Full-core high-burnup BWR LOCA fuel performance analysis and FFRD susceptibility

1 Introduction

The nuclear industry is continuously improving operations, safety, reliability, capacity factor, and economics of the current light water reactor, LWR, fleet. Improvements in plant reliability and safety have led to record capacity factors in the U.S. fleet. There are several initiatives in place to support this continuous improvement, including extended power uprates that provide more energy from the installed capacity, and fuel cycle extensions that provide capacity factor improvement. One of the technical hurdles of this effort is the fuel fragmentation, relocation, and dispersal (FFRD) during postulated loss-of-coolant accident (LOCA) scenarios (Greenquist et al., 2023). In this work, the BISON fuel performance code (Williamson et al., 2025; Williamson et al., 2021) was used to predict the full-core susceptibility of an operating BWR to FFRD during a postulated end-of-cycle large-break LOCA (LBLOCA). This analysis combined state-of-the-art multiphysics modeling and simulation tools with recent BWR-specific additions to BISON’s modeling suite (Gamble et al., 2025), detailed below, to provide the highest-fidelity predictions to date of fuel performance throughout a BWR core during normal operation and LOCA conditions.

The Nuclear Energy Advanced Modeling and Simulation (NEAMS) program is supporting the nuclear energy industry by, among other efforts, developing modern modeling and simulation tools for nuclear reactor analysis and using these tools to analyze existing and next-generation reactors. Under the direction of NEAMS, researchers at Oak Ridge National Laboratory developed a multiphysics framework that performed full-core analyses on reactor physics, thermal hydraulics, and fuel performance during normal operation and LOCAs to predict fuel susceptibility to FFRD (Greenquist et al., 2023; Hirschhorn et al., 2023; Wysocki et al., 2023; Greenquist et al., 2025b; Capps et al., 2025). The framework utilizes three codes: the Virtual Environment for Reactor Applications (VERA), the TRAC/RELAP Advanced Computational Engine (TRACE), and BISON. This framework was originally developed for PWRs but has recently been expanded to include BWRs (Capps et al., 2025; Greenquist et al., 2025b). Specifically, a VERA model of Limerick Unit 1 (LM1) Cycles 16–18 was developed and run to obtain the initial conditions for the LBLOCA. The results were ported to BISON, where 945 rods were modeled during normal operation (Greenquist et al., 2025a; Capps et al., 2024). In this work, the model was again expanded to include a postulated LBLOCA. TRACE was used to model an LBLOCA in the same reactor, with model subchannels corresponding to the same 945 rods. These results were then fed back to BISON to develop a full-core LBLOCA fuel performance and burst susceptibility study. Two separate cladding burst models were utilized to evaluate cladding integrity and FFRD susceptibility. An earlier version of this analysis was published in an ORNL Technical Memo (Greenquist et al., 2025b).

2 Limerick Unit 1

LM1 is a GE BWR operated by Constellation Energy in Limerick Township, Pennsylvania. LM1 began operation in 1984 with a power rating of 3,458 MW_th. LM1 received a power uprate to 3,515 MW_th in 2010 and a license extension in 2014.

LM1’s core is composed of 764 assemblies consisting of Global Nuclear Fuel (GNF) 10 × 10 fuel assembly designs and GNF2 and GNF3 product lines. The reactor operates on a 24-month cycle with a few weeks of refueling outages between cycles. Assemblies typically remain in the core for up to three cycles. Assemblies that are in their first, second, and third cycles are referred to as fresh, once-burned, and twice-burned, respectively, in this work. Twice-burned assemblies are placed on the core periphery, where the bundle power, linear heat generation rate (LHGR) and therefore, fuel temperatures, are lower. Each assembly contains 92 fuel rods arranged in a 10 × 10 grid. Two water rods near the center of the assemblies take up the remaining eight slots in the grid. The water rods were included in the VERA and TRACE models, but were not explicitly modeled in BISON. Several assembly types are used, but an example is shown in Figure 1. The assemblies contain short partial-length rods (SPLRs), long partial-length rods (LPLRs), and full-length rods (FLRs). Both the fuel and plenum regions are shorter in the SPLRs and LPLRs than in the FLRs. In addition, the FLR nearest the center of the nearest control blade is unenriched to reduce the neutron flux gradient.

