Burnup Credit Criticality Safety Case for AGR Spent Fuel Storage

James Ryan, Sellafield Ltd.

Albrecht Kyrieleis, Jacobs

Jennifer Bateman, Sellafield Ltd.

Dominic Winstanley, Sellafield Ltd.

21 September 2023

Introduction

• ASTOP – Operational Summary & Challenges
• Burn Up Credit
• Loading Curve & Derivation
• Fuel Depletion
• Criticality Modelling
• Uncertainties/Sensitivities
• Fuel Misload Analyses
• Cross-Checking
• Post-Irradiation Verification
• Regulatory Engagement & Learning From Experience (LFE)

AGR STOrage Programme (ASTOP) (1)

• Previous AGR spent fuel strategy (reprocessing via THORP) ceased in 2018

​

• UK shift to storage strategy (interim surface storage, awaiting GDF)

​

• >2000 tonnes of spent AGR fuel – receipts continuing over the next decade (~5000+ tonnes in total expected)

AGR STOrage Programme (ASTOP) (2)

• Single-pond storage strategy

​

• Increased fuel storage density required

​

• 20c skips -> 63c racks

​

• Increased fuel density leads to additional criticality challenges

AGR STOrage Programme (ASTOP) (3)

AGR STOrage Programme (ASTOP) (4)

• Design update to remove boronated steel from structure
• Manufacturing challenges
• Cost
• Overall optioneering and ALARP basis

​

• Challenge to fresh fuel criticality case

​

• Burn-up credit required during fault scenarios

​

Burn-Up Credit (1)

• Default assumption in majority of SL criticality cases is fresh fuel
• U235 & U238 isotopes only

​

• Spent fuel irradiation within reactor depletes fissile fuel content (U235), whilst breeding in a range of other isotopes (fissile, fissionable, poisons, etc.)

​

• Overall, depletion of fissile isotopes and build-up of poisons reduces reactivity of AGR fuels (other fuels may be more complicated with integral burnable poisons, etc.)

​

• Burn-Up Credit allows for formalisation of this reactivity reduction within criticality safety assessments

Burn-Up Credit (2)

• UK AGR fuel manufactured up to 3.78 w/o
• Historically lower, up to 2.7 w/o

​

​

• Fuel burnups targeted around 33,000 MWd/teU
• Historically lower, target ~18,000 MWd/teU

​

​

• Potentially significant reduction in k-effective

Burn-Up Credit (3)

• UK guidance recognises several methods of Burn-Up Credit

​

• Main ones are
• ‘Actinide-only’ – Credits depletion of U235, along with generation of Pu isotopes (Pu238/Pu239/Pu240/Pu241/Pu242)

​

• ‘Actinide and Fission Product’ – Extension of actinide only to include internationally recognised ‘Top 15’ fission products

​

​

• SL previous experience with ‘Actinide-only’ method
• Simplified further in this ASTOP assessment

Burn-Up Credit (4)

• Approach to burn-up credit considered in 3 stages

​

• Fuel depletion calculations
• Neutron multiplication calculations
• Sensitivity and uncertainty analyses

​

​

• These are used to derive an appropriate ‘loading curve’ for operations within the facility

Loading Curve (1)

• Loading curve defines compliant vs non-compliant fuel; typically as a function of initial enrichment and required minimum irradiation
• E.g., fuel which is greater than 3.5 w/o initial enrichment, must also have at least 10,000 MWd/teU average irradiation to remain safely sub-critical during operations

​

​

• Pre-existing calculations demonstrating safe handling & storage for fuel @ 3.0 w/o U235/U (U235 & U238 only) for normal operations & identified fault scenarios

​

Loading Curve (2) – Fuel Depletion

• Fuel depletion calculations; provide estimated inventory of U/Pu isotopes for k-eff calculation comparison

​

• Simple calculations – FISPIN/FIS2000
• Define fuel type, power rating, cooling time, enrichment, irradiation to produce isotopic inventory estimates

Loading Curve (3) – Neutron Multiplication

• Neutron multiplication calculations to determine appropriate bounding cases

​

​

​

​

• Neutron multiplication calculations to determine appropriate bounding cases

​

​

​

• IE/Burnup combinations which have a lower peak k-eff (and are otherwise statistically insignificant in difference) are considered to be sufficient as a minimum irradiation value

