FINDING 4.11: The U.S. Nuclear Regulatory Commission has completed a technical analysis of spent fuel pool accident consequences to inform a regulatory decision on expedited transfer of spent fuel from pool to dry cask storage. The analysis was carried out in accordance with prescribed U.S. Nuclear Regulatory Commission regulatory guidance and provides valuable technical information about the impacts of various accident scenarios on spent fuel storage in pools. However, the analysis did not consider spent fuel storage sabotage risks, dry cask storage risks, or certain health consequences that would likely result from a severe nuclear accident. The analysis also used simplifying bounding assumptions that make it technically difficult to assign confidence intervals to the consequence estimates or make valid risk comparisons. A risk assessment that evaluates the three questions of the risk triplet and that accounts for uncertainties in both probability and consequence estimates is needed to address Finding 4E in NRC (2006) to determine whether “earlier movements of spent fuel from pools into dry cask storage would be prudent to reduce the potential consequences of terrorist attacks on pools at some commercial nuclear plants.”
RECOMMENDATION 4.11: The U.S. Nuclear Regulatory Commission should perform a spent fuel storage risk assessment to elucidate the risks and potential benefits of expedited transfer of spent fuel from
pools to dry casks. This risk assessment should address accident and sabotage risks for both pool and dry storage. The sabotage risks should be assessed using the methodology developed in response to the present committee’s Recommendation 4.1B.
These findings and recommendations arise from the committee’s reevaluation of Finding 4E from the National Research Council (NRC) report Safety and Security of Commercial Spent Fuel Storage (NRC, 2006):
FINDING 4E of NRC (2006): Depending on the outcome of plant-specific vulnerability analyses described in the committee’s classified report, the Nuclear Regulatory Commission might determine that earlier movements of spent fuel from pools into dry cask storage would be prudent to reduce the potential consequences of terrorist attacks on pools at some commercial nuclear plants.
This chapter is organized into four sections. Sections 7.1-7.3 describe the recent U.S. Nuclear Regulatory Commission (USNRC) analyses that were undertaken to assess the need for early transfer of spent fuel from pools to dry casks at U.S. nuclear plants. Section 7.4 provides supporting discussion for the committee’s finding and recommendation.
Following the 2011 Fukushima Daiichi accident in Japan, the USNRC staff began a systematic review of the agency’s procedures and regulations to determine whether improvements were warranted (USNRC NTTF, 2011). This review is described in the committee’s phase 1 report (NRC, 2014, see especially Appendix F). The USNRC staff identified spent fuel transfer from pools to dry cask storage as having “a clear nexus to the Fukushima Daiichi event that may warrant regulatory action . . .” (USNRC, 2011b, p. 5) and subsequently determined that further study of storage arrangements was warranted (USNRC, 2012c).
A few months after the Fukushima accident, USNRC staff initiated the Spent Fuel Pool Study, which examined the consequences of a beyond-design-basis earthquake on a spent fuel pool that is similar in design to some of the pools at the Fukushima Daiichi plant (USNRC, 2014a).1 The primary objective of this study was to
1 The study was completed in October 2013 and published in September 2014 following a public comment period.
“. . . determine if accelerated transfer of spent fuel from the spent fuel pool to dry cask storage provides a substantial safety enhancement for the reference plant. The insights from this analysis will inform a broader regulatory analysis of the SFPs [spent fuel pools] at U.S. nuclear reactors as part of the Japan Lessons-learned Tier 3 plan.” (USNRC, 2014a, p. 3)
The objective of this broader regulatory analysis, hereafter referred to as the Expedited Transfer Regulatory Analysis, was to determine whether “additional studies are needed to further assess potential regulatory action on expedited transfer” (USNRC, 2013, p. iii). The Spent Fuel Pool Study and Expedited Transfer Regulatory Analysis are described in Sections 7.2 and 7.3 below.
Spent fuel pools at U.S. nuclear plants were originally outfitted with “low-density” storage racks that could hold the equivalent of one or two reactor cores of spent fuel.2 (See Appendix 7A for a discussion of spent fuel pool racking.) This capacity was deemed adequate because plant operators planned to store spent fuel only until it was cool enough to be shipped offsite for reprocessing. However, reprocessing of commercial spent fuel was never implemented on a large scale in the United States; consequently, spent fuel has continued to accumulate at operating nuclear plants.
U.S. nuclear plant operators have taken two steps to manage their growing inventories of spent fuel. First, “high-density” spent fuel storage racks have been installed in pools to increase storage capacities. This action alone increased storage capacities in some pools by up to about a factor of 5 (USNRC, 2003). Second, dry cask storage has been established to store spent fuel that can be air cooled.3 Typically, transfers of the oldest (and therefore coolest) spent fuel from pools to dry casks are made only when needed to free up space in the pool for offloads of spent fuel resulting from reactor refueling operations.
The objective of accelerated or expedited transfer would be to reduce the density of spent fuel stored in pools:
“Expedited transfer of spent fuel into dry storage involves loading casks at a faster rate for a period of time to achieve a low density configuration in the spent fuel pool (SFP). The expedited process maintains a low density pool by moving all fuel cooled longer than 5 years out of the pool.” (USNRC, 2014a, p. B-1)
2 Additionally, U.S. nuclear plant operators maintain enough open space in their spent fuel pools to offload entire reactor cores if needed for safety or maintenance actions.
3 All but three operating nuclear plants in the United States (Three Mile Island, Pennsylvania; Shearon Harris, North Carolina; and Wolf Creek, Kansas) have established or are in the process of establishing dry cask storage facilities. See http://pbadupws.nrc.gov/docs/ML1507/ML15078A414.pdf.
The low-density configuration achieved by expedited transfer would reduce inventories of spent fuel stored in pools. This might improve the coolability of the remaining fuel in the pools if water coolant was lost or if cooling systems malfunctioned.
The Spent Fuel Pool Study analyzed the consequences of a beyond-design-basis earthquake on a spent fuel pool at a reference plant4 containing a General Electric Type 4 boiling water reactor (BWR) with a Mark I containment.5 The USNRC describes this study as one in a continuing series of examinations of postulated spent fuel pool accidents (see Sidebar 7.1).
The USNRC selected an earthquake having an average occurrence frequency of 1 in 60,000 years and a peak ground acceleration of 0.5-1.0 g (average 0.7 g) as the initiating event for this analysis.6 The study examined the effects of the earthquake on the integrity of the spent fuel pool and the effects of loss of pool coolant on its stored spent fuel. The scenarios considered in the analysis are summarized in Figure 7.1.
A modeling analysis was carried out to identify initial damage states to the pool structure from this postulated seismic event. The analysis concluded that structural damage to the pool leading to water leaks (i.e., tears in the steel pool liner and cracks in the reinforced concrete behind the liner) was most likely to occur at the junction of the pool wall and floor. This leak location would result in complete drainage of the pool if no action was taken to plug the leak or add make-up water. Given the assumed earthquake, the leakage probability was estimated to be about 10 percent (see upper part of Figure 7.1), which the USNRC staff judged to be conservative.
The Methods for Estimation of Leakages and Consequences of Releases (MELCOR) code (see Sidebar 6.1 in Chapter 6) was used to model the consequences of three leak scenarios, two spent fuel loading configurations in the pool, five reactor operating cycle phases, and two mitigation actions (see Figure 7.1):
- “No leak” in the spent fuel pool.