Limerick Unit 2 (LM2), LM1’s sister unit, has the same configurations, power level, and cycle length as LM1. Eight high-burnup lead use assemblies (HBLUAs) were irradiated in LM2 (Global Nuclear Fuel, 2021). The HBLUAs are twice-burned assemblies that were irradiated for a third cycle despite exceeding their normal exposure limits in accordance with the U.S. Nuclear Regulatory Commission’s lead test assembly program guidance. For this work, these HBLUAs were included in the LM1 model to include more high-burnup rods for the analysis. One of the HBLUAs was irradiated for a fourth cycle, but that cycle is not included in this model. The core loading pattern during the cycle of interest is shown in Figure 2. Actual reactor conditions were used to inform the fuel loading pattern, assembly designs, coolant pressure, control blade positions, and other inputs. This approach made the simulations as physically accurate as possible.

3 NEAMS multiphysics framework

The multiphysics workflow developed by NEAMS is shown in Figure 3. The framework includes full irradiation cycles followed by an end-of-cycle LBLOCA. VERA was used to model the core’s neutron economy and depletion on a rod-resolved scale. Three irradiation cycles were modeled in VERA so that the full history of each rod in the core during the LBLOCA could be included. VERA’s standalone module, VERAOneWay, was used to populate a BISON input file template with rod-specific data. BISON simulations were used to analyze the fuel performance during normal operation. VERA results were also fed into TRACE. TRACE was used first to develop a normal operation model. The fuel centerline temperature predictions were compared to BISON’s corresponding predictions to ensure accuracy. The TRACE centerline fuel temperature predictions were found to match BISON’s predictions reasonably well, considering the uncertainties associated with centerline fuel temperature calculations and the differences in modeling approaches between the two codes. These differences were considered of a small enough magnitude to cause only a minor impact on the cladding heatup behavior during the LBLOCA. TRACE predictions were then ported back to BISON to develop normal operation plus LBLOCA simulations that could predict the fuel performance during the LBLOCA while accounting for the complete irradiation histories of individual rods. These BISON simulations included two cladding burst models, which predicted the time and location of rod bursts during the LBLOCA. Depending on the results of the cladding burst predictions, further analysis could be conducted to predict the reactor’s susceptibility to FFRD (Greenquist et al., 2023).

3.1 VERA

VERA is a full-core, rod-resolved, high-fidelity multiphysics solver used for analyzing LWR cores (Turner et al., 2016). VERA is a framework that couples several modular codes, including CTF, MPACT, MAMBA, and others. CTF is a rod-resolved LWR thermal hydraulics code. It uses a two-fluid, three-field modeling approach. MPACT is a 1D/2D, full-core deterministic neutron transport code capable of generating sub-pin neutron fluxes and power distributions. MAMBA predicts the growth of crud on fuel rods. VERA was originally developed for PWRs. However, beginning in 2019, DOE-NE funded a project to expand VERA’s capabilities to include BWRs. These relatively new capabilities were used to develop the model of LM1 in 2024 (Capps et al., 2024; 2025).

3.2 TRACE

TRACE is a thermal hydraulics code developed by the US Nuclear Regulatory Commission. It is built on the TRAC-P, TRAC-B, RELAP5, and RAMONA codes (Bajorek et al., 2026). It solves a six-equation formulation that includes mass, momentum, and energy conservation equations for the liquid and vapor phases. It is applicable to a wide range of nuclear reactor types—including PWRs, BWRs, and advanced reactors—in both normal operation and transient conditions.

For this work, an existing 4-loop BWR plant model was used as a starting point. Depletion and boundary conditions were imported from VERA. A normal operation model was first developed, and its fuel centerline temperature profile predictions were compared to BISON predictions for code-to-code benchmarking (Wysocki et al., 2023). Once the model was benchmarked and agreement was considered satisfactory, an LBLOCA transient was modeled. A more complete description of the TRACE model is available in the literature (Greenquist et al., 2025b). TRACE predictions were exported as text files, and a Python script was used to import those results into BISON as boundary conditions. Specifically, the transient LHGR was taken from the TRACE input while the cladding surface temperatures and coolant pressures were taken from the TRACE results. These time-dependent values were inserted into the BISON simulation as a replacement for the final shutdown step in the cycle of interest.

