The Final Report is nearly 50 pages with several illustrative tables and charts, and appendices attached with supporting information. The Panel's review of the 2008 cost estimates for Diablo Canyon and SONGS did not identify any significant issues given the assumptions and knowledge at the time. The Panel identified a substantial error in the adopted Palo Verde cost estimate but there was no associated revenue requirement approved. The Final Report's substantive findings and recommendations are described below.
The Panel emphasized that any conclusions about future decommissioning costs "involve a significant amount of informed speculation about events that will only be fully understood in the future...and which may resemble historical events to a greater or lesser degree as circumstances change."3 In other words, the Commission's interest in having the estimate be consistent with actual future decommissioning performance is limited by the reality that there are many approaches to decommissioning, even during the same time frame, and circumstances and technologies will likely change before the actual decommissioning of SONGS, Diablo Canyon, and Palo Verde.
The cost estimates submitted in each NDCTP should "consider, estimate, and document" various important cost drivers that the Panel concluded will impact the cost of decommissioning. These drivers include:4
· Decommissioning strategy;
· Timing of plant shutdown;
· Desired "end-state" (projected federal, state, local and owner requirements);
· Quantities of waste generated;
· Schedule of performance;
· Cost of waste disposition (packaging, transportation, and disposal);
· Expected spent fuel management requirements;
· Cost of labor;
· Program management;
· Security;
· Indirect costs such as taxes and insurance;
· Contingency; and
· Regulatory agency fees
The Panel examined experiences from past decommissioning of nuclear units inside the U.S. which they thought could be valuable in developing future cost estimates. However, the Panel found substantial barriers to comparing prior decommissioning experiences because reported estimates and costs from around the country are not always public, or even similar in what activities are included and the information disclosed.
During the past twenty years, owners of six commercial Pressurized Water Reactors (PWR) have completed most of the work to decommission the units, including removal and disposal of radiological waste, management of spent nuclear fuel, demolition of most structures, removal of hazardous, non-radiological materials, and some site restoration. SONGS unit 1 was the only one co-located with other units. Two more PWRs will be decommissioned in the next ten years.
The Panel attempted to provide a summary of the U.S. experience in decommissioning PWRs by identifying "Estimated Costs" and "Reported Costs" along with other pertinent information about the license, size, and status of decommissioning.5 The plants compared, with their locations, are Maine Yankee (ME), Connecticut Yankee (CN), Yankee Rowe (MA), Trojan (OR), Rancho Seco (CA), and SONGS 1 (CA). With the exception of Rancho Seco, all actual costs appear to exceed estimated costs by varying margins, e.g., Connecticut Yankee exceeded estimates by 82% and SONGS 1 by 32.5%. However, the Panel presented these results more as indications than actual factual findings due to the challenges of comparison.
As noted above, there were numerous problems in obtaining accurate and comparable figures. For example, some information is withheld as proprietary, public records can be incomplete, and estimates may not include identical activities or may even omit key elements such as site restoration. Another variable is that "Reported Costs" meant to reflect actual costs, may include both costs incurred and projected costs going forward through final steps after all Spent Nuclear Fuel (SNF) is removed.
In addition, early pre-decommissioning estimates did not anticipate actual circumstances, particularly because all six facilities compared shut down early without a transition plan, and the actual contracting differed from what had been assumed. These estimates did not account for all SNF costs of non-performance (including transfer to dry storage) or higher waste volumes arising from discovered leaks and contamination, and made widely different assumptions about availability and cost of radiological waste disposal and site restoration standards. Furthermore, estimates developed prior to September 11, 2001 did not include significantly revised security measures now in effect.
Historical experience in the U.S. has provided no consensus on the best way to decommission a nuclear plant because every site has different challenges, technology is improving, and new ideas are borne from experience. As of the date of this decision, PG&E's Humboldt Bay Power Plant Unit 3 (HBPP) is the only California nuclear plant currently undergoing decommissioning, after years of being "mothballed" (SAFSTOR).6 PG&E is managing its own decommissioning project at HBPP, as SCE did for SONGS 1.