​

​

​

Loading Curve (4) – Initial Estimate

Loading Curve (5) – Uncertainties

• With perfect data and perfect modelling, initial estimate would be sufficient

​

​

• Need to account for a number of uncertainties and physical variations

​

​

• Understanding sensitivities to range of factors also important

Uncertainties / Sensitivities (1)

• Physical variations (e.g., neutron flux variations across a reactor)

​

​

• Depletion/neutron multiplication code uncertainties

​

​

• Imperfect data (declared irradiations based on calculations and measurements with inherent uncertainties)

​

Uncertainties / Sensitivities (2) – Physical Variations

• Number of physical variations which may affect transmutation of spent fuel within a reactor during lifetime

​

​

• Spatial variation of neutron flux across fuel element
• Spent fuel data declared at elemental level

​

​

• Moderation variation within reactor
• Typically a concern for water moderator reactors, especially BWR

​

Uncertainties / Sensitivities (3) – Axial flux variation

• Spatial variations in neutron flux – affects levels of irradiation experienced by element dependent on reactor positioning

​

• AGR fuel stringers (7/8 elements) ~ 7-8 m in height, neutron flux at base/top element (element 1/7/8) significantly different from central (elements 4/5)

​

• Average irradiation for element 1 vs element 4 differs – end elements (1/7/8) also exhibit significant intra-element variation (much less effect for elements 2-6/7)

Uncertainties / Sensitivities (3) – Axial flux variation

Uncertainties / Sensitivities (3) – Axial flux variation

Uncertainties / Sensitivities (4) – Radial flux variation

• Smaller length scale (~30 cm diameter vs ~ 1m length)
• Expect impact to be smaller
• Neutron radial flux variation typically flat across bulk of reactor

​

• Larger impact on radial flux near reactor boundary, or ‘black spots’ (large control rod insertions, strong neutron interacting structural materials, etc.)

​

• Estimated 10-15% variance in flux across assembly by reactor operators

​

• Annular ring element structure and loose fuel pin storage – “shuffling” of the pins experiencing varied levels of irradiation
• Stochastic modelling of varied irradiations indicating minimal effect from radial variation. However have assumed contribution may be as large as that of axial variation.

Uncertainties / Sensitivities (5) – Code uncertainties

• Burn-up Credit requires use of fuel depletion calculation code and neutron multiplication calculation code

​

• Each code has its own associated uncertainties/bias, which must be understood and appropriately incorporated into any operational loading curve

​

• Bias in neutron multiplication code (MONK) very low for U235/U238 systems, or systems with small amounts of Pu such as those modelled in the loading curve derivation – therefore focus on depletion code uncertainties

Uncertainties / Sensitivities (6) – Depletion Code

• Depletion code used at Sellafield – FISPIN (/FIS2000)

​

• Internationally available data limited for AGR (SFCOMPO), some additional spent fuel characterisation experiments commissioned for SL but largely for High Burn-up (HBU) fuels

​

• Internal SL data available from historic reprocessing operations (previous compliance arrangements involved fuel depletion modelling and spent fuel liquor sample analysis)

Uncertainties / Sensitivities (6) – Depletion Code

• Internal FISPIN & FIS2000 Validation reports, have compared modelled isotopic content and sample analysis results

​

• Some limitations – some experimental results are from large tanks of homogenised liquor derived from multiple (12+) spent fuel elements; some from samples of individual pellets

​

• Nevertheless, may give a good overview and provide confidence into levels of expected uncertainty

​

​

Uncertainties / Sensitivities (6) – Depletion Code

• Mix of conservative and non-conservative biases – how to appropriately manage these?

​

• Derive new minimum irradiation values which may tolerate levels of bias identified – under assumption of worst-case non-conservative biases for all relevant isotopes

​

• U235/Pu239/Pu241 content “artificially” increased, U238 ignored (validation indicated extremely low fractional uncertainties due to large abundance)

​

​

Uncertainties / Sensitivities (7) – Spent Fuel Declared Data

• Compliance against loading curve requires use of declared initial enrichment and irradiation data from fuel

​

• Uncertainty on initial enrichment very low (tight manufacturing tolerances on each fuel pellet, averaged over 65 pellets per pin and 36 pins per element)

​

• Larger uncertainties on the declared irradiation data

Uncertainties / Sensitivities (7) – Spent Fuel Declared Data

• Irradiation data provided via a combination of reactor operator modelling data supported by instrument analysis throughout reactor operations

​

• Reactor modelling utilises number of codes and sub-processes to interrogate macroscopic trends (stringer power, etc.) and microscopic effects (behaviour of fuel adjacent to control rods, etc.)