3.3 BISON

BISON is a finite-element-method-based fuel performance code deployed and maintained at Idaho National Laboratory, and it is developed by BISON’s user community using the Git version-control software (Williamson et al., 2025; Rizk et al., 2025). BISON is built on the Multiphysics Object-Oriented Simulation Environment (MOOSE) (Harbour et al., 2025). BISON uses a flexible, modular design to incorporate a wide range of fuel types, materials, geometries, reactor types, and operating conditions. BISON is most commonly used for PWR UO₂ fuel rods, BWR UO₂ fuel rods (Gamble et al., 2025), and tristructural isotropic (TRISO) fuels (Hales et al., 2020); however, it can also model mixed oxides (Ferrigno et al., 2025), uranium-zirconium (Greenquist and Powers, 2021), uranium-molybdenum (Doyle et al., 2023), uranium-carbide (Williamson et al., 2021), and uranium-nitride (Williamson et al., 2021) fuels. BISON contains a robust validation suite. The LWR suite includes data from both PWRs and BWRs. Historically, BISON has modeled these reactors using very similar techniques. However, several capabilities were developed, validated, and added to the BISON repository in 2025 with the specific goal of improving BWR simulations (Gamble et al., 2025). They include:

• Burnable absorbers such as gadolinium-oxide in the fuel pellets, including their effects on burnup and thermal conductivity calculations,
• Modern cladding materials that are deployed in some BWRs (Romero et al., 2017),
• Meshing options for cladding liners,
• Two-phase flow heat transfer models, and
• Hydrogen pickup and hydride formation.

The existing BISON input files (Greenquist et al., 2025b) were updated to incorporate the applicable changes. Gadolinium was added to the fuel composition and hydrogen pickup/hydride formation models were added. However, not all of the new capabilities referenced above were included in this work. Since LM1 uses Zircaloy-2, the modern cladding materials were not necessary. Thermal boundary conditions were supplied by VERA and TRACE, so the two-phase flow model was not included. Finally, cladding liners were not included due to observed numerical instabilities they caused in the simulations. This may have some impact on cladding mechanical predictions.

4 LBLOCA simulation description

The development of the normal-operation simulations in VERA and BISON and the transient TRACE model have been documented elsewhere (Greenquist et al., 2025b; Capps et al., 2025). Therefore, this section focuses on the development of the transient BISON simulation. However, it is important to note how the 945 rods were selected for this work. The VERA simulation included 191 assemblies, which represent one quadrant of the full core. Five rods were randomly selected from each assembly using a principle of orthogonal sampling whereby it was ensured that no two rods came from the same row or column of the assembly’s 10 × 10 grid. This resulted in a set of 955 rods representing all rod types, including partial lengths and unenriched. Of these 955 simulations, 10 did not run to completion during the normal operation portion of the simulation due to numerical convergence issues and were thus not included in further analysis. An analysis of these failed rods did not find any correlations to rod-specific inputs.

TRACE divides the full reactor core into a series of channels in which the six-equation, two-phase flow formulation is solved. The channels for the model are shown in Figure 4. Figure 4a shows how the channels map onto the full core. Channels 1 and 2 cover most of the three-quarters of the core that were not modeled in VERA. Channels 3 and 4 cover the assemblies on the core periphery. The quarter core that was modeled in VERA is divided into channels corresponding to the individual assemblies. A schematic of Channel 5, which corresponds to the assembly closest to the core center, is shown in Figure 4b, which shows how the subchannels map to individual rods within the assembly. Most of the rods are covered by subchannel 1, but the five that were selected for individual fuel performance analysis are represented by subchannels 2–6. The method of appending the TRACE predictions to the BISON simulation is discussed in Section 3.2.