Other owners have turned to outside contractors with mixed results. A promising example is the proposed decommissioning of Zion units 1 and 2 in Illinois. The units sat idle for many years because the owner, now Exelon, lacked sufficient funding. Recently, the company settled on a new approach it claims will make decommissioning faster, simpler and less expensive. Instead of hiring a contractor, it has turned the job and the reactors over to a nuclear demolition company7 that owns a nuclear waste disposal site, thus drastically reducing the cost of waste disposal.8 Although this option is untested and not suitable for inclusion in current cost estimates, it may prove to be cost-effective, especially for places like California where waste disposal is a serious problem. Both HBPP and Zion will provide cost estimators with additional experience to be considered in the future.
The primary lesson the Panel drew from past experiences was for the plant operators who make modifications or replace components during operations. The Panel opined that these experiences could be useful opportunities to discover unknown contamination and better determine future waste volumes and related packaging, transportation, and disposal costs.
The Panel performed a comparison of the site-specific decommissioning cost estimates prepared in 2008 for the 2009 NDCTP. The SCE estimate for the SONGS units, prepared by ABZ, was $3.659 billion ($2008).9 The PG&E estimate for the Diablo Canyon units, prepared by TLG, was $1.828 billion ($2008).10 The Panel also obtained supplemental information from the utilities, including responses to specific questions, plant site visits, and interviews of expert utility personnel at each site.
Eight items were identified that account for 99.4% of the cost difference between SONGS 2 and 3, and Diablo Canyon 1 and 2. By a large margin, the assumed site condition at the end of decommissioning is the primary difference between the estimates, accounting for about $1.3 billion (72.9%). The scope of the projected work is very different because Diablo Canyon is on land owned by PG&E and the estimate assumes that all structures above three feet below grade are removed.11 SONGS is located on land leased from the Navy and SCE is generally required, according to the lease, to remove any and all material added to the site during operations.12 SCE also has a separate lease with the State Lands Commission (SLC) for its offshore water conduits which may need to be dismantled in decommissioning.
The cost of the estimated waste disposal difference was not separately identified in the 2008 SONGS cost estimate, but the Panel developed the estimate of $1.3 billion, reflecting a projected waste volume of 2.8 billion pounds at SONGS compared to 1.0 billion pounds at Diablo Canyon. The Panel determined it would be useful to the Commission in future NDCTPs, for SCE to provide a specific cost estimate for removal of material exceeding three feet below grade level, the scope applied to Diablo Canyon.
The second largest difference was that SCE included $178 million in personnel severance costs necessary to comply with the California Nuclear Facility Decommissioning Act of 1985. PG&E did not include such costs in its 2008 estimate but late in Phase 1 proposed inclusion of $135 million in similar costs.13 The Panel determined that to the extent state law requires these severance payments, the amounts could be calculated using similar methodology but reflect the differing staffing needs of each site.
The third largest cost difference between the estimates relates to NRC-required security costs14 during decommissioning, where SONGS costs are about $147 million higher. The security concerns are very different between the sites. SONGS is located next to a public beach and interstate highway, in contrast to the Diablo Canyon units which sit on a large parcel of PG&E land. The security requirements have increased significantly since 2001, and costs include management staff, training, supervisors, guards, and licensing compliance. The Panel viewed the input of site security personnel as useful to cost estimators to better reflect the projected security needs during decommissioning.
An additional $140 million difference relates to the estimated cost to remove and dispose of the reactor vessel, its internals, and other large components. There is limited industry experience in this category, and the estimators used different methods. ABZ assumed that scope and cost of this activity could be determined based on prior experience. TLG assumed that additional experience gained would lead to the activities becoming routine and lower cost. The Panel found both methods to be reasonable. The other differences between cost estimates were small and included costs for disposition of radiological and non-radiological waste at final shutdown, insurance costs, dry fuel storage, and survey costs.
Arizona Power Service (APS), the operating agent for the Palo Verde units, retained TLG to prepare its decommissioning cost study. According to the TLG study, SCE's share of costs, as 15.8% owner, was $324.4 million. However, for the 2009 NDCTP, SCE concluded some of the assumptions made by TLG were inconsistent with SCE's experience and risk tolerance. Therefore, SCE made substantial adjustments to the TLG cost study and then applied a 25% contingency factor to all costs. The net result was the estimate more than doubled to $708.7 million for SCE's share.