4 The reference plant was Unit 3 of the Peach Bottom nuclear plant in Pennsylvania. It is similar in design to the Fukushima Daiichi Unit 1-5 reactors and spent fuel pools.
6 The USNRC selected an earthquake as the initiating event for this analysis because its previous studies (Collins and Hubbard, 2001; Throm, 1989) concluded that “seismic events were the largest contributor to the frequency of fuel uncovery” (USNRC, 2014a, p. 8).
- A “small leak” in the pool that averages about 200 gallons per minute for water heights at least 16 feet above the pool floor (i.e., at the top of the spent fuel rack).
- A “moderate leak” in the pool that averages about 1,500 gallons per minute for water heights at least 16 feet above the pool floor.
Reactor operating cycle phases:7
- OCP1: 2-8 days; reactor is being defueled.
- OCP2: 8-25 days; reactor is being refueled.
- OCP3: 25-60 days; reactor in operation.
- OCP4: 60-240 days; reactor in operation.
- OCP5: 240-700 days; reactor in operation.
Fuel configurations in the pool:8
- A “high-density” storage configuration in which hot (i.e., recently discharged from the reactor) spent fuel assemblies are surrounded by four cooler (i.e., less recently discharged from the reactor) fuel assemblies in a 1 × 4 configuration throughout the pool (Figure 7.2).
- A “low-density” storage configuration in which all spent fuel older than 5 years has been removed from the pool.9
- A “mitigation” case in which plant operators are successful in deploying equipment to provide makeup water and spray cooling required by 10 CFR 50.54(hh)(2)10 (see Chapter 2).
- A “no-mitigation” case in which plant operators are not successful in taking these actions.
Some key results of the consequence modeling are shown in Table 7.1 and summarized in the bottom panels of Figure 7.1. Some of the loss-of-coolant scenarios examined in the study resulted in damage to, and the release of, radioactive material from the stored spent fuel. Releases began
7 Phases in a 2-year cycle of removing (defueling) one-third of the reactor core and placing that spent fuel into the pool. The phases are referenced to the start of the defueling cycle.
9 The high-density racking remained in the pool for this scenario; other racking scenarios that would allow for lateral water flow across the fuel racks (i.e., “open racking” scenarios) were not considered.
10 Available mitigation equipment includes portable pumps, hoses, and spray nozzles.
TABLE 7.1 Key Results from the Consequence Analysis in the Spent Fuel Pool Study
|Spent Fuel Pool Fuel Loading||High Dens (Regulatory Baseline)||Low Density (Proposed Alternative)|
|Seismic hazard frequencya (/yr) (PGA of 0.5 to 1.0g)||1.7E-05||1.7E-05|
|50.54(hh)(2) mitigation credited?||Yes||No||Yes||No|
|Conditionalb probability of release (%)||0.036||0.69||0.036||0.69|
|Hydrogen combustion event?||Not Predicted||Possible||Not Predicted||Not Predicted|
|Conditionalc consequences (release frequency-averagedd)|
|Cumulative Cs-137 release at 72 hours (MCi)||0.26||8.8h||0.19g||0.11|
|Measures related to health and safety of individuals|
|Individual early fatality risk||0||0||0||0|
|Individual latent cancer fatality riske Within 10 miles of plant||3.4E-04||4.4E-04||3.4E-04||2.0E-04|
|Measures related to cost-benefit analysis|
|Collective dose (person-Sv) Within 1,000 miles of planti||47,000||350,000||47,000||27,000|
|Land interdictionf (mi2) Within 1,000 miles of planti||230||9,400||230||170|
|Long-term displaced individualsf Within 1,000 miles of planti||120,000||4,100,000||120,000||81,000|
|Spent Fuel Pool Fuel Loading||High Density (1 × 4) (Regulatory Baseline)||Low Density (Proposed Alternative)|
|Consequences per year (release frequency-weightedd)|
|Release frequency (/yr)||6.1E-09||1.2E-07||6.1E-09||1.2E-07|
|Measures related to health and safety of individuals|
|Individual early fatality risk (/yr)||0||0||0||0|
|Individual latent cancer fatality riske Within 10 miles of plant (/yr)||2.1E-12||5.2E-11||2.1E-12||2.4E-11|
|Measures related to cost-benefit analysis|
|Collective dose (person-Sv/yr) Within 1,000 miles of planti||2.9E-04||4.1E-02||2.9E-04||3.2E-03|
|Land interdictionf (mi2/yr) Within 1,000 miles of planti||1.4E-06||1.1E-03||1.4E-06||2.0E-05|
|Long-term displaced individualsf (persons/yr)
Within 1,000 miles of planti
NOTE: The individual early fatality risk estimates and individual latent cancer fatality risk estimates shown in the table were not derived from a risk assessment. They were computed using the postulated earthquake and scenario frequencies shown in the table. PGA = peak ground acceleration.
a Seismic hazard model from Petersen et al. (2008).
b Given that the specified seismic event occurs.
c Given atmospheric release occurs.
d Results from a release are averaged over potential variations in leak size, time since reactor shutdown, population distribution, and weather conditions (as applicable); additionally, “release frequency-weighted” results are multiplied by the release frequency.
e Linear no-threshold and population weighted (i.e., total amount of latent cancer fatalities predicted in a specified area, divided by the population that resides within that area).
f First year post-accident; calculation uses a dose limit of 500 mrem per year, according to Pennsylvania Code, Title 25 § 219.51.
g Mitigation can moderately increase release size; the effect is small compared to the reduction in release frequency.
h Largest releases here are associated with small leaks (although sensitivity results show large releases are possible from moderate leaks). Assuming no complications from other spent fuel pools/reactors or shortage of available equipment/staff, there is a good chance to mitigate the small leak event.
i Kevin Witt, USNRC, written communication, December 22, 2015.
SOURCE: USNRC (2014a, Table 33).
anywhere from several hours to more than 2 days after the postulated earthquake. The largest releases were estimated to result from high-density fuel storage configurations with no mitigation (Figure 7.1). The releases were estimated to be less than 2 percent of the cesium-137 inventory of the stored fuel for medium-leak scenarios, whereas releases were estimated to be one to two orders of magnitude larger for small-leak scenarios with a hydrogen combustion event. Hydrogen combustion was found to be “possible” for high-density pools but “not predicted” for low-density pools.
Operating-cycle phase (OCP) played a critical role in determining the potential for fuel damage and radioactive materials release. The potential for damage is highest immediately after spent fuel is offloaded into the pool (OCP1) because its decay heat is large. The potential for damage decreases through successive operating-cycle phases (OCP2-OCP5). In fact, only in the first three phases (OCP1-OCP3) is the decay heat sufficiently large to lead to fuel damage in the first 72 hours after the earthquake for complete drainage of the pool. These three “early in operating cycle” phases (Figure 7.1) constitute only about 8 percent of the operating cycle of the reactor.