Several other changes were made to the normal operation input files to convert them to transient input files. First, the previous plenum temperature model was replaced with a more advanced, recently-developed version. Two cladding failure models were added: the Chapman correlation (CC) (Powers and Meyer, 1980; Erbacher et al., 1982) and the strain rate criterion (SRC) (Di Marcello et al., 2014). The CC works by calculating a burst temperature based on an empirical correlation of hoop stress and cladding heating rate. The rod is predicted to fail if the current cladding temperature exceeds the burst temperature. The CC’s BISON implementation was modified for this work. The data upon which the correlation is based come from relatively high temperatures (Powers and Meyer, 1980). Therefore, a minimum burst temperature of 700 °C was added to prevent the correlation from straying beyond its valid temperature range. The SRC predicts burst if the hoop strain rate exceeds 100 h⁻¹.

The fuel was significantly modified compared to previous simulations of these rods (Greenquist et al., 2025b; Capps et al., 2025). Individual GNF rods contained multiple fuel compositions such that the rod’s fuel properties (i.e., density, enrichment, and gadolinium concentration) vary axially. Originally, these values were averaged to produce a uniform fuel rod. In this work, a separate block was meshed for each fuel composition along with a unique set of material blocks that match the fuel composition. The updated mesh is depicted in Figure 5. The composition values were taken from the VERA input files using a Python script. These changes were implemented using BISON’s built-in mesh generator. The cladding secondary creep model was changed from the Limback-Andersson formulation (Hoppe, 1991) to the LOCA-specific Erbacher formulation (Erbacher et al., 1982) to better predict creep during the portions of the simulation of most interest in the current evaluation. Fuel creep was added using the MATPRO FCREEP material model (Allison et al., 1993). The UO₂ thermal conductivity function was changed from the NFIR formulation (OECD and Nuclear Energy Agency, 1999) to the Ronchi et al. formulation (Ronchi et al., 2004) because it is more specific to the burnup and temperature conditions in this work than the NFIR model.

Fuel-cladding contact was modeled using a frictionless kinematic formulation. The frictionless formulation is known to cause unrealistically high hoop stress in the cladding as the fuel expands axially on the interior cladding surface. A frictional model would be more realistic. However, frictional models are significantly more computationally expensive and were not feasible for the number of simulations being performed in this work. The frictionless model is not expected to affect the cladding burst predictions because the excessive stresses occur at the top of the fuel rod, which is cool enough to remain below the CC’s burst temperature.

Gap heat transfer is calculated in a cylindrical geometry (Toptan et al., 2019) with the Brokaw fill gas thermal conductivity equation (Brokaw, 1958). Incremental formulations were used to determine the stress and strain profiles in the fuel and claddings meshes.

The UO₂ Young’s modulus and Poisson’s ratio were set to the BISON defaults of 2 × 10¹¹ Pa and 0.345, respectively. The Zircaloy values were set to 7.5 × 10¹⁰ and 0.3. UO₂ fragmentation and relocation were predicted using the ESCORE model (Oguma, 1983). Zircaloy thermal conductivity and specific heat were set to 16 W m^-1 K⁻¹ and 330 J kg^-1 K⁻¹, respectively.

Fission gas generation, diffusion, and release were governed by the Simple Integrated Fission Gas Release and Swelling (Sifgrs) model (Pastore et al., 2013; Pastore et al., 2014; Barani et al., 2017). Sifgrs includes modules for fission product generation, intra-granular bubble formation, itergranular bubble migration (due to diffusion and grain boundary sweeping), bubble growth and interconnection, and fission gas release. It includes several models for burst release during transients (Simon et al., 2023). For this work, the Pastore et al. transient microcracking model was used (Pastore et al., 2014).

5 Simulation results

The simulations ran on a BISON version compiled on 15 January 2026. The simulations were performed on INL’s Sawtooth cluster. Out of 945 individual simulations, 943 ran to completion without incident, typically in under 2 h of elapsed real time. The two unfinished rods timed out after 24 h. This is a completion rate of 98.7%, including rods that failed during the normal operation portion of the simulation.