The Panel examined each of the adjustments and found that the largest one, for dismantling and disposal of waste, mistakenly included a significant volume of Class A waste which TLG projected would instead be handled through processing and conditioning, rather than packaging and shipping for disposal. This error had been carried forward in SCE's reviews of Palo Verde decommissioning cost estimates for many years, resulting in an incorrect adjustment to the 2008 Palo Verde estimate of $330.1 million. The Panel advised this error be corrected and SCE, which helped the Panel determine the error, has agreed.15
The other adjustments made by SCE related to the cost of large component removal, the contingency factor, and SNF dry storage. The Panel found these adjustments to be reasonable, assuming adequate documentation and explanation is supplied by the utilities.
The Panel selected four U.S. nuclear power plants of similar reactor type, reactor vendor, and construction period for comparison of decommissioning cost estimates with those for California's nuclear units. Other factors considered included similar sized plants, geographic diversity, and the availability of somewhat current, public data on decommissioning costs. The selected plants are Indian Point Unit 2 (IP),16 St. Lucie Units 1 and 2 (St. Lucie),17 South Texas Project Units 1 and 2 (STP), and Vogtle Units 1 and 2 (Vogtle).18 Similarities and differences between the plants are described in the Final Report and summary information for these plants and SONGS, Diablo Canyon, and Palo Verde are listed in Table 5.1.19
The Panel carefully scrutinized the cost estimates to extract and compare data from cost estimates developed by different estimators, in part by converting all data to $2008 and assuming an annual 3% inflation rate. All of the estimates were adjusted to be equivalent to the cost of decommissioning a two-unit site and to account for varying contingency factors. The Final Report focused on four areas of quantitative comparison, (1) total cost, (2) annual cash flows, (3) Low Level Radioactive Waste (LLRW) burial volumes, and (4) staffing, as well as examining the important topic of scheduling.
The Panel carefully compared available total cost studies because of differences in definition, format, and data organization. Chart 5.1, which provides a comparison of all sites, shows that all have estimated decommissioning costs between $1- $2 billion, except for SONGS and Palo Verde, as previously adjusted by SCE, which both exceed $3 billion.20 However, after the Panel corrected the Palo Verde estimate, as discussed above, the revised estimate was just under $2 billion. The SONGS estimate remains the highest among the reviewed sites, in excess of $2 billion, even after the estimated difference for site restoration pursuant to the Navy lease is removed. The Panel said that the SONGS estimate remains higher due, in part, to the challenges of its small site and a higher contingency factor than the non-California plants.21
The Panel created several charts to illustrate the projected annual cash flows to show comparisons between all of the selected plants for total expenditures and in each phase of decommissioning: pre-decommissioning, major decommissioning, wet fuel storage, dry fuel storage, and ISFSI (dry storage) decommissioning.22 There are two areas where the total projected annual cash flows are significantly different for SONGS and Diablo Canyon from other units examined. These are the extended site restoration activities for SONGS and assumptions about how long SNF will remain in wet storage before moving to dry storage, or "ISFSI." The result is substantial expenditures for SONGS that extend from years 3 to 12 for site work, and to both SONGS and Diablo Canyon in years 12 to 17 due to longer wet fuel storage than other sites.
The pre-decommissioning cash flows are similar between comparable plants. Cash flows during major decommissioning, where the bulk of the work is done and most expenses are incurred, are similar except for SONGS. For all units there is a rapid rise in costs that peaks about the third year of decommissioning when most waste is packaged and removed from the sites. SONGS costs are higher and go on years longer due to its expected additional waste volumes arising from the Navy lease requirements for site restoration.
