ML23116A214

Nureth 20 - Nrc'S Methodology to Estimate Fuel Dispersal During a Large Break Loss of Coolant Accident

ML23116A214
Person / Time
Issue date:	04/26/2023
From:	Andrew Bielen, James Corson, Joseph Staudenmeier NRC/RES/DSA
To:
References
Download: ML23116A214 (1)

Text

NRC's Methodology to Estimate Fuel Dispersal during a Large Break Loss of Coolant Accident Andrew Bielen, James Corson, and Joseph Staudenmeier United States Nuclear Regulatory Commission Washington, DC 20555-0001 andrew.bielen@nrc.gov; james.corson@nrc.gov; joseph.staudenmeier@nrc.gov

[leave space for DOI, which will be inserted by ANS]

ABSTRACT Recent experimental findings indicate that under certain transient conditions, high-burnup fuel operating at sufficiently high power can fragment and escape from burst fuel cladding, as discussed in the Nuclear Regulatory Commissions Research Information Letter (RIL) 2021-13. The escaped fuel fragments could subsequently be distributed throughout a reactor coolant system (RCS), potentially challenging core coolability during a loss of coolant accident (LOCA). Consequently, the NRC has developed a methodology to estimate the mass of fuel that could be dispersed during a large-break LOCA. The methodology involves several NRC-sponsored computer codes, including SCALE/Polaris for lattice physics, PARCS/PATHS for full-core neutronics, TRACE for core and RCS thermal hydraulics, and FAST for steady-state and transient fuel performance calculations. Output from FAST is combined with the models in RIL 2021-13 to estimate fuel dispersal to the coolant.

NRC staff have exercised this methodology using a prospective Westinghouse four loop core design with high burnup and extended enrichment that was obtained from the Department of Energys NEAMS program. Results demonstrate that the NRC methodology can be exercised to provide fuel dispersal estimates that can subsequently be used to study the potential consequences of FFRD. However, this exercise has also identified several areas for improvements to both the codes and to the modeling approaches used here, including the pin power reconstruction methods in PARCS, the modeling approach used to connect the Cartesian and cylindrical vessel components in TRACE, transient fission gas release and fuel rod upper plenum models in FAST, and tighter coupling between FAST and TRACE.

KEYWORDS LOCA modeling, TRACE, fuel dispersal

1. INTRODUCTION In the United States, there is recent heightened interest by the nuclear power industry to increase allowable maximum fuel burnup to improve economics [1, 2]. So-called high burnup (HBU) fuel often requires enrichment of Uranium-235 beyond the current 4.95 weight percent (wt/o) limit (i.e., extended enrichment (EE)) to meet cycle energy requirements. Prior to implementation, there are several safety concerns which must be addressed. For example, recent experimental findings indicate that under certain transient conditions, HBU fuel operating at sufficiently high power can fragment and escape from burst fuel cladding. The escaped fuel fragments could subsequently be distributed throughout a reactor coolant system (RCS) in adverse ways [2]. This so-called fuel fragmentation, relocation, and dispersal (FFRD) is a serious concern for the US Nuclear Regulatory Commission (NRC) and the industry. Consequently, methods development to understand and disposition the effects and consequences is being actively pursued. This article describes a methodology developed at the NRC to estimate the mass of fuel ejected

to the RCS during transient events via a convolution of burst size and location with pre-transient core-wide burnup distribution. At this time, FFRD concerns are primarily directed at pressurized water reactors (PWRs) during large-break loss of coolant accidents (LBLOCAs), although other event sequences, and other reactor types, may also be of importance. The work described here should be understood as a demonstration of our methodology, which could be employed to support review of eventual HBU/EE licensing submittals.

