Validation of a Computational Fluid Dynamics Method Using Horizontal Dry Cask Simulator Data (NUREG/CR-7274)

On this page:

• Publication Information
• Abstract

Download complete document

• NUREG/CR-7274 (PDF – 6.37 MB)

Publication Information

Manuscript Completed: November 2020
Date Published: December 2020

Prepared by:
Kimbal Hall, P.E.

Alden Research Laboratory, Inc.
Holden, MA 01541

Abdelghani Zigh, Senior Technical Advisor, U.S. NRC

Michelle Bales, NRC Project Manager

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

Availability Notice

Abstract

Applicants submit spent nuclear fuel dry storage cask designs to the U.S. Nuclear Regulatory Commission (NRC) for certification under Title 10 of the Code of Federal Regulations Part 72, "Licensing requirements for the independent storage of spent nuclear fuel, high-level radioactive waste, and reactor-related greater than Class C waste" [1]. The NRC staff performs its technical review of these designs in accordance with 10 CFR Part 72 and NUREG-2215, "Standard Review Plan for Spent Fuel Dry Storage Systems and Facilities," issued April 2020 [2]. To ensure that the cask and fuel material temperatures of the dry cask storage system remain within the allowable limits or criteria for normal, off normal, and accident conditions, the NRC staff performs a thermal review as part of the technical review.

Recent applications have increasingly used thermal-hydraulic analyses using computational fluid dynamics (CFD) codes (e.g., ANSYS-Fluent) to demonstrate the adequacy of the thermal design. The applicants are also looking to license casks with decay heat close to 50 kilowatts (kW), resulting in a peak cladding temperature (PCT) close to the temperature limit of 400 degrees Celsius (C) (752 degrees Fahrenheit (F)) suggested in NUREG-2215 [2]. It is part of CFD best practice guidelines to present CFD predictions for the PCT or any other target variable supported by an uncertainty quantification (UQ) to provide assurance and confidence that the obtained margin is adequate. For this reason, the NRC Office of Nuclear Material Safety and Safeguards asked the NRC Office of Nuclear Regulatory Research to be part of a dry cask storage system numerical modeling validation study to assist it in making regulatory decisions to ensure adequate protection for storage and transportation casks [3].

The NRC recognizes that CFD using finite volume is one of the methods for the applicants to perform dry cask thermal modeling and to demonstrate adequate margins with the temperature limits suggested in NUREG-2215 [2]. Additionally, when demonstrating the thermal margins, it is valuable to quantify the uncertainty in the simulation result as a function of the computational mesh and simulation inputs. When finite volume CFD is applied carefully through the use of CFD best practice guidelines in NUREG-2152, "Computational Fluid Dynamics Best Practice Guidelines for Dry Cask Applications," issued March 2013 [4], and UQ, there will be confidence and certainty in the obtained margins. As a participant in this CFD validation exercise, the NRC led the effort to include UQ and followed the CFD best practice guidelines in NUREG-2152 [4].

The numerical modeling validation study was based on data collected at Sandia National Laboratories (SNL) in an experiment sponsored by the U.S. Department of Energy (DOE) on a horizontal dry cask simulator (HDCS) [3]. The HDCS simulates one prototypic 9x9 boiling-water reactor fuel assembly, which uses electrical resistance heaters to simulate the decay heat. The power input in the experiment ranged from 0.5 kW to 5 kW. The fill gas was either helium or air. The fill gas pressure was either 100 kilopascals or 800 kilopascals (helium only). Extensive temperature measurements were made throughout the cask, including throughout the axial length of the fuel assembly and on different wall surfaces of the HDCS inside and outside the pressure vessel representing the dry cask canister. The measurements also included PCT and induced air mass flow rate.

Ten tests were conducted in the HDCS at SNL. Two tests were considered "open," in that the measured temperatures and airflow rates were provided to the participant modelers to benchmark and validate their CFD models. Following submission of the results of these two open cases, modelers predicted the air mass flow rate and temperature at 21 locations within the case for the other eight cases. The simulation of these eight cases was considered "blind" in that the modelers submitted their simulation results before the experimental results were released.

This report discusses the validation and UQ of the HDCS CFD model using the experimental data gathered by SNL. Uncertainty quantification follows the procedures outlined in American Society of Mechanical Engineers (ASME) Verification and Validation (V&V) 20-2009, "Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer" [5]. Sources of uncertainty examined in the analysis include iterative uncertainty, spatial discretization, experimental uncertainty, and uncertainty due to eight input parameters including ambient temperature, emissivity values, decay heat, fuel region porous media hydraulic resistance (i.e., a porous media model was used to model the fuel region), ventilated air straighteners porous media hydraulic resistance (i.e., porous media were used to model straighteners that were applied to obtain uniform airflow before mass flow rate measurements), sensitivity to certain gaps, orientation angle with respect to gravity, and external heat transfer coefficients. This report documents the HDCS CFD model validation exercise and is not intended to be used for regulatory guidance.