The primary difference between sites during the wet fuel storage phase is that SONGS and Diablo Canyon assume a 12-year storage period for spent nuclear fuel, rather than five-year period assumed by the non-California sites. Costs rise again when the spent fuel is moved out into dry storage and site restoration of the empty wet fuel pool occurs. The dry fuel storage phase is generally low cost mostly due to a major reduction in staffing. Removal of the ISFSI and final site restoration are also generally low cost. When the estimated $50 million for SCE to dismantle its offshore water conduits is excluded, all sites are comparable.23
The totals of LLR Class A volumes vary significantly between sites. Chart 5.8 illustrates this comparison and breaks it down into waste expected to be processed for reuse and waste expected to be buried. Except for SONGS, all units anticipate significant processing and conditioning of waste. SONGS has the second highest volume at 1.5 million cubic feet (surpassed only by Indian Point with 2 million cubic feet) and projects 100% of its waste will be buried.24 The Panel notes that a utility may consider processing and conditioning waste, rather than burial, based on the cost of labor for sorting, decontaminating, and "otherwise reducing the volume of waste" for burial. Some waste is easier to process, and labor costs which escalate faster than burial costs are a motivation for the utilities to achieve a lower unit cost by adapting to more cost efficient methods.
Staffing costs during decommissioning are an important cost driver. The Panel broke labor costs into two main categories: craft and utility/contractor. Chart 5.9 illustrates estimated man-hours by category for decommissioning each site.25 Most use total labor in the range of 10,000,000 man-hours, but SONGS estimates more than twice as much at 25,000,000 total man-hours. The high estimate is primarily the result of the large volumes of waste due to full site restoration requirements of the Navy and State Lands Commission leases.
The Panel found that in all cases, the majority of labor was for utility/contractor hours, particularly for SONGS and Indian Point. Decommissioning cost estimates divide tasks into "activity" costs driven by the scope of the task, while "period" costs derive from the time needed to complete the task. Craft man-hours primarily support activity costs, and utility/contractor hours are generally for period costs. Thus, the efficiency of the utility/contractor man-hours and completion of tasks on schedule will significantly affect period costs.
Given the prominent role of "period" costs, keeping to schedule becomes a very important factor in whether actual decommissioning costs will be similar to estimated costs. Furthermore, it matters where the delays occur. For example, a delay in the major decommissioning phase has more serious impacts on actual costs than a delay in a later phase where less work remains to be done.
The Panel determined that the cash flows for each site, except SONGS, reflect similar schedules for the major decommissioning phase. Again, the explanation is the unique site restoration requirements for the leased lands at SONGS.
The Panel was asked to evaluate emerging radiological contamination issues and whether the utilities could take action while in operation to minimize the effects, including the future costs of waste removal. They found that every domestic decommissioning project has had to deal with unexpected contamination which increased actual costs.26
For this report, the Panel identified two specific contaminants of concern: tritium and carbon 14. Tritium is a radioactive isotope of hydrogen that is created both naturally in the atmosphere and also as a byproduct of nuclear power, normally in reactor coolant and spent fuel pool water. Tritium combines with oxygen to form titrated water which diffuses easily into concrete and soil. The NRC requires a licensee to consider tritium contamination but it can generally be left onsite under NRC standards.27 The issue for cost estimators is whether lower dose limits set by the EPA or state agencies will require additional waste removal.
Carbon 14 is also naturally occurring and is produced inside the reactor as one of the primary gaseous effluents from PWRs. Its presence at decommissioning can become a cost factor. For example, at Yankee Atomic, carbon 14 contamination led the licensee to ship and bury tons of concrete rather than use it as onsite fill resulting in a direct increase to decommissioning costs.
The utilities monitor both tritium and carbon 14 during operations but only limited data is collected about the distribution of these agents into concrete and soil. The Panel advocated for collection of additional data to facilitate a more accurate estimate of related decommissioning costs, "where the opportunity is provided through routine maintenance and plant modifications."28
As noted before, the estimated costs for decommissioning may not mirror actual results implemented many years later under new and unexpected circumstances. Nonetheless, the Panel identified several actions that the utilities could take into account in future decommissioning cost estimates to reduce costs or minimize uncertainty.
The following are items that have the potential to reduce future decommissioning costs, along with the Panel's comments:
· Operate the plant to minimize end-of-life contamination levels and uncontained environmental releases; this is already happening and NRC has recently promulgated new requirements on this topic at 10 CFR 20.1406(c).
· Consider mitigating actions if LLRW disposal costs are expected to escalate at a rate substantially higher than labor or general inflation; NRC has said with approval that if rates for disposal of LLRW grow faster than labor or inflation rates, then it may be economical to use labor and equipment to reduce the volume of disposable waste.29
· Incorporate economies of scale when decommissioning a two-unit site; certain activity costs can be shared, e.g., procedure development, waste packaging analysis, some program management functions.