In traditional safety analysis, the limiting fuel rod is typically the highest powered rod in the core during the transient, which tends to be a fuel rod within a freshly loaded assembly. Given the relatively intact state of low-burnup fuel, this simplifies the regulatory and safety metrics considerably. In contrast, it is necessary to incorporate both the pre-transient power level and burnup to properly assess the safety implications of FFRD. The experimental evidence shows the fuel becomes susceptible to fine fragmentation starting at a pellet-average burnup of 55 GWd/MTU. In the past, the NRC has concluded that fuel dispersal would be minimal under existing burnup limits (roughly corresponding to 62 GWd/MTU rod-average burnup) and fuel management practices. However, NRC staff recognized that changes to either the burnup or the typical linear heat generation rate (LHGR) of rods above 55 GWd/MTU could lead to fuel dispersal that constitutes a safety concern [3]. With fuel vendors and utilities expected to seek burnup limit increases up to 75-80 GWd/MTU peak rod-average over the next few years, it is even more important that methodologies exist to, at the very least, conservatively bound the amount of fuel that may be released to the RCS during accident sequences. This allows the estimation of consequences in terms of long-term coolability of the system, and of potential dose to workers and the public.

The interaction between core loading strategy, pre-transient fuel rod state, and transient thermal-hydraulics and fuel performance necessitates a comprehensive multi-physics approach to properly represent the phenomena of interest. NRC staff have thus developed an analytical workflow called WHAM (Workflow for Holistic Accident Multi-physics) to evaluate the potential for FFRD within a given reactor core design. WHAM incorporates elements from the full spectrum of design-basis analytical tools currently under development at NRC:

• The Polaris lattice physics sequence from the SCALE code package [4] is used to generate few-group nuclear data, including power form factors, for each assembly segment present in the core.

• PARCS, the NRCs few-group nodal core simulator [5], uses the nuclear data produced by Polaris to simulate the required core operating cycles and generate power histories for every fuel rod within the core using its pin power reconstruction methodology.

• TRACE, the NRCs flagship design-basis thermal-hydraulic analysis code [6], is used to simulate the conditions in the RCS and reactor core during the transient of interest.

• FAST, the NRCs fuel performance code [7], is used to simulate the evolution of each fuel pin during normal operations based on the power histories generated by PARCS. It is also used to calculate transient fuel performance during the LBLOCA using thermal-hydraulic boundary conditions generated by TRACE.

A detailed description of WHAM is provided in the next section, followed by an example application to a prospective HBU/EE core design developed by a utility under Department of Energy (DOE) sponsorship and documented in the public domain by the DOEs Nuclear Energy Advanced Modeling and Simulation (NEAMS) program [8]. We acknowledge that this core design is of preliminary nature and has not undergone the full gamut of analysis required for production reloads. However, it does provide a publicly available testbed for the WHAM methodology and allows us to demonstrate its capabilities in (1) calculating key figures of merit and (2) performing sensitivity studies which elucidate the effects of key inputs on calculated mass release from the fuel rods.

2. DESCRIPTION OF METHODOLOGY To adequately model transient scenarios where FFRD is a concern, the initial condition of the fuel at the core state point of interest must be comprehensively defined. One necessary component of this work is thus to resolve the pre-transient fuel state with all the parameters that could affect its transient performance (i.e., local burnup, rod internal pressure (RIP), etc.) with enough detail to accurately make meaningful fuel dispersal predictions. Another is representing the core thermal-hydraulics in sufficient detail to adequately capture local effects of heat transfer on thermo-mechanics of the fuel during the accident progression. NRCs design-basis code suite allows for such modeling detail, which was therefore leveraged for this work, combined with auxiliary codes and Python scripts for efficient data transfer from model to model. A flow chart describing WHAM is provided in Figure 1.

Figure 1. Summary of WHAM Workflow; blue boxes indicate simulations performed at the assembly level and orange boxes indicate simulations performed at the pin level The Polaris/PARCS code system is used to simulate the operating history of the reactor fuel which is in the core during the transient of interest, with GenPMAXS being the conversion tool from Polaris lattice physics outputs to few-group nuclear data libraries readable by PARCS. PARCS solves for the power distribution at each point in core life by solving the nodal diffusion equations on a three-dimensional cartesian grid. In PARCSs case, each radial node in a plane corresponds to a single assembly, for which the effective homogenized nuclear data is provided by the Polaris-generated library. Pin power reconstruction (PPR), which convolutes the three-dimensional core power distribution with two-dimensional power form factors in each lattice pre-computed by Polaris, is used to generate power histories for each individual rod within the core.

