Publications
Nuclear Systems Enhanced Performance (NSP) Program
Thermal Striping in LWR Piping Systems
L.-W. Hu, M.S. Kazimi, and A. Sonin
MIT-NSP-PR-014
Abstract
This report summarizes the progress of the MIT/TEPCO project entitled "Thermal Striping in LWR Piping Systemsa" during the second year. Progress during the first year was summarized in the MIT CANES report MITR-NSP-TR-007. Thermal striping has raised safety concerns due to incidents at some nuclear power plants. The current study investigates thermal striping at tee junctions of light water coolant systems through numerical simulations. A series of benchmark and sensitivity study was performed through the first and second year of this project.
Further improved results were obtained from FLUENT simulations during the second year due to several refinements. These include: use of a smaller time step (~ 1 ms vs. 10-50 ms), an improved sub-grid scale model, a refined tee junction geometry, and the use of a central differencing spatial discretization scheme. Sensitivity studies of these refinements and the flow velocity ratios are described in this report.
A new version of FLUENT, FLUENT6.0, was released in January 2002. All calculations that were performed since March 2002 used FLUENT6.0. Although modifications in FLUENT6.0 do not involve enhancements of LES subgrid models, one feature related to LES in FLUENT6.0 is the incorporation of the second-order accurate central-differencing scheme. The central-differencing scheme is available for the momentum equations when using the LES turbulence model. This scheme provides improved accuracy for LES calculations. A sensitivity study was performed to compare the calculated results using the second order upwind and the central differencing schemes. Comparisons of the normalized temperatures show that the central differencing scheme produces significantly higher temperature fluctuations than the secondary upwind scheme at all three azimuthal positions. It is also noted to have a significant effect in turbulent mixing which in turn promotes a more uniform temperature distribution in the main coolant flow down stream the tee junction. However, one problem associated with use of the central-differencing scheme is that it can produce unbounded solutions and can lead to stability problems. Overshoot and undershoot of the calculated coolant temperatures were both found when the central-differencing scheme was chosen.
Comparisons of the two sub-grid scale models offered by FLUENT, the Smagorinsky and the RNG sub-grid scale models, were performed. The RNG model has favorable comparisons with the experimental data for predicting the locations of the peak normalized fluctuating temperatures. It also showed better comparisons of normalized mean temperatures. A smaller RNG constant leads to higher fluctuating temperatures at all locations. This trend is similar to that of the of the sensitivity study of the Smagorinsky constant which was performed in the first ear.
Experiment results obtained at Hitachi seem to show a correlation of normalized mean and fluctuating temperatures with respect to the flow velocity ratios, i.e., the normalized mean/fluctuating temperatures plotted along the main pipe have similar trends of those with the same flow velocity ratio. Sensitivity studies were performed to simulate 4 experimental conditions, each two share the same flow velocity ratio. Calculated results of case#17 and #18 normalized mean temperatures have very good agreements with experimental data. The calculated peak fluctuating temperature locations are within about 0.5D to those obtained in experiments. The calculated normalized fluctuating temperatures are somewhat higher than measurements. Calculated normalized mean temperatures for cases#10 and #11 have good agreements with experimental data. The calculated peak fluctuating temperature locations are within about 2D to those obtained in experiments. The calculated normalized fluctuating temperatures are generally higher than the experiments. The trends of the peak normalized fluctuating temperatures of the same flow velocity ratio obtained from FLUENT simulations are similar to those of measurements.