· Adjust "actual experience" costs to be consistent with the cost estimate contingency model; when the estimator incorporates these costs into the estimate, the estimator should also determine whether it is appropriate to remove some or all of the contingency cost.
· Consider the potential impact of a 20-year license renewal; if decommissioning is deferred for 20 years, it could raise or lower cost estimates, depending on the timing of decommissioning, increased contamination, and spent fuel storage costs linked to DOE performance. If after-tax trust fund earnings lead to balances that exceed inflation rates, license renewal could result in lower net present value of costs.
· Consider the potential impact on recovery of spent fuel damages from the utilities' legal action against DOE for non-performance of its duty to accept SNF; some of the post-shutdown SNF costs will likely be recovered from DOE, however, how much and when is uncertain. NRC requires licensees to manage SNF, regardless of DOE performance for five years without reliance on DOE recovery.
The Panel identified other items that may have the potential to reduce uncertainty with future decommissioning costs, which might lead to reduced funding requirements:
· Upon opportunity, identify the extent of radionuclides that could impact NRC license termination and state or other standards; this is consistent with new NRC regulations 10 CFR 20-1501(a) and (b).
· Solicit and include input from site security in developing a site-specific decommissioning security model; this is a substantial cost both for direct costs for security department personnel and equipment, and indirect costs for implementation including contractor screening. (The Panel did not suggest any protected security related information be publicly disclosed.)
· Minimize stored LLRW to be disposed of during decommissioning; disposing of waste generated during operations reduces uncertainty of disposal costs in the future where costs are expected to escalate.
When a cost estimate is made for a future activity, it inherently includes various risks. The Panel identified and discussed four areas of risk that have resulted in added costs to recent decommissioning projects. These are performance risk, scope risk, regulatory risk, and financial risk. ABZ and TLG estimates typically address only performance risk, and assign contingency on a line-by-line basis (i.e., a lower contingency for a routine activity).30
Performance risk is associated with completing the defined scope of activities within the allotted schedule and cost, and is expected to be fully spent by the end of the project. Types of conditions that reflect this risk include weather delays, equipment and tool breakage, waste packaging problems, and personnel turnover. The cost estimates from ABZ and TLG typically assign a performance risk contingency between 17% and 22%.31 In the 2009 NDCTP, the Commission adopted an overall 25% contingency value for both SONGS and Diablo Canyon, based in part on what the NRC requires for funding assurances.32
Scope risk is the risk that the defined scope did not consider all required activities, leading to unforeseen additional work. For example, an estimator may rely on plant drawings that do not reflect the actual configuration of the site, such as underground pipes and cables, or the distribution of contamination. Previously unknown or unidentified contamination can also lead to unexpected costs and delays, as it did at SONGS and Yankee Atomic where previously unknown contamination vastly increased cleanup activities.33
Regulatory risk relates to the risk that estimators cannot accurately predict the specific rules that will govern future decommissioning work. Estimators determine costs based on the current regulatory scheme. The last type of risk, financial, is the risk that one or more of the financial assumptions used for the funding analysis was wrong due to the estimator's inability to forecast financial variables e.g., escalation, rates of return for the trust funds, and taxes.
The Panel identified and supported three methods for the utilities to mitigate these risks, including (1) use of performance risk contingency, (2) conservative assumptions for cost escalation and rates of return, and (3) periodic reassessment of costs.
SONGS and Diablo Canyon are both PWRs located in California and have similar license expiration dates. Despite separate ownership, the Panel concluded there are assumptions that should be reasonably consistent between SONGS and Diablo Canyon. These cost drivers are identified below:
· DOE Spent Fuel performance - significant assumptions including the projected start date for DOE performance and the annual acceptance rate;
· Waste management options and costs - the assumed estimated unit rates because it is reasonable to assume both sites will have access to the same disposal facilities;
· Severance Costs - state-mandated labor termination costs, although there may be owner-specific assumptions and different staffing needs;
· Federal and State "End State" acceptance criteria - there are multiple layers of criteria for site restoration some of which would be consistent, e.g., requirements by the NRC, EPA, and California state agencies; in contrast, local and owner-specific requirements would be unique, e.g., Navy lease; and
· Share existing "actual" decommissioning-related activity performance and cost data - utilities should consider sharing performance and cost data, particularly where relevant to future decommissioning costs, e.g., waste packaging densities, production rates for dismantling components, manhours for packaging large components, use of alternate waste disposal sites. (The Panel did not advocate public disclosure of proprietary, confidential, or commercially sensitive materials.)