The individual rod powers are then used in two ways. First, they are used to directly simulate the operating history of each rod in the core to initialize the rod-by-rod transient fuel performance simulations. Second, they provide a basis for representative rods per assembly which are used in the TRACE model. The latter is necessary to (1) leverage TRACEs advanced fuel thermo-mechanical capabilities to provide for some fuel feedback during the transient event, while (2) keeping the number of heat structures in the thermal-hydraulics model reasonable while still maintaining the degree of detail needed to resolve local heat transfer details, at least to the assembly level. Within each assembly, the fuel

rods are grouped by design into representative rods depending on enrichment, burnable absorber loading, etc., by averaging their power histories together. For PWRs, this typically means 2-4 fuel rod heat structures per assembly in the TRACE model. Additionally, the highest-powered rod within each assembly (i.e., the hot rod) is represented by a supplemental rod to allow for evaluation of peak cladding temperature (PCT). The data provided to TRACE from the FAST operating history simulations includes:

• Radial and axial burnup profiles within the fuel

• Fuel swelling and densification

• Cladding permanent deformation due to creep

• Cladding outer surface corrosion thickness and hydrogen content

• Gap gas constituents and amounts In addition to the advanced fuel capabilities, TRACEs cartesian VESSEL component is used to represent the active core region in the thermal-hydraulics model to provide for assembly-by-assembly heat transfer characteristics. This is considerably more detailed than the legacy TRACE core modeling approach, which was to group the core into 12-24 radial wedges (i.e., two to three rings and six to eight azimuthal sectors) within a cylindrical VESSEL and allows for a more explicit treatment of the flow distribution within the core. It also allows for a more explicit representation of the local fuel characteristics and its impact on the heat transfer. The TRACE core model is initialized with the radial and axial power distribution read from the PARCS model at the statepoint of interest but can be modified manually to perform sensitivity studies as desired. With these features incorporated, TRACE is used to simulate the accident event of interest and calculate time histories of core thermal-hydraulics throughout.

Upon completion of the TRACE simulation, the bundle-by-bundle heat transfer coefficients and coolant conditions from the TRACE thermal-hydraulic transient calculations are combined with the rod-by-rod fuel performance simulations to predict the performance of each fuel rod during the accident. If a fuel rod is predicted to burst, the burnup at the burst location can be identified, and, combined with the burst strain, the fraction of fuel dispersal can be estimated based on models available in the RIL. Additional quantities of interest (e.g., burst temperatures, locations, LHGRs) can also be determined from the code output. This information is integrated from every fuel rod to generate core-wide estimates of mass dispersal into the RCS.

3. DESCRIPTION OF EXAMPLE APPLICATION: HIGH BURNUP 4-LOOP CORE DESIGN To exercise the WHAM methodology, NRC staff performed demonstration calculations using a prospective increased enrichment, high burnup loading pattern for a Westinghouse four-loop core design that we obtained from the DOEs NEAMS program. The report from which this core design was taken describes the prospective transition from 18-month, standard enrichment operating cycles to 24-month equilibrium HBU/EE cycles for a 3626 MWth, 193 assembly reactor. At equilibrium, the end-of-life assembly burnups in the HBU/EE core are greater than 71 GWd/MTU. It uses Westinghouse VANTAGE+ fuel assemblies [9] with enrichments of 5.95, 6.2, or 6.6 weight percent U-235, up to 200 integral fuel burnable absorber (IFBA) rods per assembly, and heavy use of wet annular burnable absorber (WABA). The high IFBA and WABA loadings are needed to control the high reactivity of the core design and to limit the early-cycle boron concentration. The core loading pattern is presented in Figure 2.

Figure 2. Core loading pattern used in this work (quarter-core representation). Red are feed (fresh) assemblies, green are once-burned and blue are twice-burned.