A limited-scope human reliability analysis (HRA) was conducted to estimate the likelihood of successful operator actions to prevent spent fuel damage following the earthquake (USNRC, 2014a, Chapter 8). The analysis estimated that the probability of failure to successfully complete required mitigating actions was highest in OCP1-OCP3, particularly for moderate-leak scenarios. (The probability of failure of successful mitigation was estimated to be 1 in the case of a moderate leak in both OCP1 and OCP3.) The probability of failure of mitigating action ranged from 0.15 to 0.75 for the moderate-leak scenarios in OCP2, depending on whether a station blackout with and without DC power is assumed (USNRC, 2014a, Table 49). The HRA suggests that “no mitigation” is a prudent assumption for moderate-leak scenarios in OCP1-OCP3.11
The USNRC summarized the results of the consequence analysis as follows:
11 Chapter 8 of USNRC (2014a) notes that the human-error probabilities were estimated under the assumption that mitigation equipment was available, there was no simultaneous core damage or primary containment failure that caused inaccessibility of the refueling floor, and there was sufficient staff to deploy the spent fuel pool mitigation strategy. If the earthquake caused damage in multiple reactors and spent fuel pools, such as occurred at the Fukushima Daiichi plant (see Chapter 2 of this report), then these assumptions might not hold. The authors indicate that examination of these additional considerations would require performance of a more comprehensive probabilistic risk assessment and associated HRA. The more comprehensive risk analysis recommended by the present committee (Recommendation 4.11) would be able to address the impact of these types of considerations on the likelihood of successful mitigating actions.
“. . . in a high-density loading configuration, dispersing hotter fuel throughout the pool or successful mitigation generally prevented or reduced the size of potential releases [of radioactive material from stored spent fuel]. Low-density loading reduced the size of potential releases but did not affect the likelihood of a release. When a release is predicted to occur, early and latent fatality risks for individual members of the public do not vary significantly between the scenarios studied because protective actions, including relocation of the public and land interdiction, were modeled to be effective in limiting exposure. The beneficial effects in the reduction of offsite consequences between a high-density loading scenario and a low-density loading scenario are primarily associated with the reduction in the potential extent of land contamination and associated protective actions.” (USNRC, 2014a, p. xxix)
The Spent Fuel Pool Study (USNRC, 2014a, Appendix D) also included a regulatory analysis for the reference plant. This entailed a comparison of the consequences to the public from the postulated releases from the reference plant against the quantitative health objectives (QHOs) in the USNRC’s Safety Goal Policy Statement (Sidebar 7.2) and also the development of a cost-benefit analysis to determine whether expedited transfer is cost beneficial.13
The USNRC estimated that releases from the reference plant would not result in any early fatalities from acute radiation exposures within 1 mile (1.6 kilometers) of the plant boundary (QHO 1 in Sidebar 7.2). The USNRC also estimated the individual latent cancer fatality risk from the accident within a 10-mile (16-kilometer) radius of the plant boundary to be in the range 10–11–10–12 (Table 7.1). This is a miniscule risk—about six orders of magnitude lower than the 2 × 10–6 per year objective (QHO 2 in Sidebar 7.2).
The USNRC summarized the results of these analyses for the reference plant as follows:
“. . . expediting movement of spent fuel from the pool does not provide a substantial safety enhancement for the reference plant. . . . The [US]NRC continues to believe, based on this study and previous studies that high density storage of spent fuel in pools protects public health and safety.” (USNRC, 2014a, p. xxix)
13 That is, to determine whether the benefits of the proposed regulatory action equal or exceed its costs.
A second analysis, the Expedited Transfer Regulatory Analysis, extended the Spent Fuel Pool Study to the fleet of commercial spent fuel pools and independent spent fuel storage installations in the United States. The analysis considered two storage alternatives:
- A regulatory baseline alternative that maintains existing storage requirements, that is, “storage of spent fuel in high-density racks in the SFP [spent fuel pool], a relatively full SFP, and compliance with all current regulatory requirements” (USNRC, 2013, p. 5). As noted in Chapter 6 of this report, current regulations require that fuel be dispersed in a 1 × 4 pattern of high- and low-decay-heat assemblies (see Figure 7.2) following each fuel offload from the reactor not later than 60 days after reactor shutdown, unless such configuration can be shown to be inapplicable or unachievable.
- A proposed alternative that “would require older spent fuel assemblies to be expeditiously moved from SFP storage to dry cask storage beginning in year 2014, to achieve and maintain a low-density loading of spent fuel in the existing high-density racks as a preventive measure” (USNRC, 2013, p. 6). The USNRC identified three benefits of this alternative: less long-lived radionuclide inventory in the spent fuel pool, lower heat load in the pool, and a small increase in the initial water inventory in the pool (because water would displace the fuel assemblies that were moved from the pool to dry cask storage).
The analysis utilized seven groupings of nuclear plants and their spent fuel pools based on “conservative estimates and assumptions to bound the variations in SFP parameters across the fleet . . .” (USNRC, 2013, p. iv and Table 1):
- BWR Mark I and II reactors with nonshared pools (31 reactors, 31 pools),
- BWR Mark III reactors and pressurized water reactors (PWRs) with nonshared pools (49 reactors, 49 pools),
- AP1000 reactors,14
- Reactors with shared spent fuel pools (20 reactors, 10 pools),
- Spent fuel pools located below grade (a subset of the PWR reactors in Group 2),
14 No AP1000 reactors are currently operating in the United States, but four reactors are currently under construction.
- Decommissioned reactors with spent fuel pools (7 reactors, 6 pools), and
- Decommissioned reactors with only dry cask storage (21 reactors, no pools).
The regulatory analysis focused on the first four groups of spent fuel pools. The pools in group 5 were excluded from the analysis because they were below grade and therefore deemed to be “less susceptible to the formation of small or medium leaks due to the absence of open space around the pool liner and concrete structure” (USNRC, 2013, p. 11).
The analysis considered eight types of initiating events that were judged to have the potential to lead to the loss of cooling in spent fuel pools: seismic events, drops of casks and other heavy loads on pool walls, loss of offsite power, internal fire, loss of pool cooling or water inventory, inadvertent aircraft impacts, wind-driven missiles,15 and failures of pneumatic seals on the gates in the spent fuel pools (USNRC, 2013, Table 43). These initiating events could lead to partial or full drainage of the spent fuel pools. If full drainage occurs, then air cooling of the fuel to prevent its runaway oxidation (i.e., a zirconium cladding fire) was assumed to be feasible 60 days following its discharge from the shutdown reactor. If partial drainage occurs in pools with racks that block natural air circulation, then air cooling was assumed not to be feasible (USNRC, 2014a, p. 18).
The expedited transfer regulatory analysis was carried out in two parts:
- The potential safety benefits of expedited transfer were screened using the QHOs in the USNRC’s Safety Goal Policy Statement (Sidebar 7.2).
- A cost-benefit analysis was carried out to determine whether expedited transfer would be cost beneficial.
7.3.1 Safety Goal Screening
The pool-weighted averages of release frequencies of fission products to the environment from the seven types of initiating events described previously range between 7.39 × 10–7 and 2.88 × 10–5 per pool per year without successful mitigation (USNRC, 2013, Table 43). Even though some releases were large, the USNRC concluded that they would not result in any early fatalities from acute radiation exposures within 1 mile (1.6 kilometers) of the plant boundary (QHO 1 in Sidebar 7.2). The USNRC estimated
15 That is, the wind-driven impacts of heavy objects (e.g., storm debris) on external walls of spent fuel pools.
the individual latent cancer fatality risk16 within a 10-mile (16-kilometer) radius of the plant boundary to be 1.52 × 10–8 per year. This is less than 1 percent of the 2 × 10–6 per year objective (QHO 2 in Sidebar 7.2).
