Prandtl number effects in abruptly separated flows : LES and experiments on an unconfined backward facing step flow
(2018)
Files
Thesis_SBuckingham.pdf
Open Access - Adobe PDF
- 14.98 MB
Details
- Authors
- Supervisors
- Abstract
- Liquid metal cooled reactors have been relying on CFD analyzes for their design and safety evaluation. However, the major challenge is the modeling of the turbulent heat transfer in the low Prandtl number coolant (Pr≃0.001-0.01). Appropriate RANS models have been developed but these require to be tested, calibrated and validated on a wide variety of flows to make them applicable to Liquid Metal Reactors. In this thesis, high-fidelity data is generated and analyzed in order to provide insight concerning the way heat is transported across the low Prandtl number coolant, crucial for the improvement of suitable RANS models. To represent simply the detached flow areas and wall interactions that dominate the upper plenum, an unconfined backward facing step is chosen as benchmark test-case. This work proposes an efficient approach to generate inflow turbulence by extending the temperature perturbation method (TPM) to all types of wall-bounded flows. The optimized TPM enables to minimize the recovery distance without the need to impose second order statistics. To evaluate the accuracy of the LES, PIV and probe measurements are performed in a He-Xe gas mixture (Pr≃0.2), which is close to the lowest achievable value for a transparent medium. This work has demonstrated the importance of integrating the probe limitations while processing the LES data, resulting in a meaningful and satisfactory comparison between the LES and the experiment. The LES results obtained at Pr=0.025, 0.2 and 0.71 provide a better understanding of low Prandtl number effects, while confirming the necessity of resorting to anisotropic turbulent heat transfer models for RANS.
- Affiliations
Citations
Buckingham, S. (2018). Prandtl number effects in abruptly separated flows : LES and experiments on an unconfined backward facing step flow. https://hdl.handle.net/2078.5/47810