This core design employs fuel management consistent with current practice in the operating fleet: feed fuel is placed in a ring of fire adjacent to the core periphery and checkerboarded with once-burnt fuel on the interior. Higher burnup fuel is placed in low-power regions on the periphery. It should be noted that the transition described in the NEAMS report was to a two cycle, even/odd equilibrium in which the loading alternated between 84 and 85 feed assemblies, respectively. Only the odd pattern is explicitly presented in the report, so we obtained the even cycle through communication with NEAMS staff [10].

The IFBA patterns within each lattice were not made available, so we used patterns from previous NRC work [11]. PARCS models were built of the even and odd cycles, and the shuffles between them. We did not explicitly simulate the transition from the 18-month cycle as in the NEAMS report. Rather, we used an equilibrium cycle approach common with PARCS in which we initialized the odd cycle with fully fresh fuel, depleted it, and iteratively applied the odd-to-even and even-to-odd loading patterns at the end of each cycle depletion until the maximum difference in nodal burnup at end of cycle (EOC) between successive even and odd cycles converged within 0.1 GWd/MTU. This produced cores with neutronics characteristics comparable to the reference values, as will be seen in the next section.

The TRACE model explicitly models all four loops, with emergency core cooling system (ECCS) configuration consistent with an operating reactor of that rated power level. The core model is initialized based on EOC conditions. Each assembly has two fuel rod heat structures (Non-IFBA/IFBA), which results in 386 fuel heat structures that interact with the hydraulics, as well as 193 supplemental hot rods which provide for PCT estimation. The cartesian core VESSEL has 193 active radial nodes and 24 axial levels and is connected to a cylindrical VESSEL that represents the rest of the reactor vessel (RV). The RV has four radial rings, the inner two of which are directly above and below the core, and 8 azimuthal sectors. Ring 1 is connected to the interior assemblies, whereas ring 2 is connected to the peripheral assemblies. The placement of the core within the RV is presented in Figure 3.

Sensitivity studies were conducted with the TRACE model to examine the impact of various modeling parameters on the quantity of dispersed fuel. In particular, the effect of axial power distribution, decay

heat uncertainty, and number of available ECCS trains was studied. These results are presented in the next section.

Figure 3. Placement of core VESSEL within reactor vessel. White are active flow channels, red and green are mapped to reflector nodes

4. RESULTS 4.1. Steady-state Results The first step was to generate the fuel rod power histories used in the FAST steady-state calculations.

NRCs cycle depletion results are in good agreement with the NEAMS results obtained using VERA and utility results using NRC-approved codes and methods, as shown in Table I. Figure 4 provides further comparison between the critical boron concentration and maximum fuel rod peaking factor (FdH) calculated by NEAMS and NRC tools. The results are in reasonable agreement, particularly given the fidelity of the VERA core simulator and the fact that the IFBA patterns were not likely entirely consistent.

The fuel rod power histories from Polaris/PARCS shown in Figure 5 are used for the steady-state fuel performance calculations with FAST. NRC staff performed two sets of calculations: a smaller set of representative non-IFBA and IFBA rods and a hot rod for each assembly, and a larger set containing every rod in the core. The smaller and larger sets of steady-state FAST results were used to initialize the TRACE heat structures and the FAST rods for the transient systems analysis and fuel performance calculations, respectively. It is clear from Figure 5 that pin-by-pin calculations cover a much broader range of power and burnup conditions than the representative rods used in TRACE, which demonstrates the power of our methodology to capture features lost when collapsing the analysis to a smaller set of fuel rods. The calculations for the additional pins not modeled in TRACE assume that the thermal hydraulic conditions calculated by TRACE for the fuel pins it models are representative of what the additional pins would see.

Table I. Comparison of NRC core depletion results to utility and NEAMS results Utility Parameter NEAMS NRC Simulation Maximum fuel 1.49 1.51 1.52 rod power (-)

Maximum local 1.76 1.85 1.91 pin power (-)

Maximum 1.39 1.39 1.40 assembly power Maximum critical boron 1568 1586 1573 concentration (ppmB)

Figure 4. Core depletion results from VERA (left) and from Polaris/PARCS (right). The VERA results are taken from Figure 10 of Ref. [8].