The USNRC staff concluded that
“. . . the continued operation of nuclear power plants with high-density loadings in their SFPs [i.e., the regulatory baseline alternative] does not challenge the [US]NRC’s safety goals or related QHOs. Therefore, in the staff’s judgment, a regulatory action to require reducing the inventory of spent fuel in the pools would provide no more than a minor safety improvement.” (USNRC, 2013, p. 10)
A regulatory analysis would normally be terminated once USNRC staff determined that the alternative action (in this case expedited transfer) did not provide a sufficient safety enhancement relative to the Commission’s safety goals and objectives. However, the USNRC staff performed a cost-benefit analysis even though the computed risks were well below the QHOs. The USNRC staff stated that it provided the cost-benefit analysis “to provide the Commission additional information” (USNRC, 2013, p. 2).
7.3.2 Cost-Benefit Analysis
The cost-benefit analysis considered several attributes that could be affected by the two storage alternatives (i.e., the regulatory baseline alternative and proposed alternative) consistent with the USNRC’s regulatory guidance (USNRC, 1997b, 2004) and U.S. Office of Management and Budget (OMB) guidance (1992) (Sidebar 7.3). The attributes are described in Chapter 5 of USNRC (1997b) and are summarized below:
- Public Health17 (accident and routine). In the base case, changes in estimated exposures of the public to radiation caused by changes in accident frequencies or consequences associated with the alternative action measured over a 50-mile (80-kilometer) radius from the plant site boundary are considered. Exposures can result from continued occupation or reoccupation of radioactively contaminated land following a release from a spent fuel pool as well as worker exposures resulting from cleanup and decontamination of offsite land.
16 As noted previously, these risks were not derived from a risk assessment but are based on postulated earthquake and scenario frequencies.
17 The USNRC’s public health attributes focus exclusively on exposures of workers and the public to radiation during routine (i.e., normal operating) and accident conditions. They do not include other public health effects such as psychological effects.
- Occupational Health (accident and routine). Two occupational health attributes are considered: (1) changes in exposures of workers to radiation caused by changes in accident frequencies or consequences associated with the alternative action and (2) routine radiological exposures to workers resulting from dry storage cask loading and handling associated with the alternative action.
- Property. Two property attributes are considered: (1) monetary effects on offsite property resulting from radiological releases associated with the alternative action, including direct (e.g., land, food, and water) and indirect (e.g., tourism) effects and (2) monetary effects on onsite property, including replacement power costs18 and decontamination and refurbishment costs associated with the alternative action.
- Industry. Two industry attributes are considered: (1) net economic effects on nuclear plant licensees resulting from the implementation of any mandated changes associated with the alternative action and (2) net economic effects resulting from recurring operational activities associated with the alternative action.
- USNRC. Two USNRC attributes are considered: (1) net economic effect on the USNRC resulting from implementation of the alternative action and (2) net economic effect from recurring activities (e.g., inspections and enforcement activities) associated with the alternative action.
The USNRC staff developed expected values for each cost and benefit:
“The expected value is the product of the probability of the cost or benefit occurring and the consequences that would occur assuming the event happens. For each alternative, the staff first determines the probabilities and consequences for each cost and benefit, including the year the consequence is incurred. The [US]NRC staff then discounts the consequences in future years to the current year of the regulatory action for purposes of evaluating benefits and costs (i.e., providing a net present value). Finally, the [US]NRC staff sums the costs and the benefits for each alternative and compares them.” (USNRC, 2013, p. 14)
Sensitivity analyses were performed to assess the effects of four factors on the cost-benefit analysis (USNRC, 2013, Table 3):
- Discount rate for determining net present value. 7 percent rate for the base case and 2 percent and 3 percent rates for the sensitivity analysis;
- Averted dose conversion factor. $2,000 per person-rem for the base case and $4,000 per person-rem for the sensitivity analysis;
18 The cost of replacement power is the difference between the cost of electricity from the shutdown nuclear reactor and the next least-costly available generating source. The Spent Fuel Pool Study assumed that only the nuclear plant where the accident occurred would be taken out of operation. This was not the case in Japan, where all Japanese plants were shut down following the Fukushima Daiichi accident.
- Replacement energy costs. $2.3 million for the base case and $729,000 to $57.3 million for the sensitivity analysis; and
- Consequences extending beyond 50 miles. 50 miles for the base case and beyond 50 miles for the sensitivity analysis.
USNRC staff concluded that
“. . . the added costs involved with expedited transfer of spent fuel to dry cask storage to achieve the low-density SFP [spent fuel pool] storage alternative are not warranted in light of the benefits from such expedited transfer. The combination of high estimates for important parameters assumed in some of the sensitivity cases presented in this analysis result in large economic consequences, such that, the calculated benefits from expedited transfer of spent fuel to dry cask storage for those cases outweigh the associated costs. However, even in these cases, there is only a limited safety benefit when using the QHOs and the expected implementation costs would not be warranted. In addition, in the staff’s judgment, the various assumptions made in the analysis of the “base case” result in an overall cost-benefit assessment that is appropriately conservative for a generic regulatory decision and justify using the “base case” as the primary basis for the staff’s recommendation.” (USNRC, 2013, pp. v-vi)
The USNRC staff recommended “that additional studies and further regulatory analyses of this issue [expedited transfer] not be pursued, and that this Tier 3 Japan lessons-learned activity be closed” (USNRC, 2013, transmittal memo, p. 2). A majority of the Commissioners accepted the staff’s recommendation.19 In other words, the Commission decided not to require its licensees to expedite the transfer of spent fuel from pools to dry casks because the Expedited Transfer Regulatory Analysis showed that the costs of such transfer exceeded the benefits.
The USNRC staff put a great deal of thought and effort into the development of the Spent Fuel Pool Study and Expedited Transfer Regulatory Analysis and the explication of their results. The staff also spent a good deal of time presenting the results of these analyses to the committee and responding to follow-up questions. The presentations helped to sharpen the
19 Commission Voting Record on COMSECY-13-0030. Available at http://www.nrc.gov/reading-rm/doc-collections/commission/comm-secy/2013/2013-0030comvtr.pdf; and Memo from the Secretary of the NRC Commission to the Executive Director of Operations, Staff Requirements–COMSECY-13-0030–Staff Evaluation and Recommendation for Japan Lessons-Learned Tier 3 Issue on Expedited Transfer of Spent Fuel. Available at http://www.nrc.gov/reading-rm/doc-collections/commission/comm-secy/2013/2013-0030comsrm.pdf.
committee’s understanding of these USNRC analyses and its assessment of their usefulness for reevaluating Finding 4E in NRC (2006).
The Spent Fuel Pool Study and Expedited Transfer Regulatory Analysis are valuable technical contributions to understanding the consequences of spent fuel pool accidents. However, the USNRC’s analyses are of limited use for assessing spent fuel storage risks20 because
- Spent fuel storage sabotage risks are not considered.
- Dry cask storage risks are not considered.
- The attributes considered in the cost-benefit analysis (Section 7.3.2) are limited by OMB and USNRC guidance and do not include some expected consequences of severe nuclear accidents.
- The analysis employs simplifying bounding assumptions that make it technically difficult to assign confidence intervals to the consequence estimates or make valid risk comparisons.
The present committee’s recommended risk analysis (Recommendation 4.11 in Table 4.1) would provide policy makers with a more complete technical basis for deciding whether earlier movements of spent fuel from pools into dry cask storage would be prudent to reduce the potential consequences of accidents and terrorist attacks on stored spent fuel. This recommended risk analysis should
- Consider accident and sabotage risks for both pool and dry cask storage.
