI said.
Looking forward, the NRC is expected to publish NUREG-2117 (2012), Practical Implementation Guidelines for SSHAC Level 3 and 4 Hazard Studies, that provides guidance on the need to update a regional model. This guidance covers updating regional and site-specific assessments. It addresses the “refinement” process of starting with a regional model and refining it for site-specific applications, EPRI said.
This article was published in the April 2012 issue of Nuclear Enginering International magazine. Follow Caroline Peachey on Google+