Thermal Analysis of Horizontal Storage Casks for Extended Storage Applications (NUREG/CR-7191)

On this page:

Download complete document

Publication Information

Manuscript Completed: February 2014
Date Published: December 2014

Prepared by:
Kaushik Das, Debashis Basu, and Gary Walter

Center for Nuclear Waste Regulatory Analyses
Southwest Research Institute®
6220 Culebra Road
San Antonio, TX 78238-5166

S. Gonzalez, NRC Project Manager
A. Velazquez-Lozada, NRC Project Manager
G. Zigh, NRC Technical Lead

Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission
Washington DC 20555-0001

Availability Notice


A computational fluid dynamics (CFD) model that can be used to reliably predict temperature distributions for long-term storage was developed for the ventilated horizontal dry storage system containing pressurized water reactor (PWR) fuel at the Calvert Cliffs Independent Spent Fuel Storage Installation (ISFSI). The model was developed systematically by (i) constructing a three dimensional (3-D) model for two horizontal modules at this ISFSI, (ii) comparing model results with measured temperature data, (iii) performing a parametric analysis to assess the impact of model inputs on temperature predictions, (iv) using the developed model to predict temperature distributions of critical components for a storage period of 300 years, and (v) conducting numerical uncertainty analyses of the solutions using the grid convergence index (GCI) method.

Three-dimensional (3-D) numerical models were developed using the commercial CFD software FLUENT, Version 14.5 [ANSYS Inc. 2012]. The horizontal storage modules (HSM) modeled in the analysis are the HSM–1 and HSM–15 units at Calvert Cliffs. Each of the storage modules contains a horizontal dry shielded canister (DSC) loaded with 24 fuel assemblies. The concrete storage units and DSC internal structures such as spacer plates, tie rods, fuel assemblies, and supporting rails were included in the model. Individual fuel rods were not explicitly modeled; instead, the volume within each fuel assembly was represented as a porous medium with specified frictional and inertial flow resistances to helium movement. An orthotropic temperature-dependent equivalent thermal conductivity was used to model conductive and radiative heat transfer in the porous zone. The discrete ordinate (DO) model was used for radiation, while the low Reynolds number k-ε model with buoyancy was used to simulate turbulence.

The baseline simulations were compared to measured temperature data for the HSM–1 and HSM–15 units from June 27 and 28, 2012 [Suffield et al. 2012]. The steady state temperature distributions on different components of the storage modules, and the corresponding air and helium circulation patterns, were calculated using ambient conditions representative of the measurement period. The models predicted a high-temperature region on the upper half of the curved surface of the DSC shell, resulting from a combination of temperature-driven natural convection flow of the helium coolant inside the DSC and buoyancy-driven upward flow of heated air around the DSC. Computed temperatures at selected locations on the DSC shell were compared with measured data. The simulated temperatures for both modules were, higher than the measured values, but the agreement between the modeled and measured data was better near the end cap of the DSC shell. Aspects of the temperature data collection suggest that there were a number of uncertainties involved with the method used for temperature measurement and the observed values were not representative of a normally operated closed module.

Sensitivity analyses of the simulated results included evaluations of the near wall mesh refinement, turbulence model, porous media resistance, insolation, and evaluation of the heat transfer coefficient using an extended domain analysis. The study showed that near wall mesh refinement above the baseline mesh had minimal impact on the temperature distribution. Therefore, the baseline mesh was deemed adequate to perform the benchmark studies, other parametric analyses, and the 300-year thermal evolution study. The turbulence models evaluated included the standard k-ε model, low Reynolds number k-ε model, standard SST k-ω model, and SST k-ω model with buoyance effects. Among the studied turbulence models, simulated results with the low Reynolds number k-ε model predicted the lowest temperatures. The study of porous media resistance showed that as long as some flow resistance exists within the porous zone of the fuel assemblies, the temperature distribution is not very sensitive to the specified porous media resistance; however, the fundamental pattern of temperature distribution is altered if resistance is completely eliminated from this zone. Accordingly, reasonable resistance coefficients are necessary in this zone in order to accurately capture the temperature field. The sensitivity study also showed that insolation impacts the temperatures calculated on the internal components by a few degrees. To obtain an independent verification of the heat transfer coefficient on the exposed concrete surface of the module, a simulation was performed using an extended domain that included the surrounding atmosphere. Results indicated that the heat transfer coefficients used for the baseline simulation were reasonable.

The 300-year simulations for both the HSM–1 and HSM–15 configurations showed that (i) the drop in maximum cladding temperature along the profile is relatively rapid in the first 100 years and becomes more gradual thereafter and (ii) the minimum cladding temperature varies by a small amount as time progresses. All the component temperatures showed an asymptotic decline towards the ambient temperature over time.

The GCI method developed by Roache [1998, 2003] and adopted in the ASME standard V&V 20-2009 [Celik et al. 2008; ASME 2009] was used to calculate the observed order of accuracy and associated numerical uncertainty for the simulated results. Four levels of computational grids were used in the GCI analysis. The simulated results obtained from the different grid levels showed little variation. The apparent order of accuracy calculated using different combinations of solutions was not constant, which can be attributed to minor numerical oscillations in the results and possibly round-off errors. An alternative estimate of numerical uncertainty was made using an order of accuracy of 1 and a higher factor of safety. From this estimation, it was found that the GCI is less than 4 percent, indicating that there was no significant deviation of the predicted variables due to mesh refinement.

Page Last Reviewed/Updated Thursday, March 25, 2021