- Consider societal, economic, and health consequences of concern to the public, plant operators, and the USNRC.
- More fully account for uncertainties in scenario probabilities and consequences.
These points are discussed further in the following sections.
7.4.1 Sabotage Risks
The Expedited Transfer Regulatory Analysis considered a large number of initiators for spent fuel pool accidents (see Section 7.3); the analysis did not include initiators for spent fuel pool sabotage. The USNRC staff asserted that it was unnecessary to include sabotage initiators because
20 As noted in Section 7.4.5 of this chapter, USNRC staff characterized the Spent Fuel Pool Study as “a limited-scope consequence assessment that utilizes probabilistic insights” (USNRC, 2014a, p. 6).
“For nuclear power plants, security requirements are established to provide high assurance of adequate protection from radiological sabotage of the nuclear power plant reactor and SFP [spent fuel pool]. The [US]NRC continually monitors threat conditions and, as was done after the September 11, 2001 attacks, makes adjustments, as appropriate in the governing security requirements and in actions to oversee their effective implementation. Based on the staff’s view that security issues are effectively addressed in the existing regulatory program, they are not part of this analysis.” (USNRC, 2013, p. v)
The USNRC staff did not provide the committee with a technical analysis to support its assertion that security requirements are being effectively addressed in its regulatory program.21 Moreover, the staff’s approach for handling sabotage risks is logically inconsistent with how it handled accident risks in the Spent Fuel Pool Study and Expedited Transfer Analysis: In those analyses, staff assumed a nonzero probability that regulatory requirements for mitigation capabilities in 10 CFR 50.54(hh)(2) were not effective (the “no-mitigation” case), even though these requirements are a condition of every nuclear plant operating license in the United States.
The Spent Fuel Pool Study and Expedited Transfer Regulatory Analysis considered beyond-design-basis severe accidents where required mitigation actions failed to be completed successfully. A complete analysis would also include similar considerations for sabotage threats, including the consequences should a design-basis-threat (DBT) event fail to be mitigated, as well as the consequences should beyond-DBT events occur and fail to be mitigated. A complete analysis would consider a broad range of potential threats including insider and cyber threats.
Sabotage initiators can differ from accident initiators in important ways: For example, most accident initiators occur randomly in time22 compared to the operating cycle of a nuclear plant. Sabotage initiating events can be timed with certain phases of a plant’s operating cycle, changing the conditional probabilities of certain attack scenarios as well as their potential consequences (Sidebar 7.4). There may be additional differences between accident and sabotage events with respect to timing, severity of physical damage, and magnitudes of particular consequences, for example radioactive material releases.
21 USNRC staff provided briefings to the committee on security at U.S. nuclear plants and described several security changes that have been implemented since the September 11, 2001, terrorist attacks on the United States; some of these changes involved spent fuel pools. Most of these changes have not been publicized for security reasons.
22 For example, earthquake and storm initiators occur randomly in time, although some storm initiators are seasonal (e.g., hurricanes). Cask-drop initiators, on the other hand, would typically occur prior to reactor refueling outages.
The committee judges that it is not technically justifiable to exclude sabotage risks without the type of technical analysis that is routinely performed for assessing reactor accident risks. Such an analysis would consider both design-basis and beyond-design-basis threats. The likelihoods of these threats could be assessed through elicitation of experts who are knowledgeable about the intents and capabilities of potential saboteurs and who have the appropriate personnel security clearances to access sensitive national security information on terrorist threats. See Chapter 5 for a more detailed discussion of sabotage risks and assessments.
7.4.2 Dry Cask Storage Risks
The Spent Fuel Pool Study (USNRC, 2014a) examined previous studies of dry cask storage safety risks and developed an updated analysis, the results of which are shown in that study’s Table 68. The USNRC staff concluded that
“Comparison of this [Spent Fuel Pool] study to dry cask storage studies (NUREG-1864 [USNRC, 2007b] and supplemental analyses from [this report]), indicates that in some circumstances, the conditional individual LCF [latent cancer fatality] risk within 0 to 10 miles would be similar due primarily to the conservative upper bound estimate of the dry cask release as well as the expected effectiveness of protective actions in response to an SFP release. However, conditional results for metrics such as population dose or condemned or interdicted lands are several orders of magnitude lower for dry cask scenarios than the low end of consequences of pool accidents, due to the substantially smaller amount of released material.” (USNRC, 2014a, p. 254-255)
The Expedited Transfer Regulatory Analysis did not examine the safety or sabotage risks of dry cask storage. The study did not consider sabotage risks and “conservatively ignored” the risks of handling and loading dry casks to calculate the maximum potential benefits and implementation costs (USNRC, 2013, p. 33). The committee judges that a more in-depth examination of the risks associated with dry-cask storage are needed to fully inform an analysis of spent fuel storage risks. The committee encourages the USNRC to develop a risk assessment that explicitly accounts for the risks associated with both pool and dry cask storage.
7.4.3 Expected Consequences
The USNRC uses safety goal screening (Sidebar 7.2) in all of its regulatory analyses that impact nuclear plant licensees to determine whether a substantial safety enhancement exists. The safety goals used in this screening were developed following the 1979 accident at the Three Mile Island (TMI) plant in Pennsylvania. The goals and objectives were influenced substantially by the characteristics of that accident, which involved modest leakage of radioactive material from the reactor’s containment. The possibility of spent fuel pool accidents was recognized but largely unanalyzed.24
 “Condemnation is the permanent relocation of the affected population if decontamination, natural weathering, and radioactive decay cannot adequately reduce contamination levels to habitability limits within 30 years” (USNRC, 2013, p. 103).
24 Only one study (Benjamin et al., 1979) was available at the time the policy statement was developed. The first USNRC regulatory analysis dealing with spent fuel pools (Throm, 1989) was carried out after the policy statement was issued.
The USNRC staff recognized the limitations of the USNRC’s screening goals for analyzing spent fuel pool accidents:
“The QHOs [quantitative health objectives] effectively establish expectations related to the frequency of severe accidents associated with nuclear reactors and the potential for release of radioactive materials from an operating reactor core. . . . Some considerations in comparing SFP [spent fuel pool] risks to the QHOs are that the potential consequences of a SFP accident can exceed those of reactor accidents in terms of the amount of long-lived radioactive material released, the land area affected, and the economic consequences.” (USNRC, 2013, p. 9)
In fact, a spent fuel pool accident can result in large radioactive material releases, extensive land contamination, and large-scale population dislocations. For example, Figures 7.3A-C show the estimated radioactive
material releases, land interdiction, and displaced persons for the reference plant in the Spent Fuel Pool Study (see Table 7.1). Also shown for comparison purposes are the same consequences for the Fukushima Daiichi accident taken from the committee’s phase 1 report (see NRC, 2014, Chapter 6).
These figures illustrate three important points:
- A spent fuel pool accident can result in large releases of radioactive material, extensive land interdiction, and large population displacements.
- Effective mitigation of such accidents can substantially reduce these consequences for some fuel configurations (cf. the bars in the figures for 1 × 4 mitigated and unmitigated scenarios) but can increase consequences for others (cf. the bars in the figures for low-density unmitigated and unmitigated scenarios).25
- Low-density loading of spent fuel in pools can also substantially reduce these consequences and also reduce the need for effective mitigation measures.