Figure 5. Linear heat generation rates for the TRACE heat structures (left) and for the FAST pin-by-pin calculations (right).

Figure 6 shows the end-of-cycle fission gas release and rod internal pressure results from FAST. The results show that a significant fraction of the fuel rods have internal pressures above the system pressure.

This is especially true for the IFBA rods, though there are some non-IFBA rods that also exceed system pressure. Furthermore, several rods have greater than 10% fission gas release, which may be due in part to cladding liftoff and the fuel-cladding gap reopening at these extreme rod internal pressures. Again, the intent of this demonstration is not to design a realistic core that can meet all the reload safety checks, but these results highlight the challenge of meeting rod internal pressure limits and preventing cladding liftoff for high burnup PWR core designs with 24-month cycles.

Figure 6. Fission gas release and rod internal pressure at end-of-cycle calculated by FAST.

4.2. Transient Results As mentioned in Section 3, NRC staff ran several TRACE calculations to explore the impacts of various parameters on fuel rod behavior during a double-ended guillotine break of a cold leg, including the effects of the three power profiles shown in Figure 7. All four loops were modeled separately, and the break was in a loop that did not contain the pressurizer. The calculations assumed a loss of offsite power coincidental with the break. Calculations were performed with both one and two trains of ECCS available and the pressure boundary condition on the break was set by a coupled containment backpressure calculation to capture the significant impact of containment pressure on core reflood rates and reflood heat transfer. Using a single train of ECCS extends the reflood phase of the accident compared to the cases using two trains of ECCS. Calculations using two trains of ECCS have a slightly lower containment pressure in the long term because of the additional cold water spilling from the break. Figure 8 shows the TRACE results for the maximum cladding temperature throughout the LOCA for each sensitivity case.

The base case resulted in cladding temperatures that were low compared to the 1204 C acceptance criterion in 10 CFR 50.46, as did scenarios with higher decay heat and with a single train of ECCS. The biggest difference in peak cladding temperature behavior results from changing the axial power profile.

The peak cladding temperatures during blowdown are largely determined by the LHGR. The peak cladding temperatures during reflood are dependent on the location of the peak power axially in the core and the reflood rate in addition to the LHGR. In the most extreme case of a top peaked power profile with a single train of ECCS, the peak cladding temperature approached the 1204 C limit. This case also resulted in the largest number of burst rods in the FAST transient calculations, as will be discussed shortly.

Figure 7. Axial power profiles used in the TRACE and FAST transient calculations.

Figure 8. Maximum TRACE heat structure temperatures during a LOCA.

The TRACE coolant conditions and heat transfer coefficients were applied as boundary conditions for the FAST transient fuel performance calculations. The results are summarized in Table II. For all cases, FAST predicts that a large percentage of rods would burst during the LOCA. In general, the percentage of IFBA rods that burst is higher than the non-IFBA percentage, but both are still significant. Figure 9 shows a clear correlation between pre-transient linear heat generation rate and fuel rod burst for the base case.

The results are likely conservative, for reasons that will be discussed later. With that said, the rod internal pressures of many of the high burnup rods are quite high at the start of the transient, so it is still expected that a significant percentage of high burnup rods would burst.

The mass of fuel that would be dispersed from the burst rods to the coolant was estimated using burnup and strain results from FAST and models presented in Appendix A to RIL 2021-13 [2]. The results are included in Table II. All models require the fuel to be susceptible to fine fragmentation, which is defined in RIL 2021-13 as fuel with a pellet-average burnup greater than 55 GWd/MTU and cladding hoop strain greater than 3%. The first model (model A in RIL 2021-13) assumes that all fuel in nodes susceptible to fine fragmentation that are contiguous with the burst node would disperse. The second model (model C in RIL 2021-13) assumes that only a fraction of the fuel in the nodes susceptible to fine fragmentation would disperse. The fraction increases from 0.0 at 55 GWd/MTU to 1.0 at 80 GWd/MTU, based on available data on the fragment size distribution described in RIL 2021-13. The final model is a variation of model C where only fuel in and above the burst node disperses, which reflects the fact that it would take a significant amount of force to expel fragments below the burst location from the rod.