​

• Combine with measured analysis of reactor channel powers, temperatures etc. (useful analogues for fuel burn-up)

​

• Uncertainty on declared irradiation value estimated to be 5-10% (supported also by decay heat analysis) 🡪 Assumed 10%

​

Updated Operational Loading Curve

• Taking into account range of uncertainties discussed, the previously determined ‘minimum irradiation curve’ is updated to the ‘operational loading curve’.

​

​

Fuel Misload Analyses (1)

• Utilisation of burn-up credit introduces a new fault sequence group – fuel misloads

​

• Can occur either via operational error (e.g., operator selects wrong can from donor system to load into receipt system) or documentation error (e.g., paperwork gets associated with incorrect fuel at some point in transit)

​

• Worst case result – fresh fuel (at max IE) present instead of compliant fuel

Fuel Misload Analyses (2)

• Range of international guidance on extent of misload which should be considered

​

​

• Also informs extent of requirement for any post-irradiation verification measurements (samples/radiometrics etc)

​

​

• Scenarios include chronic issues leading to large transfers of moderately underburned fuel or acute issues leading to transfer of smaller amounts of severely underburned fuel (fresh or near-fresh)

Fuel Misload Analyses (2)

• Range of international guidance on extent of misload which should be considered

​

​

• Also informs extent of requirement for any post-irradiation verification measurements (samples/radiometrics etc)

​

​

• Scenarios include chronic issues leading to large transfers of moderately underburned fuel or acute issues leading to transfer of smaller amounts of severely underburned fuel (fresh or near-fresh)

Fuel Misload Analyses (3)

• Upstream failure of controls may result in presence of non-compliant fuel with no mechanism for this to be revealed during operations
• Therefore must also consider conjunction of fuel misload scenario and other fault sequences (e.g., mechanical handling accident) to understand tolerance

​

​

​

Fuel Misload Analyses (4)

• Overall, significant tolerance to fuel misloads either chronic or acute, and in conjunction with other independently unlikely fault scenarios

​

​

• Supports the overall confidence in the proposed arrangements

Cross-Checking (1)

• Calculations discussed thus far utilised the ‘ANSWERS’ suite of codes (specifically, MONK & FISPIN) – well-established set of codes with significant use in the UK, and the JEF-2.2 data libraries

​

• Best practice to include appropriate cross-checking to support the overall case.

​

• This has been achieved via use of parallel, independent set of calculations – utilising a separate code suite (SCALE) and separate nuclear data library (ENDF)

Cross-Checking (2)

Cross-Checking (3)

• Additional tools in SCALE (which were not readily available in FISPIN/FIS2000) also enable greater interrogation of uncertainties derived due to nuclear data and other reactor processes (e.g., power rating variability)

​

• Depletion calculations covering range of initial enrichments (3-3.82 w/o) and average irradiations (up to 18 GWd/teU)

​

• Neutron multiplication calculations via CSAS to then produce k-eff estimates, combined with above to cross-check the operational loading curve previously derived. Demonstrated good agreement.

Cross-Checking (4)

• Use of SCALE has also provided estimated contribution that extending to full fission product burnup credit would provide

​

​

​

​

​

​

• Overall uncertainty on k-eff for 3.78 w/o and 15,000 MWd/teU fuels (i.e. those with most significant amounts of non-U235/U238 contributions) is Δk/k = 0.6009% ± 0.0026%

Cross-Checking (5)

• Also provides estimates for nuclear data library covariances and their impact on the k-eff estimates (3.78 w/o & 15000 MWd/teU)

​

​

​

​

​

​

• Indicates that even for modelled scenarios with the greatest proportion of Pu present, the dominant isotopes and reactions are those also associated with a fresh fuel condition

Cross-Checking (6)

• Overall the cross-checking in SCALE has provided confidence that the MONK & FISPIN/FIS2000 process has achieved suitable results.

​

• The SCALE results also demonstrate a significant source of unused margin (Pu240 + FPs presence) which would provide only a relatively small impact to the overall results uncertainties

​

​

Post-Irradiation Verification (1)

• UK and international guidance for use of burn-up credit recommend consideration of post-irradiation verification measurements. This may typically be achieved via samples, or radiometrics.