The above points are not obvious when consequence estimates are presented only after being weighted by release frequencies. The committee judges that it is important to present the full risk triplet (scenarios, frequencies, and consequences) separately, as well as their product, in cost-benefit analyses.
Note that the Fukushima estimate includes land that is both interdicted26 and likely condemned (see footnote 23 for the definition of condemned land); the Spent Fuel Study (USNRC, 2014a) reports only interdicted land. One of the difficulties with USNRC (2014a) is that, unlike previous studies, the condemned land is not reported. Of the 430 mi2 (1,113 km2) that were
25 This increase in consequences is the result of larger water inventories in the pools from removal of spent fuel assemblies to attain low-density configurations. It takes longer to drain the pools below the baseplates at the bottoms of the fuel racks because the pool contains more water, which delays the establishment of natural air convection through the fuel assemblies to prevent them from reaching runaway-oxidation conditions.
26 Interdiction is defined in footnote 12. Interdicted land is temporarily evacuated during the first year due to the dose level exceeding 500 mrem/year (5 mSv/yr); see footnote f to Table 7.1. We report two values for interdicted land as a result of the Fukushima accident: the mandated evacuation area of 430 mi2 (1,113 km2) in 2013 (NRA, 2013) and the additional area of 690 mi2 (1787 km2) contaminated to an excess dose level of 100 mrem/yr (1 mSv/yr) evaluated from the projected excess dose mapping by IAEA (2015). The interdicted land reported here is much smaller than that reported in NRC (2014, p. 6-4) because that value was based on an estimate made immediately after the accident and before detailed radioactivity surveys were available. The government of Japan has lifted evacuation orders in some regions that have been decontaminated to projected dose levels less than 20 mSv/yr (MOE, 2015, pp. 8-9) and aims to lift more orders as decontamination efforts warrant. In some instances, the projected dose levels in these areas are higher than the government’s proposed 1 mSv/yr long-term cleanup target (see MOE, 2015, p. 28).
evacuated as of May 2013, 124 mi2 (320 km2) was reported as “difficult to return,” which gives an indication of the amount of land that may ultimately be condemned.
A similar point can be made by examining the unweighted results from the Expedited Transfer Regulatory Analysis (USNRC, 2013) for a “sensitivity case” that removes the 50-mile limit for land interdiction and population displacements and raises the value of the averted dose conversion factor from $2,000 per person-rem to $4,000 per person-rem.27 This scenario postulates the evacuation of 3.46 million people from an area of 11,920 mi2, larger than the area of New Jersey (Table 7.2).28
In fact removing the 50-mile limit and raising the value of the averted dose conversion factor to $4,000 per person-rem increased the base-case average estimated benefits of expedited transfer by a factor of 5.9, that is, from about 13 percent of the estimated costs of expedited transfer to about 80 percent.29 Moreover, for the 20 reactors with shared spent fuel pools and the four AP1000 reactors currently under construction (see Section 7.3), the base-case benefits were found to exceed the costs of expedited transfer (i.e., expedited transfer would have been cost beneficial), even though the base case had a limited safety benefit when assessed against the QHOs.
The numbers presented in Table 7.2 are not weighted by frequency. Consequently, they are not expected values and cannot be compared directly with the cost-benefit results in the Expedited Transfer Regulatory Analysis (USNRC, 2013).
The cost-benefit analysis did not consider some other important health consequences of spent fuel pool accidents, in particular social distress. The Fukushima Daiichi accident produced considerable psychological stresses within populations in the Fukushima Prefecture over the past 4 years, even in areas where radiation levels are deemed by regulators to be acceptable for habitation. Radiation anxiety, insomnia, and alcohol misuse were significantly elevated 3 years after the accident (Karz et al., 2014). The incidence of mental health problems and suicidal thoughts also were high among residents forced to live in long-term shelters after the accident (Amagai
27 Current USNRC guidance (USNRC, 2004) specifies the use of $2,000 per person-rem for the averted dose conversion factor. The USNRC is in the process of revising this factor. The $4,000 per person-rem value used in the USNRC’s sensitivity analysis is based on the updated Environmental Protection Agency value of a statistical life and the International Commission on Radiation Protection mortality risk factor coefficient (USNRC, 2013, p. 120). However, the value of the averted dose conversion factor is a matter of Commission policy.
29 These average estimated benefits were obtained using the Group 1-4 pool frequencies in Table 1 and the cost-benefit estimates in Tables 10, 27, 28, 29, and 30 for the 7 percent discount rate case in USNRC (2013).
TABLE 7.2 Sensitivity Scenario of Pool-Averaged Consequences and Benefits for Expedited Transfer
|Cost or Benefita||Sensitivity Study Base-Case Average (range)b|
|Area interdicted (mi2)||11,900 (5,220-18,500)|
|Population interdicted (million)||3.46 (1.34-8.68)|
|Population dose cost ($billion)||435 (84-1,133)|
|Property loss ($billion)||265 (85-668)|
|Total benefits from expedited transfer ($billion)||701 (170-1,802)|
NOTE: These results are averaged over the four spent fuel pool groups, weighted by the number of pools in each group, and have not been weighted by release frequencies.
a Costs and benefits are in 2012 dollars, and no discount factors have been applied. The changes in the costs of potential dry cask storage accidents are not included.
b This sensitivity case eliminates the 50-mile restriction on land contamination and population displacements and uses a $4,000 per person-rem averted dose conversion factor. The pool-weighted release of cesium-137 from a high-density pool accident for the base case in USNRC (2013) is estimated to be 43 MCi. This estimate was obtained by multiplying the cesium-137 inventories in high-density pools (Table 35) by the release fractions from high-density pools (Table 2) and weighting the results by the Group 1-4 pool frequencies in USNRC (2013).
SOURCE: USNRC, written communications, July 15, 2015, and March 8, 2016.
et al., 2014). Complex psychosocial effects were also observed, including discordance within families over perceptions of radiation risk, between families over unequal compensatory treatments, and between evacuees and their host communities (Hasegawa et al., 2015).
These findings are not new. Ten years after the 1979 TMI accident, for example, worries about personal and children’s health were still elevated among women who had lived within 10 miles of the plant prior to the accident (Bromet and Licher-Kelly, 2002), despite the fact that radioactive releases from that accident were small.
Well-documented mental health impacts have also been seen in populations affected by the 1986 Chernobyl accident. Danzer and Danzer (2014) analyzed a sample of adults drawn from the population that was not relocated out of areas contaminated by the accident. They used survey and economic data to estimate the increase in national income that would be needed to compensate for the impact of the accident on life satisfaction: about 6 percent of Ukraine’s gross domestic product. Masunaga et al. (2014) found that even well-educated people born after the Chernobyl accident in areas that were only modestly contaminated had anxiety about their radiation exposures, which has affected their mental health.
It is too soon to know what the long-term mental health impacts will be in the Japanese populations affected by the Fukushima Daiichi accident.
It is clear, however, that mental health impacts are a major if not dominant effect from nuclear accidents involving land contamination and have potentially large attendant costs. Many of these impacts are not readily monetizable at present but could be assessed qualitatively.
Additional research might be needed to develop quantitative metrics for social distress. The development of such metrics might at first glance seem daunting given the myriad ways social distress can be displayed in human populations. On the other hand, relatively simple metrics might be developed based on the underlying drivers for social distress, namely land contamination and population relocations. Such metrics could include, for example, a cost metric based on land areas contaminated above certain thresholds that would require temporary or permanent relocations or remedial actions, as well as a population metric based on the numbers, ages, and employment status of affected people.