Table II. FAST transient fuel performance results for several sensitivity cases, all with 2 trains of ECCS Top Peak Chopped Top Peak Power Cosine Parameter Base Case Power Shape (1 Power Shape ECCS Shape train)

Burst rods (%)

IFBA 64 68 76 78 Non-IFBA 40 32 69 80 Total 58 58 74 78 Fuel dispersal (%)

All fragment sizes 2.3 2.8 3.5 3.4 Fragments < 1 mm 1.1 1.9 2.1 2.1 Fragments < 1 mm 0.6 1.3 1.1 1.1 above burst Figure 9. Core map showing burst and non-burst rods for the base case. Red and magenta indicate burst IFBA and non-IFBA rods, respectively; blue and cyan indicate non-burst IFBA and non-IFBA rods, respectively. Bursts are generally correlated with the pre-transient linear heat generation rate shown on the right, though the system thermal hydraulic response clearly impacts burst as well.

Note that LHGR values are in kW/m.

Our results show relatively little variation in the number of rods that burst for all cases except those with a top peaked power shape. This is not surprising, given the similar peak cladding temperatures for all cases with the nominal and chopped cosine power shapes. However, there were some notable differences in burst timing because of the higher temperatures calculated by TRACE during the blowdown phase for the chopped cosine and top peaked power profiles. Figure 10 and Figure 11 highlight these differences between the base case and chopped cosine power shapes.

There was more variation in the estimated fuel dispersal shown in Table II. Compared to the base case, the top peak power shape results in greater estimated dispersal due to the greater number of high burnup rods that burst, while the chopped cosine power shape results in more dispersal because the burst node is closer to the core midplane (see Figure 10), where the burnup is greater than near the top of the rod. The differences in burst behavior also impact the timing of fuel dispersal shown in Figure 11. This has implications for transport of dispersed fuel particles in the reactor coolant system since flow rates during the blowdown are much higher than during the reflood phase of the accident.

Figure 10. Burst elevations for second cycle rods predicted by FAST for the base case (left) and the chopped cosine power shape (right).

Figure 11. Dispersed mass fraction for the base case (left) and the chopped cosine power shape (right). The dispersed mass fraction is calculated using different models in RIL 2021-13 [2] that are described in the main body of this paper. Note that the trends in the dispersed fraction versus time for the top peaked cases (not shown) are similar to the base case, though the total mass dispersed was greater in the top peaked case.

As noted, the burst results from the FAST calculations may be conservative. First, FAST is predicting anywhere from 0 to 10% transient fission gas release during the LOCA. This is in addition to the fission gas released during normal operations. Note that for these calculations make use of the default modified Forsberg-Massih model for fission gas release, which is described in detail in the FAST code documentation. For now, it is worth pointing out that the Massih model assumes that once the fission gas

concentration on the grain boundaries in each radial node exceeds the saturation concentration, all grain boundary and re-solved fission gas in that node is released to the internal void volume. During transients, the saturation concentration is cut in half to match comparisons to power ramp calculations. As a result, a significant amount of fission gas is released early in the transient for rods with relatively high linear heat generation rates. While the overall magnitude of the release is consistent with past transient fission gas release measurements under conditions more typical of a LOCA [12, 13, 2], the timing of the release is likely inaccurate. Furthermore, it must be noted that the fuel temperatures for the base case are relatively low compared to the peak temperatures in past transient fission gas release tests. For these reasons, FAST may be conservatively predicting both the magnitude and the timing of transient fission gas release.

Clearly, more work is needed to develop and validate a transient fission gas release model that is applicable to loss-of-coolant accident conditions.