The CFD results and experimental data for PCT and location of PCT agreed very favorably for all the collected cases within the calculated validation uncertainty using best estimate CFD analysis that includes the most likely scenario analysis supplemented by UQ. The uncertainty in the wall emissivity values, and hence radiation heat transfer between surfaces within the fuel assembly and the pressure vessel, was found to be the principal source of uncertainty in the HDCS experiment. The PCT validation uncertainty was calculated to be higher for the air fill cases than for the helium fill cases. This result is attributed to the larger role that radiation heat transfer plays with air fill gas than it does with helium fill gas. Air thermal conductivity is almost one order of magnitude less than the conductivity of helium, so radiative heat transfer contributes more to the overall heat exchange to compensate for the minimal heat conduction of the air. The PCT validation uncertainty in helium cases varied between 6 to 20 degrees C (11 to 36 degrees F) for decay heat varying between 0.5 to 5 kW respectively. In the air fill cases, the PCT validation uncertainty varied between 8 to 40 degrees C (14 to 72 degrees F) for decay heat between 0.5 to 5 kW respectively.

Even though the validation uncertainty in this experiment is much less than the one obtained for the DOE cask demonstration documented in NUREG/CR-7260, "CFD Validation of Vertical Dry Cask Storage System," issued May 2019 [6], a smaller value would be preferred, especially for the air fill cases. Cask vendors have been submitting applications within PCT margins of approximately 10–20 degrees C (18–36 degrees F) from the specified temperature limit of 400 degrees C (752 degrees F) in NUREG-2215 [2]. To demonstrate that the CFD modeling process can produce this level of accuracy, the validation uncertainty should be minimized to the desired margin of 10–20 degrees C (18–36 degrees F) or less. Therefore, it is difficult to classify this experiment as CFD-grade for this explicit purpose with such a low temperature margin, particularly for the cases with air fill gas. If the cask vendors use a worst-case scenario and proven documented conservative input to perform their CFD predictions, they can avoid the UQ for input CFD variables. As such, the margins obtained by applicants would be compared with only the numerical uncertainty. The CFD predictions presented in analyses of this report, using CFD best practice guidelines documented in NUREG-2152 [4], showed that numerical uncertainty for all the cases varied between 0.4–4.9 degrees C (1–9 degrees F). Consequently, thermal models using documented conservative input parameters as usually used by applicants using CFD modeling guidelines documented in NUREG-2152 [4] would generally be deemed acceptable with adequately demonstrated thermal margins.

The results show that an ANSYS-Fluent thermal model using NUREG-2152 CFD best practice guidelines [4], including simulating the fuel assembly using a porous media approach, can demonstrate the safety of the storage of spent nuclear fuel by accurately predicting the PCT with accurate model inputs. This report also looks at the quality of the data collected in the HDCS experiment document [3] using the calculated validation uncertainty. The HDCS experiment was designed to minimize the validation uncertainty—a key factor and the basis for thermal model validation. Consequently, a well-validated thermal model will enable thermal reviewers to have confidence in the predictions, even with decreased margins.

The discussions and conclusion section of the Electric Power Research Institute (EPRI) DOE cask demonstration (demo) project report validation exercise [7] indicated that "CFD dry cask thermal models are generally conservative." However, the NRC cask demonstration CFD analysis as documented in NUREG/CR-7260 [6] showed that the EPRI report statement is acceptable only when conservative analysis is used. NUREG/CR-7260 [6] showed that if a best estimate model (i.e., most likely or base case scenario supplemented by UQ as done in this report) was used, the PCT and temperature field overpredictions can be explained in detail. In that thermal model round robin involving the DOE cask demonstration project [6] [7], geometrical uncertainties (i.e., gaps) were the main reasons for temperature prediction deviations as explained in detail by NUREG/CR-7260 [6], which showed a PCT validation uncertainty of 62 degrees C (112 degrees F) for the DOE cask demonstration exercise. This unusually high validation uncertainty was mainly caused by the uncertainty in the knowledge of the fluid gaps existing in the cask geometry. The validation exercise contained in this report, NUREG-2238, "Validation of a Computational Fluid Dynamics Method using Vertical Dry Cask Simulator Data," issued June 2020 [8], NUREG/CR-7260 [6], and the SNL validation synthesis report [9] [10] showed that CFD thermal model as described in this report, using CFD best practice guidelines as documented in NUREG-2152 [4] resulted in PCT predictions that agreed very favorably with the experimental data within the calculated validation uncertainty.

Three different modeling groups participated in the numerical validation and benchmarking exercise, including the NRC, Pacific Northwest National Laboratory, and Empresa Nacional del Uranio, S.A., S.M.E. [9]. The SNL validation synthesis report [9] states the following about the models used to validate the HDCS cases as presented in this report: "Based on the combined RMS error results, NRC model offered the best overall fit to the experimental data." The report also adds the following:

NRC was the only institution that accompanied the base case model with uncertainty quantification…the NRC submission was an extensive effort that captures the effect of introducing simulation uncertainty bounds in the comparison of model results to experimental data. The method, which is derived from the ASME verification and validation approach in ASME V&V 20-2009, "Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer" [5], is explored in the validation uncertainty section in Chapter 3 of the synthesis report [9], which shows how the uncertainty quantification can be used to provide a better measure of model prediction accuracies. Overall, this model validation method takes both measurement and simulation uncertainties into account and serves as an example of how the model validation uncertainty quantification can be further explored. By definition, NRC thermal model is considered validated if the combined RMS error normalized by the validation uncertainty is less than 1, and this was shown to be the case.