7.4.4 Bounding Assumptions
The USNRC staff used numerous bounding assumptions in the Expedited Transfer Regulatory Analysis to “ensure that design, operational, and other site variations among the new and operating reactor fleet were addressed and to generally increase the calculated benefits from the proposed action” (USNRC, 2013, p. 7 in Memorandum to Commissioners). Bounding assumptions were used, for example, for
- Frequency of damage to spent fuel pools from accident initiators that could challenge pool cooling or integrity,
- Loss of AC power following an accident initiator,
- Potential drainage paths from pools,
- Potential for natural-circulation air cooling following drainage, and
- Conditional probability for the failure to successfully mitigate an accident.
Bounding assumptions are commonly used in safety assessments to account for variabilities in model parameters and unanalyzed risks. However, the use of such assumptions can make it difficult to determine whether the results of an analysis are truly bounding. Moreover, it can be difficult or impossible to assign confidence intervals30 to the results when parameter uncertainties are not propagated through the analysis.
30 A confidence interval expresses the degree of uncertainty associated with a model result. It is usually expressed in terms of a probability, for example, a 95 percent probability that the model result falls within the stated uncertainty range.
Sensitivity tests can be used to examine the effects of particular bounding assumptions on the results of an analysis. However, these tests are usually carried out by varying one parameter at a time while holding the other parameters at fixed values. This approach, which was used in the Expedited Transfer Regulatory Analysis, does not account for potential parameter covariability. This approach also makes it difficult to propagate parameter value uncertainties through the analysis to estimate uncertainties in the expected consequences.
It can be difficult to perform valid comparisons of analysis results without reliable uncertainty estimates. To illustrate, consider two spent fuel pool accident scenarios that yield similar best-estimate probability-weighted consequences. The first scenario involves a high-probability, low-consequence event that has a small uncertainty of occurrence. (The uncertainty is small because the event occurs frequently enough to be observed and measured.) The second scenario involves a low-probability, high-consequence event that has a large uncertainty of occurrence. (The uncertainty is large because the event occurs very infrequently and may not have been observed or measured.)
The best-estimate consequences for these two scenarios might have similar numerical values. However, their confidence levels are different—the high-probability event has a high confidence level (i.e., low uncertainty) compared to the low-probability event. Consequently, one would need to know both the best-estimate values and their uncertainty ranges to make useful risk comparisons.
Table 7.2 shows selected accident consequences and cost estimates for a base-case scenario for the Expedited Transfer Regulatory Analysis. Also shown are the ranges of low and high estimates from the analyses. It is immediately apparent that the ranges are large. When weighted by probability, these ranges overlap the cost estimates in the regulatory analysis. This example illustrates the limitations of using best estimates in isolation for making policy decisions.
7.4.5 Concluding Comments
The committee judges that the most effective means to assess the need for expedited transfer would be through a risk assessment that addresses the three questions of the risk triplet (see Chapter 5) and that accounts for uncertainties in both probability and consequence estimates. Such an assessment could include qualitative assessments of currently nonquantifiable consequences such as mental health impacts, or an effort could be made to quantify such impacts. The Spent Fuel Pool Study is “a limited-scope consequence assessment that utilizes probabilistic insights” (USNRC, 2014a, p. 6). It is not a risk assessment. The study is, however, a useful step toward a risk assessment of spent fuel storage arrangements.
The committee’s recommended assessment of spent fuel storage risks would go beyond the Expedited Transfer Regulatory Analysis to include
- Use of established methods to evaluate risk from accidents and sabotage in terms of the risk triplet: scenarios, likelihoods, and consequences. These methods are discussed in the committee’s phase 1 report (NRC, 2014)—see especially Chapter 5 and Appendix I in that report—and in Chapter 5 of the present report. The committee anticipates that the accident and sabotage risk assessments would be carried out separately because they use different analytical approaches. However, there would likely be some commonalities in the event progression and consequence analysis portions of the two assessments.
- The safety and sabotage risks for dry cask storage.
- The range of expected economic and health consequences that would likely result from a severe nuclear accident, as seen most recently in Japan following the Fukushima Daiichi accident. Cost and health impacts associated with land interdiction and population relocation need to accurately reflect the implications of the Japanese experience for U.S. conditions.
The committee-recommended risk assessment would be particularly valuable for analyzing pool storage risks in plants that are in outage or undergoing decommissioning. During plant outages, the reactor core may be moved into the pool to facilitate refueling or maintenance, substantially increasing pool heat loads. During plant decommissioning, the pool may be filled to near capacity and some plant safety systems may be inoperable.
The committee discusses how the USNRC might carry out a sabotage risk assessment in Chapter 5. Although there remain differences of opinion regarding the extent to which risk assessment methods can be meaningfully applied to terrorist threat, clear progress is being made in developing and applying risk assessment methods to terrorist threat. Chapter 5 documents several examples of quantitative assessments of the risks associated with terrorist threats. There are important insights to be gained from more in-depth analysis of these risks, particularly the risks associated with insider cyber threats.
Risk analysis tools that focus on the risk triplet—scenarios, likelihoods, and consequences—can contribute to those insights. The numerical results of such analyses can be used to make relative comparisons, for example, to compare differences in design or operational alternatives within a particular system or facility or between facilities, particularly when the analyses are conducted by the same group of people applying comparable assumptions.
Even if the USNRC staff were to determine that substantially more thorough quantification of sabotage risks is not feasible at this time it could undertake qualitative or partially quantitative analyses. Whichever approach is used, the risk assessment should identify, communicate, and account for the uncertainties in the analyses.
The USNRC staff informed the committee that it is already thinking about how to expand its risk assessment methodologies to include sabotage risks. The committee strongly encourages the staff to continue this important effort.
The decision to expedite the transfer of spent fuel from pools to dry casks is a policy decision for the USNRC, not the task of this study. The committee’s critiques of the Spent Fuel Pool Study and Expedited Transfer Regulatory Analysis are intended to strengthen the quality of any future analyses of spent fuel pool storage risks to support sound decision making by the USNRC and nuclear industry.