Second, FAST is predicting burst at lower temperature than expected based on the Chapman correlation, as seen in Figure 12. The Chapman correlation is based on internally heated burst tests described in NUREG-0630 [14] and is used in TRACE. FAST performs a more mechanistic calculation of fuel rod ballooning that is triggered when the cladding effective plastic strain in an axial node exceeds the instability burst strain from MATPRO. The ballooning model calculates the localized non-uniform strain in the cladding in that node. It predicts failure when the true stress exceeds a temperature-based limit or when the permanent hoop strain exceeds a temperature-based limit. Results from the ballooning model are in reasonable agreement with the data in NUREG-0630 for the more idealized heat up conditions in those tests [7]. With that said, FAST does tend to predict earlier failure at lower temperature when compared to integral LOCA experiments [15], so there is room for improvement in the code predictions.

This is particularly true for rods with a slower average heat up rate, since FAST is in better agreement with the Chapman correlation for rods that burst during the blowdown when the average heat up rate is high.

Figure 12. Base case burst predictions compared to the Chapman correlation (left), using axes consistent with those shown in Figure 3 from NUREG-0630 (right) [14]. The heat map shows the average heat up rate between the start of the accident and burst.

5. CONCLUSIONS A methodology has been developed to enable the estimation of fuel dispersal during accidents in PWRs in as much explicit detail as can be supported by the code suite currently used at NRC for LWR analysis.

This methodology has been demonstrated using a prospective core loading pattern developed by a utility with DOE support. For this loading pattern, with the sensitivities that were explored here, dispersal estimates range from 0.6% to 3.4% of the mass of the fuel in the core. Other loading patterns, with other

fuel management strategies (i.e., batch loading fraction, burnable absorber type/distribution) may lead to very different FFRD behavior. One of the most challenging aspects to this problem is the difficulty in drawing universal conclusions based on specific examples.

In addition to exploring other potential core designs, there are areas where additional methodological work is required:

• PARCS currently models one radial node per assembly for purposes of diffusion and exposure/history tracking. This is coarse relative to some of the other core simulators in the industry and does not allow for intra-bundle radial burnup gradients to be captured. PARCS should be upgraded to allow finer resolution, which is particularly important for ring-of-fire/peripheral assemblies.

• The cartesian core VESSEL represents bundle-averaged flow within each assembly during the transient and provides this as boundary conditions to rod-by-rod FAST transient simulations. There may be important local effects at the subchannel level that are lost in this averaging procedure.

• The cartesian core VESSEL in this work only represents the flow in the active core region. This does not allow for explicit communication between the control rod guide tubes (CRGTs) in the upper core structures and the assemblies they sit above, which may affect top-down quench during reflood within those assemblies. The core vessel should be extended to the upper core plate, and the CRGTs explicitly modeled.

• There are only two rings below the core in this demonstration problem, connected to the peripheral and interior assemblies, respectively. It may be important to have a higher radial resolution of the lower plenum in the reactor vessel.

• The transient fission gas release models in FAST are highly empirical and are based on data from reactivity-initiated accident and power ramp tests. Additional work is needed to develop and validate models appropriate to LOCAs.

• As noted, FAST is predicting lower burst temperatures than what one would expect from the Chapman correlation. As a result, this methodology may be overestimating the number of rods that burst in a LOCA.

This should be understood as a demonstration project. We recognize that more validation is needed to increase confidence in our results. With that said, the methodology described here and the expertise gained through its utilization could be used to support future licensing reviews that involve potential dispersal.

ACKNOWLEDGEMENTS NRC staff wish to thank Andrew Godfrey and Ugur Mertyurek, Oak Ridge National Laboratory, for providing the required loading patterns and running the Polaris models, respectively. We also acknowledge Andrew Ireland, Lucas Kyriazidis and Steven Muller, NRC Office of Nuclear Regulatory Research, for incorporating the cartesian VESSEL into the TRACE model, setting up and running the FAST steady-state calculations used to initialize the TRACE heat structures, and assisting in post-processing the FAST pin-by-pin transient calculations, respectively. Finally, they thank their colleagues in the NRC Office of Nuclear Reactor Regulation for feedback during the development and exercise of this methodology.