To understand the transient behaviors of the fuel rods, it is important to understand the state of the fuel prior to the LBLOCA. Therefore, nine end-of-cycle (EOC) simulation predictions are shown in Figure 6. All predictions are compared to the rod-average burnup and are shaped and colored according to the number of irradiation cycles and HBLUA status. These predictions include the peak fuel temperature (PFT), peak cladding temperature (PCT), peak cladding hoop strain (PHS), the locations (from the bottom of the cladding) of PCT and PHS, LHGR, fission gas release (FGR), plenum pressure, and void volume (also known as plenum volume). Note that these values are all peaks and are all measured while the reactor is at full operating power. They should not be compared to average values or post-irradiation-examination measurements.

The rod-average burnup ranged from 14.2 to 66.1 MWd/kgU. The results show clear behavior differences between fresh, once-burned, and twice-burned rods. Fresh and once-burned rods are located in the central core and thus have higher bundle powers, LHGRs, and temperatures. Twice-burned rods are at the periphery, where the bundle power, LHGR, and temperature are lower. The cladding initially expands due to thermal expansion, but as burnup increases the strain decreases due to creep-down effects until it contacts the expanding fuel and begins to expand again. It is common for once-burned rods to have higher hoop strains, more FGR, and higher plenum pressures than those of twice-burned rods because the most aggressively burned rods are removed after two irradiation cycles rather than three, so the twice-burned population is biased towards rods that experience relatively lower burnups during the first two cycles. Moreover, twice-burned rods experience lower LHGRs and temperatures due to their positions on the core periphery.

The LoPHS results in Figure 6 warrant additional discussion. A significant number of the peak hoop strains occur slightly above or slightly below the active fuel region. In both cases this signifies that the cladding hoop strain in the fuel region is negligible. Cladding outside the active fuel region experiences lower temperatures and lower neutron fluxes, which results in lower creep-down. These locations then become the LoPHS by default. Furthermore, in cases where the peak hoop strain occurred at multiple locations, the Python script used to analyze these results defaults to the lowest location. The lowest location was consistently selected as the default because plotting multiple values for a single rod or plotting averaged values would add confusion to the results.

Figure 7 shows the same data as Figure 6 but colored according to the four rod types from Figure 1 (unenriched, SPLR, LPLR, and FLR). Once again, there are clear groupings based on rod length. SPLRs tend to have lower LHGRs, lower temperatures, and lower plenum pressures. They also have smaller void volumes to match their smaller fuel volumes. LPLRs have higher burnups, tend towards higher LHGRs and temperatures, and have the highest PHSs. Interestingly, the rod with the highest FGR percentage is a SPLR, despite SPLRs having lower average LHGRs and peak fuel temperatures than LPLRs and FLRs. The rod in question had the highest LHGR and fuel temperature of the SPLRs modeled, and it is postulated that due to its small size, it had a more uniform power and temperature distribution than the longer rods and thus experienced more uniform FGR.

Time-dependent predictions of the considered physical quantities for the 943 rods are shown in Figures 8–16. Since the time scales for normal operation and the LBLOCA are so different, the two periods are plotted separately. Figure 8 shows the LHGR. During normal operation, LHGR tends to be highest at the beginning of cycle (BOC) and slowly decrease until the EOC. It tends to jump suddenly as control blades are adjusted. The reactor is scrammed during the LBLOCA, so the LHGR drops by about 99% almost immediately. Figure 9 shows the peak fuel temperature. During normal operation, it generally mirrors the LHGR, but during the LBLOCA the temperature drops to the 250 °C–425 °C range, rises slightly due to a lack of cooling, peaks at 240 °C–560 °C at 100 s, and then drops to a stable temperature range of 75 °C–280 °C as the emergency core cooling system successfully cools the core. Figure 10 shows the peak cladding temperature. During normal operation, the cladding temperature is stable in the 280 °C–350 °C range. During the LBLOCA, the temperature increases to 240 °C–540 °C at 100 s as the fuel and cladding temperatures equalize without cooling, then drops to 80 °C–135 °C at 200 s. Previous studies that used a single-code approaches predicted peak PCTs in the range of 700 °C–1000 °C (Trivedi et al., 2012; Nikitin et al., 2016; Hwang, 2021). However, the peak PCT predicted by BISON is about 540 °C. This is lower than the temperature cutoff implemented in the CC, so no cladding bursts can be predicted by that model. There are several possible explanations for this temperature discrepancy, including differences in the fuel and core designs evaluated in each case, differences in code methods, and the current multiphysics approach requiring fewer conservative assumptions.