The Panel was asked to develop a common format for decommissioning cost estimates that would result in greater transparency and comparability. However, the fact that cost estimators use proprietary and substantially different decommissioning cost models to develop their estimates, combined with the unique aspects of decommissioning SONGS, make a common cost model impractical. Instead, the Panel concluded the objectives could be achieved through a common summary format.
The Panel identified a list of information it deemed the most pertinent and useful, divided into estimate assumptions and estimate results. Although the Panel created a sample format (included as Appendix E to the Final Report), they encouraged the utilities to develop together an actual common summary format to be used in the next NDCTP. The Panel also suggested that cost estimate information for SONGS without the lease requirements should be included in the common summary.
3 Report on Nuclear Decommissioning (February 2011) prepared by Geoffrey Griffiths, Nicholas Capik, and Bruce Lacy (Final Report).
4 Final Report at 3.
5 Table 3.1, Final Report at 8.
6 HBPP was not included in the Panel's analysis because it is a Boiling Water Reactor, not a PWR, and it is already in the process of decommissioning. Therefore, it was not easily comparable to the units examined at San Onofre, Diablo Canyon, and Palo Verde.
7 Exelon transferred its Part 50 operating license to the contractor until after completion of decommissioning.
8 Transcript of Workshop (February 2, 2011) at 16-17, 19-20, 31, 32.
9 Final Report at 12.
10 Ibid.
11 Id. at 14, fn19. In its cost estimate, PG&E assumes compliance with NRC requirements, i.e., sufficient radioactive contamination is removed to show compliance with NRC standards, and that remaining clean material remains onsite and may be used as fill material.
12 Ordering Paragraph 10 of D.10-07-047 requires the Commission's Executive Director, on behalf of the Commission, to make a formal written request to the Navy to clarify the applicable site restoration and remediation standards that will be required to terminate the site lease.
13 D.10-07-047 acknowledged that labor termination costs overlooked by PG&E may be reasonable costs at 35-36.
14 10 CFR 73.
15 The erroneous adjustment did not lead to a revenue requirement for these units in Phase 1.
16 Indian Point units are located in New York.
17 St. Lucie units are located in Florida.
18 Vogtle units are located in Georgia.
19 Final Report at 19-20.
20 Final Report at 22.
21 Id. at 23.
22 Charts 5.2 through 5.6, Final Report at 24-30.
23 SCE's lease with the State Lands Commission (SLC) requires removal of the water cooling system and is included in the SONGS cost estimate. However, SCE has said that they are in discussions with SLC to try to negotiate leaving the water conduits in place.
24 Final Report at 31.
25 Id at 32.
26 Final Report at 34, e.g., fuel at the Connecticut Yankee plant, from an early fuel failure, had contaminated portions of the plant systems, making removal more difficult; at Yankee Rowe, tritium and carbon 14 contamination of concrete converted plans to treat and reuse the concrete on site to disposal as LLRW; during the SONGS steam generator replacement, metal rebar used in the concrete containment walls was found to be radiologically activated.
27 Final Report at 34, fn 29. Tritium is one of the least dangerous radionuclides because it leaves the body relatively quickly.
28 Id. at 35.
29 NUREG-1307, Rev. 14, Section A.3., "Report on Waste Burial Charges: Changes in Decommissioning Waste Disposal Costs at Low Level Waste Burial Facilities" (November 2010).
30 Final Report at 40.
31 Ibid.
32 10 CFR 50. 75.
33 In Phase 1, PG&E argued for a higher contingency factor for Diablo Canyon than SONGS based on the assumption that the full site restoration requirements for SONGS meant a lower risk of scope changes.