Development of an advanced Hybrid (U)RANS/LES turbulence models using Scale Adaptive Simulation (SAS) method
Date
2022Metadata
[+] Show full item recordAbstract
Turbulence is a ubiquitous phenomenon occurring in most practical fluid flows. The accurate prediction of turbulent flows is evident in optimizing and developing highly efficient modern fluid systems. Several turbulence models are available to investigate practical flows numerically and are broadly described among Reynolds-Averaged Navier- Stokes (RANS) and Large Eddy Simulation (LES) models. The RANS models are de- vised to model all the length scales of the associated flow field and provides often inaccu- rate results in complex flows such as flow separation and reattachment. In contrast, LES is primarily formulated to directly resolve the large scales and model the effect of the smaller scales. However, the mesh constraints become stringent in the near-wall region resulting in an exorbitant cost of LES methodology. In such cases, hybrid (U)RANS-LES models play a vital role as they perform as RANS in the near-wall region and LES in the regions away from the wall. However, currently available hybrid (U)RANS-LES model such as Improved Delayed Detached Eddy Simulation (IDDES) does not provide consis- tent performance under different fluid flow cases and suffers from several issues such as log-layer mismatch in attached flows, sensitivity to anisotropic grid design, and need for an ad-hoc turbulence generation mechanism for separated flows.
Hence, this dissertation comprehensively assesses the IDDES methodology in dif- ferent flow cases. The response of grid designs/resolutions and different Reynolds num- bers are investigated. This study helps determine critical criteria to assess the ideal hybrid (U)RANS-LES model performance. Further, this investigation enables a deeper under- standing of various shortcomings encountered in the IDDES methodology and delivers better comprehension of the ideal hybrid (U)RANS-LES model. Finally, to overcome the shortcoming of the IDDES methodology, an advanced hybrid (U)RANS-LES model is devised.
This dissertation introduces a modified hybrid (U)RANS-LES model, sensitized to non-equilibrium flow conditions such as flow separation. The numerical experiment of Decay of Homogeneous Isotropic Turbulence (DHIT) provides proof of accurately pre- dicting the non-equilibrium effects and returning to equilibrium at a high Reynolds num- ber. Further, the model is tested on the attached flows (turbulent channel flow) and sepa- rated flows (periodic hill and hump flow). Next, a new length scale equation (L-equation) is devised to mitigate the issue of log-layer mismatch (LLM) in attached flows, which provides an accurate description of the gray region (region which is neither (U)RANS nor LES). Further, a new automatic triggering mechanism is formulated utilizing the dy- namic behavior of the L-equation, and in conjunction with an advanced (U)RANS model (k − ε − ζ − f ), provides an accurate prediction of the turbulent channel flows.
Additionally, the modified hybrid (U)RANS-LES model provides accurate pre- dictions on the highly anisotropic grid designs. In the case of separated flows, the mod- ified model accurately captures the separation and reattachment locations and provides a consistent response to different grid designs/resolutions. Moreover, for hump flow, the IDDES model requires ad-hoc turbulence fluctuations to sustain turbulence resolving ca- pabilities. The modified hybrid (U)RANS-LES model develops turbulent fluctuations arising from underlying flow instabilities after the flow separation and accurately predicts first and second-order flow quantities.
Table of Contents
Introduction -- Effects of near wall modeling in the improved-delayed-detached-eddy-simulation (IDDES) methodology -- Towards understanding and improving the transition between unsteady Reynolds-Averaged Navier-Stokes (URANS) and large eddy simulation (LES) in hybrid URANS-LES methodology -- Conclusion and outline of the dissertation
Degree
Ph.D. (Doctor of Philosophy)