REFERENCES

1. "Project Plan to Prepare the U.S. Nuclear Regulatory Commission for Efficient and Effective Licensing of Accident Tolerant Fuels, Version 1.2," U.S. Nuclear Regulatory Commission, Washington, DC (2021).

2. M. Bales, A. Chung, J. Corson and L. Kyriazidis, " Interpretation of Research on Fuel Fragmentation, Relocation, and Dispersal at High Burnup," RIL 2021-13, U.S. Nuclear Regulatory Commission, Washington, DC (2021).

3. "Evaluation of Fuel Fragmentation, Relocation and Dispersal Under Loss-Of- Coolant Accident (LOCA) Conditions Relative to the Draft Final Rule on Emergency Core Cooling System Performance During a LOCA (50.46c)," SECY-15-0148, U.S. Nuclear Regulatory Commission, Washington, DC (2015).

4. W. A. Wieselquist and R. A. Lefebvre, "ORNL/TM-SCALE-6.3.0, Version 6.3.0," Oak Ridge National Laboratory, Oak Ridge, TN (2022).

5. T. Downar, A. Ward, V. Sekar, Y. Xu and N. Hudson, "PARCS v3.3: U.S. NRC Core Neutronics Simulator Theory Manual," Department of Nuclear Engineering and Radiological Sciences, University of Michigan, Ann Arbor, MI (2021).

6. "TRACE V5.0 Patch 7 Theory Manual: Field Equations, Solution Methods and Physical Models,"

Division of Systems Analysis, US Nuclear Regulatory Commission, Washington, DC (2022).

7. K. Geelhood, D. Colameco, W. Luscher, L. Kyriazidis, C. Goodson, J. Corson and J. Whitman, "FAST-1.1: A Computer Code for Thermal-Mechanical Nuclear Fuel Analysis under Steady-state and Transients," Pacific Northwest National Laboratory, Richland, WA (2022).

8. N. Capps, A. Wysocki, A. Godfrey, B. Collins, R. Sweet, N. Brown, S. Lee, N. Szewczyk and S.

Hoxie-Key, "Full Core LOCA Safety Analysis for a PWR Containing High Burnup Fuel,"

ORNL/TM-2020/1700, Oak Ridge National Laboratory, Oak Ridge, TN (2020).

9. P. Doshi, D. Chapin and L. Scherpereel, "Westinghouse VANTAGE+ fuel assembly to meet future PWR operating requirements," Transactions of the American Nuclear Society, San Diego, CA (1998).

10. Personal communication, Godfrey, Andrew (Oak Ridge National Laboratory) to Bielen, Andrew (NRC), September 2021.

11. J. Hu, U. Mertyurek and W. Wieselquist, "Assessment of Core Physics Characteristics of Extended Enrichment and Higher Burnup LWR Fuels using the Polaris/PARCS Two-Step Approach. Vol. I:

PWR Fuel," Oak Ridge National Laboratory, Oak Ridge, TN (2022).

12. Y. Pontillon, M. P. Ferroud-Plattet, D. Parrat, S. Ravel, G. Ducros, C. Struzik, I. Aubrun, J. Eminet, J.

Lamontagne, J. Noirot and A. Harrer, "Experimental and theoretical investigation of fission gas release from UO2 up to 70 GWd/t under simulated LOCA type conditions: the GASPARD program,"

Proceeding of the 2004 International Meeting on LWR Fuel Performance, Orlando, FL (2004).

13. A. Bianco, C. Vitanza, M. Seidl, A. Wensauer, W. Faber and R. Macián-Juan , "Experimental investigation on the causes for pellet fragmentation," Journal of Nuclear Materials, 465, pp. 260-267 (2015).

14. D. Powers and R. Meyer, "Cladding Swelling and Rupture Models for LOCA Analysis," NUREG-0630, U.S. Nuclear Regulatory Commission, Washington, DC (1980).

15. K. Geelhood, D. Richmond, D. Colameco, T. Zipperer, W. Luscher, L. Kyriazidis, C. Goodson and J.

Corson, "FAST-1.1: Integral Assessment," Pacific Northwest National Laboratory, Richland, WA (2022).