​

• Previous SL implementation of burn-up credit utilised radiometric instruments known as the ‘Feed-Pond Fuel Monitors’ (FPFMs) to achieve this.

​

• Review instigated to determine feasibility of re-appropriating the FPFMs to support ASTOP loading arrangements, and what is the appropriate ALARP solution

Post-Irradiation Verification (2)

• Requirement for verification considered a function of a number of factors
• Levels of burn-up credit required
• Tolerance to secondary contingencies
• Safety margins beyond those formally credited within the assessment

​

​

• Guidance also available for scenarios whereby it may be feasible to forego verification measurements
• Requisite confidence in reactor records
• Supporting reactor operational measurements
• Sufficiently robust misload analyses.

​

​

Post-Irradiation Verification (3)

• Engineering-led review of the FPFMs concluded that the monitors had reached beyond their service life, and significant costs, additional risk and programme time/delay would be required to provide a suitable replacement.

​

• Overall, we can demonstrate good confidence against the 3 alternative arrangements outlined previously, indicating that the gap between Relevant Good Practice (RGP) would be small if foregoing the replacement of the verification measurements

​

• The ALARP balance therefore is judged that replacement of the FPFMs would grossly disproportionate compared to the benefit they would provide in this scenario.

Regulatory Engagement & LFE (1)

• Relatively infrequent application of burn-up credit within UK criticality safety cases, therefore particularly interesting to the regulator

​

• Frequent specialist-to-specialist engagement sessions between SL and the regulator

​

• Helped early identification of technical challenges, allowing them to be addressed prior to formal submission of the case to the regulator

Regulatory Engagement & LFE (2)

• Case has required significant technical underpinning and the multiple stakeholders have been involved in generating this.

​

• The early and frequent engagement has also facilitated discussions between 3^rd parties (e.g. reactor operators) and helped produce any additional required data as efficiently as possible

​

• Overall, significant benefit to the permissioning process has been obtained via this approach and adopting a similar method is strongly recommended in any future assessments utilising burn-up credit at SL

Regulator Engagement & LFE (3)

• Discovery, during safety case development, of significant manufacturing defect present in the spent fuel cans. Design stipulates a cylindrical vessel of fixed maximum diameter, manufacturing process resulted in some regions exceeding this diameter by up to 5%

​

• Requirement of significant criticality safety interest, as it limits the available arrangements of fissile material during operations

​

• Risk to development of ongoing case

Regulator Engagement & LFE (4)

• Review of existing inventory and a ‘safe state’ position justified (i.e. ongoing operations prior to introduction of 63c safety case and operations)

​

• Once safety of ongoing operations demonstrated, urgent strategic review of future operations (i.e. 63c racks). This determined the original burn-up credit approach was no longer suitable, as the diameter issue had eroded the previously understood safety margins

​

• Multi-discipline optioneering process to identify appropriate route forward

Regulator Engagement & LFE (5)

• Optioneering needs to consider and balance wide range of drivers
• Programmatic drivers (time, capacity, site and national strategy)
• Engineering capability (new designs may be challenging, etc.)
• Operability and maintenance (reliance on operator protections, need to support as much as possible to provide robust safety case)
• Radiological and criticality safety

​

• Balanced solution –increasing the minimum burn-up requirement, and additionally blanking off the central channel of the rack to lower the fissile loading.

​

• Reduced fissile loading provides significant reduction in burn-up credit requirements (from 36% of extant population to 10%), and also increases tolerability to misloading scenarios.

​

• Further blanking of channels provides diminishing returns (e.g., 1 -> 2; 10% -> 7.5%)

Regulator Engagement & LFE (6)

• Importance and benefit of prompt strategic review

​

​

• Rapid engagement of all stakeholders to de-risk further issues down the line

​

​

• Documentation and maintenance of strategic/optioneering processes should be thorough and reviewed at appropriate junctures

Summary/Conclusion

• Overall, have utilised a simplified approach to actinide-only burn-up credit – acknowledges depletion of U235 and production of Pu239/Pu241 to determine reduction in overall spent fuel reactivity.

​

• Generation of minimum and operational loading curves – taking appropriate consideration of physical variations, data uncertainties, and impact of validation data

​

• Cross-check supports the adequacy of the case and interrogates further the unused safety margins present

​

• Extensive misloading analyses demonstrates robust tolerance for failures/faults during implementation or operation of this burn-up credit arrangement