The FGR predictions are shown in Figure 11. The FGR variation among similar rods is significant. Some rods go well into their third cycle before experiencing any meaningful FGR, whereas others reach 3% or higher within the first 40 days. Many rods experience burst release during reactor startup steps due to the inclusion of the transient FGR model. The peak FGR is 12.7% at 1,950 days. Note that the percentage of FGR can decrease as fission gas is produced but not released. The FGR during the LBLOCA is negligible. It stays steady in the 0%–11.6% range throughout. This is attributed to the low fuel temperature, which does not activate the empirical microcracking model used in this work (Pastore et al., 2014; Barani et al., 2017). Figure 12 shows the plenum temperature. The predictions are similar to the peak cladding temperatures except that the temperature drops sooner during the transient due to coolant coming from the top of the core. Figure 13 shows the plenum pressure. Fresh rods have initial pressures between 2 and 3 MPa, but due to FGR, the pressure range increases to 2.5–5 MPa by the end of their first irradiation cycle. However, aggressively burned once-burned rods have pressures up to 8.9 MPa by the time of the LBLOCA. During the LBLOCA, the pressure follows the corresponding temperature behavior, which is expected because there is not significant FGR and the cladding volume is stable.

Figure 14 shows the peak cladding hoop strain. Hoop strain is an important prediction because it is the basis of the SRC burst model. During normal operation, the cladding expands thermally, then creeps down. Creep down is caused by the coolant pressure pushing in on the cladding. Eventually the cladding contacts the fuel, at which point it may begin to expand again as the fuel swells. The peak cladding hoop strain during normal operation is 1.1%. By EOC, the strain varies between 0% and 0.8%. During the LBLOCA it expands to up to 1.6% at 100 s but contracts as the fuel cools to −0.2%–0.8%.

The void volumes are shown in Figure 15. The three rod lengths also have different plenum lengths and, therefore, different initial void volumes. To account for this, the values are normalized to their initial volumes. The void volumes quickly shrink to about 77%–92% of their original sizes due to thermal expansion of the fuel. Throughout the irradiation the void continues to decrease due to fuel swelling and cladding creep down. At EOC, the void volumes are 60%–78% of their original sizes. During the LBLOCA, the void volumes expand due to fuel cooling to ranges of 67%–87% of their original sizes.

The hydride concentrations in weight ppm are shown in Figure 16. The initial concentration was assumed to be 18 ppm by weight (Gamble et al., 2025). As the rods heated up, the hydrides dissolved and the hydrogen existed as a solution within the cladding. During normal operational shutdowns, the hydrogen again precipitated into hydrides. Negligible amounts of hydrogen were picked up by the cladding, so the total hydrogen content remained at 18 ppm. During the LBLOCA, the cladding cooled too quickly for hydrides to form, so it remained in solution despite the low temperature. Hydrides would be expected to form over longer timescales.

By comparing the predictions of the present work to a previous study (Greenquist et al., 2025b), the effects of the new BISON capabilities enumerated in Section 3.3 could be measured. It was found that individual rod predictions were affected (e.g., Gd-bearing rods saw temperature increases), but the effects on the whole-core distributions were minimal.

Finally, neither the CC nor the SRC predicted any cladding ruptures during the simulation. Test cases with artificially elevated temperatures were used to ensure that the models were working properly. Due to this uniform result, no fragmentation models or statistical analyses were needed to calculate an FFRD susceptibility range. Note that these results are theoretical and no comparisons to actual or experimental LOCA data have been made.

As mentioned in Section 1, results from a previous version of these simulations have been published (Greenquist et al., 2025b). That work did not include the new BISON capabilities listed in Section 3.3. Therefore, the effects of the new BISON capabilities on the simulations can be quantified by comparing them against the previous results. Select EOC prediction distributions of select values are shown in Figure 17. It shows the peak fuel temperature, FGR, and plenum pressure. These were the three predictions that were most affected by the updated capabilities. Both distributions include the same selection of rods with identical input and boundary conditions. The only differences are the new BISON capabilities and other updates to existing BISON and the BISON version used to run the simulations. The range of predicted values only changed minimally, but the shapes of the distributions were moderately affected.