The Spent Fuel Pool Study (USNRC, 2014a) did not analyze the effects of “open” or “low-density” racking on the coolability of spent fuel in air. USNRC staff noted that
“Re-racking the pool would represent a significant expense, along with additional worker dose, and was not felt to be the likely regulatory approach taken based on consultation with the Office of Nuclear Reactor Regulation. Much of the benefit of low density racking is achieved by the implementation of a favorable fuel pattern (1 × 4). Additionally, to get the full benefit of low-density racking, BWR fuel would likely need to have the channel boxes removed.” (USNRC, 2014a, p. 23)
“Based on insights from the SFPS [Spent Fuel Pool Study], the [USNRC] staff believes that within the first few months after the fuel came out of the reactor, the decay heat in the freshly unloaded spent fuel is high enough to cause a zirconium fire even in the presence of convective cooling. Therefore, reracking the SFP [Spent Fuel Pool] to install open frame racks even with channel boxes removed to allow potential crossflow, would not necessarily prevent a radiological release during this time.” (USNRC, 2013, p. 31)
In response to a question from the public about whether the results of the Expedited Transfer Regulatory Analysis (USNRC, 2013) would change if open-frame racks were considered, USNRC staff noted the following:
“For the reference plant studied, the BWR fuel assemblies channel boxes would impede crossflow even with open-frame racks. Furthermore, even for the high-density racking, the study showed that without mitigative actions, fuel is estimated to be air-coolable for at least 72 hours for all but roughly 10% of the operating cycle. Based on the insights from the accident progression analyses in the SFPS, within the first few months after the fuel comes out of the reactor, the decay heat in the freshly unloaded spent fuel is high enough to cause a zirconium fire even in the presence of any additional convective cooling once natural circulation is established (see Figures 90 and 93 in the SFPS for the high-density and low-density pool loadings and a moderate leak). Therefore, open frame racks even with channel boxes removed to allow potential crossflow, would not necessarily prevent a radiological release during this time.” (USNRC, 2013, p. 139)
The USNRC’s concerns about the “significant expense” of reracking, the need to remove channel boxes to obtain an appreciable benefit in BWR pools, and increased worker exposures are plausible; however, there is no supporting analysis in USNRC (2013, 2014a) regarding the potential benefits of low-density racks. On the other hand, the argument against considering low-density racking has merit: Within a certain time period after the fuel is removed from a reactor, single isolated fuel assemblies cannot be safely cooled by natural air convection alone. This suggests that a limited benefit would be obtained by going to a low-density rack configuration.
Evaluating the efficacy of open racking requires modeling natural convective cooling of widely spaced assemblies in air, water, or multiphase mixtures, particularly under conditions where oxidation may take place. Flow between widely spaced fuel assemblies will be countercurrent and three-dimensional, driven by buoyancy differences between water or air masses in the pool. The control-volume approach of MELCOR, which was used in the Spent Fuel Pool Study (USNRC, 2014a), is poorly suited for modeling these types of flows. This model treats the large open portions of the pool and building as single volumes with well-defined mixing properties. One needs a computational fluid dynamics (CFD) model that solves the field equations for conservation of mass, momentum, and energy to properly represent the flows that would be expected to occur in pools with low-density racks (see Sidebar 6.1 in Chapter 6). This CFD model needs to be validated with experiments.
Flow in the pool with low-density racks will be turbulent. Given the characteristic dimensions of the pools and fuel racks, significant approximations of unknown fidelity will have to be used to model the fluid dynamics and heat transfer from the fuel rods to pool water. Even greater modeling difficulties will be encountered for partially drained pools because there will be two-phase three-component flow (liquid water, vapor water, air) within the open spaces of the fuel assemblies and a combination of vaporization and buoyancy-driven mixing between assemblies.
Sandia National Laboratories and the USNRC have carried out separate CFD studies on natural convection processes in fully drained pools; there have also been studies on water-filled pools by other researchers (e.g., Boyd, 2000; Chen et al., 2014; Hung et al., 2013; Wagner and Gauntt, 2008). Boyd (2000) discusses the limitations of natural convection for cooling spent fuel and uses a CFD model (FLUENT) to model air convection in a fully drained pool. None of these studies assessed the effects of fuel dispersal in the pool or open versus closed racking.
Benjamin et al. (1979) modeled a loss-of-coolant accident in a spent fuel pool with several rack configurations using a modeling approach similar to MELCOR. One of the configurations considered was an open frame that represented an early design spent fuel pool rack. They note that
“The open frame configuration . . . is more difficult to analyze because of the lack of defined flow paths. On the other hand, it is obviously a very coolable configuration because of the openness of the structure and the large spacings between elements, so that a detailed exact flow calculation was not deemed necessary from a practical viewpoint.” (p. 105)
Benjamin et al. used an abbreviated version of their model (SFUEL) to assess air circulation in an open frame rack, but they cautioned that
“The calculations for the open frame configuration should be viewed as very approximate, with minimum allowable decay times being accurate, perhaps, to within a factor of two.” (p. 106)
Sailor et al. (1987) used a modified version of SFUEL to estimate the risks (likelihoods) of zirconium cladding fires as a function of racking density. They estimated that risks could be reduced by a factor of 5 by switching from high- to low-density racks. This estimate was based on the reduction of minimum decay times before the fuel could be air cooled, and also on the reduction in the likelihood of propagation of a zirconium cladding fire from recently discharged fuel assemblies to older fuel assemblies in the low-density racks compared to high-density racks. However, Sailor et al. (1987) cautioned that “[t]he uncertainties in the risk estimate are large.”
The regulatory analysis for the resolution of Generic Issue 821 (Throm, 1989) was intended to determine whether the use of high-density racks poses an unacceptable risk to the health and safety of the public. The analysis concluded that no regulatory action was needed; that is, the use of high-density storage racks posed an acceptable risk. The technical analysis was based on the studies of Benjamin et al. (1979) and Sailor et al. (1987) and used the factor-of-5 reduction in the likelihood (i.e., the conditional probability of a fire given a drained pool) of a zirconium cladding fire for switching to low-density racks from high-density racks. A cost-benefit analysis analogous to that employed in USNRC (2014a) found that the costs associated with reracking existing pools (and moving older fuel in the pool to dry storage to accommodate reracking) substantially exceeded the benefits in terms of population dose reductions.
The assumptions and methodology used in the regulatory analysis for Generic Issue 82 are similar to those used in USNRC (2014a): A seismic event is considered the most likely initiator of the accident and spent fuel pool damage frequency is taken to be about 2 × 10–6 events per reactor-year. Moreover, USNRC (2014a) reached essentially the same conclusions as the regulatory analysis for the resolution of Generic Issue 82 (Throm,
1 This study was undertaken in response to the issues identified in the Benjamin et al. (1979) study.
1989). However, USNRC (2014a) took more credit for the operating cycle in reducing the risks of zirconium cladding fires.
A more pessimistic view on the uncertainties of modeling spent fuel pool loss-of-coolant accidents was expressed by Collins and Hubbard (2001):
“In its thermal-hydraulic analysis . . . the staff concluded that it was not feasible, without numerous constraints, to establish a generic decay heat level (and therefore a decay time) beyond which a zirconium fire is physically impossible. Heat removal is very sensitive to these additional constraints, which involve factors such as fuel assembly geometry and SFP rack configuration. However, fuel assembly geometry and rack configuration are plant specific, and both are subject to unpredictable changes after an earthquake or cask drop that drains the pool. Therefore, since a non-negligible decay heat source lasts many years and since configurations ensuring sufficient air flow for cooling cannot be assured, the possibility of reaching the zirconium ignition temperature cannot be precluded on a generic basis.” (p. 5-2)
The older studies of Benjamin et al. (1979) and Sailor et al. (1987) simulated open racking configurations and showed the potential for increased air coolability for those configurations. More recent analyses by Sandia National Laboratories (see Chapter 6) and the Spent Fuel Pool Study (USNRC, 2014a) have not carried out simulations of these configurations. There have been substantial advances over the past decade in understanding the complex phenomena involved in the prediction of critical conditions for fuel assembly ignition. Consequently, these older studies will need to be revisited as part of any future consideration of reracking as a spent fuel pool management strategy. As discussed in this Appendix, the modeling approach and software (MELCOR) used in USNRC (2014a) have limitations that will need to be addressed as part of any study of reracking. Accurate modeling of natural convective cooling of widely spaced fuel assemblies will require careful examination of the fundamental assumptions in the modeling and validation against test data.
SOURCE: Middle column, USNRC (2014a, Fig ES-2); right-hand column, committee generated.