6 Conclusion and future work

To predict a BWR’s susceptibility to FFRD during an LBLOCA, a set of 945 simulations constituting a statistically representative sample of rods in the LM1 BWR were run using a multiphysics framework that includes the codes VERA, TRACE, and BISON. The BISON simulations also incorporated several BWR-specific models that were recently developed and are in the process of being fully validated and incorporated into BISON.

Cladding integrity was assessed using the CC and SRC models already included in BISON. Both failure models predicted that no rods would burst in the core despite large variations in burnup, temperature, plenum pressure, cladding strain, and other conditions that can affect cladding integrity. The multiphysics framework used to achieve them is specific to the as-modeled operating conditions and safety system response of LM1 and should not be generalized to other reactors.

Legacy fuel performance methods typically involve selecting rods with the highest burnups or operating temperatures. This is sufficient for identifying limiting behaviors. However, the methodology used in this work can evaluate the overall fuel performance and susceptibility to cladding burst and FFRD more accurately by including a more complete and holistic view of fuel behavior throughout the core. The predicted fuel behaviors followed some general trends, such as a reduction in cladding strain due to creep and temperature clustering based on the number of irradiation cycles and the rod lengths. However, individual rod performance varied significantly based on individual irradiation histories and other factors. These rod-specific variations cannot be adequately captured with single-rod techniques.

Additional work is needed to continue to improve these models and FFRD predictions and quantify their uncertainties. First, several other updated BISON models are still being developed, tested, validated, and deployed, such as fuel relocation due to fracture and high-burnup structure. These should be implemented once they are fully developed. The cladding liner mesh option will also be included in future work once the numerical instabilities are resolved. In addition, several of the new and updated models apply equally well to PWR simulations (Greenquist et al., 2023)—especially the empirical-bounds-based temperature limit that was applied to the CC. Therefore, the PWR simulations will be re-run to determine whether these updates affect the FFRD susceptibility predictions. Finally, uncertainty propagation in the multiphysics framework has not been studied. A study to determine how uncertainty in reactor conditions and simulation input values propagate across the three codes is an important step to determining the accuracy of these results. Performing such a study in a computationally-cost-effective manner could help the nuclear energy industry make informed design and operation decisions.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnuen.2026.1822290/full#supplementary-material

Supplementary material Supplementary material at the source

Glossary

BOC Beginning of Cycle BWR Boiling water reactor CC Chapman Correlation EOC End of cycle FFRD Fuel fragmentation, relocation, and dispersal FGR Fission gas release FLR Full-length rod GNF Global Nuclear Fuel HBLUA High-burnup lead use assembly INL Idaho National Laboratory LBLOCA Large-break loss-of-coolant accident LHGR Linear heat generation rate LM1 Limerick Unit 1 LM2 Limerick Unit 2 LOCA Loss-of-coolant accident LoPCT Location of peak cladding temperature LoPHS Location of peak hoop strain LPLR Long partial-length rod LWR Light water reactor MOOSE Multiphysics Object-Oriented Simulation Environment NEAMS Nuclear Energy Advanced Modeling and Simulation NFIR Nuclear Fuel Industry Research ORNL Oak Ridge National Laboratory PCT Peak cladding temperature PDE Partial differential equation PFT Peak fuel temperature PHS Peak hoop strain PWR Pressurized water reactor SPLR Short partial-length rod SRC Strain Rate Criterion TRACE TRAC/RELAP Advanced Computational Engine TRISO Tristructural isotropic VERA Virtual Environment for Reactor Applications

Footnotes

Edited by: Gert Van den Eynde, Belgian Nuclear Research Centre (SCK CEN), Belgium

Reviewed by: Paul Van Uffelen, European Commission, Germany

Gašper Žerovnik, Bundesamt für die Sicherheit der nuklearen Entsorgung (BASE), Germany