Michael K. Stoellinger

Dynamic unified RANS-LES simulations of high Reynolds number separated flows

The development of hybrid RANS-LES methods is seen to be a very promising approach to enable efficient simulations of high Reynolds number turbulent flows involving flow separation. To contribute to further advances, we present a new, theoretically well based, dynamic hybrid RANS-LES method, referred to as DLUM. It is applied to a high Reynolds number flow involving both attached and separated flow regimes: a periodic hill flow is simulated at a Reynolds number of 37 000. Its performance is compared to pure LES, pure RANS, other hybrid RANS-LES (given by DLUM modifications), and experimental observations. It is shown that the use of this computational method offers huge cost reductions (which scale with Re/200, Re refers to the Reynolds number) of very high Reynolds number flow simulations compared to LES, it is much more accurate than RANS, and more accurate than LES, which is not fully resolved. In particular, this conclusion does also apply to the comparison of DLUM and pure LES simulations on rather c...

Simulation of Turbulent Channel Flow Using a Linear and Non Linear realizable Unified RANS-LES Model

Large Eddy Simulation (LES) is computationally expensive at high Reynolds numbers (Re), especially for the accurate resolution of the near-wall region in wall-bounded flows. Hence, hybrid turbulence models have been developed to reduce the computational cost of LES. In the hybridmodeling approach, the near-wall region is modeled using Reynolds-averaged Navier-Stokes (RANS) methods while regions away from the wall are modeled using LES. The most commonly used hybrid models do not accurately predict the mean streamwise velocity and the Reynolds normal stresses for attached flows without empirical modifications. Hence, there is a need for the development of unified turbulence models which can be used continuously as LES or RANS methods for both attached and separated flows. The current study investigates the performance of a unified turbulence model derived using stochastic analysis. Turbulent channel flow is used as the test case to evaluate the performance of the unified turbulence model. The current study provides encouraging results for the use of the unified turbulence model considered to investigate complex turbulent flows.

A RANS-LES Study of Swirling and Non-Swirling Jets

Studying swirling turbulent jet flows is important for both application-driven reasons and increasing our fundamental understanding of turbulence. Although experimental and numerical studies of swirling jet flows provided some insights into the processes causing the mixing enhancement of momentum and scalars, a definite understanding is still not achieved. The paper describes the development of a combined RANS/LES method for the investigation of the structure of swirling turbulent jet flows. Comparisons with experimental data provide evidence for the suitability of this combined RANS/LES approach.

Swirl-Induced Mixing Enhancement in Turbulent Jet Flows

The paper reports the results of numerical investigations of swirling turbulent jet flows by large eddy simulation (LES). Inflow data are provided on the basis of Reynolds-averaged Navier-Stokes (RANS) simulations. In particular, instantaneous inflow data are produced by a forcing that generates correlated noise. In this way, the characteristic length and time scales of inflowing instantaneous turbulent eddies are in consistency with the corresponding RANS profiles imposed at the inlet. The performance of several subgrid-scale stress models is studied. The mechanism of swirl effects is investigated on this basis. It is shown that swirl breaks apart the typical ring structurs of non-swirling turbulent jets into two modes: a helical mode and streamwise braid structures. The interaction of these two modes generates less organized turbulence structures. Correspondingly, swirl may imply a significant increase of the efficiency of the turbulent mixing of scalars.

Atmospheric Boundary Layer Studies with Unified RANS-LES and Dynamic LES Methods

The paper reports the results of applications of nonlinear dynamic LES and unied RANS-LES models to neutrally stratied atmospheric boundary layer (NABL) simulations. The advantages of dynamic LES methods are their accuracy and the ability to account correctly for anisotropy and backscatter. The advantage of unied RANS-LES models is their ability to provide results like LES for much lower computational cost. The models are applied to two ows. First, a turbulent Ekman layer with a smooth wall at the bottom is considered at a low Reynolds number Ref = 1140. The results are compared with available DNS results. Second, the results of simulations of the NABL on coarse grids are presented. On coarse grids the subgrid-scale models have to provide a signicant contribution, this means they are much more important than in ne grid resolution studies.

Stochastic-Based RANS-LES Simulations of Swirling Turbulent Jet Flows

Abstract Many turbulent flow simulations require the use of hybrid methods because LES methods are computationally too expensive and RANS methods are not sufficiently accurate. We consider a recently suggested hybrid RANS-LES model that has a sound theoretical basis: it is systematically derived from a realizable stochastic turbulence model. The model is applied to turbulent swirling and nonswirling jet flow simulations. The results are shown to be in a very good agreement with available experimental data of nonswirling and mildly swirling jet flows. Compared to commonly applied other hybrid RANS-LES methods, our RANS-LES model does not seem to suffer from the ’modeled-stress depletion’ problem that is observed in DES and IDDES simulations of nonswirling jet flows, and it performs better than segregated RANS-LES models. The results presented contribute to a better physical understanding of swirling jet flows through an explanation of conditions for the onset and the mechanism of vortex breakdown.

Reynolds Stress Closure in Hybrid RANS-LES Methods

The feasibility of using the elliptic blending Reynolds stress model (EB-RSM) in hybrid RANS-LES methods is investigated in this paper. The advantage of the EB-RSM is that it does not use any geometrical wall distance or wall normal vector information which makes it well suited for application in flows with complex wall geometries. A slight modification to the original EB-RSM is proposed to improve the performance for flows with separation. The model is also extended to a sub-grid scale model for fully resolved LES and several possibilities for use as a hybrid RANS-LES model are presented. The RANS EB-RSM model performed overall well in plane channel flows, the periodic hill flow and the flow over a NACA 4412 airfoil with trailing edge separation. In LES, the EB-RSM model provided very good results in a plane channel flow at low Reynolds number. When used as a zonal hybrid RANS-LES model, the EB-RSM displayed a significant log-layer mismatch although the relevance of the modeled and resolved stresses switched right at the prescribed interface.

Monte Carlo Simulations of Turbulent Non-premixed Combustion using a Velocity Conditioned Mixing Model

Non-premixed turbulent combustion in a laboratory scale flame (Delft III flame) is studied using a statistical description at the one-point one-time joint velocity—scalar composition probability density function (PDF) level. The PDF evolution equation is solved using a stochastic Lagrangian Monte Carlo method. The PDF equation requires a so called micro-mixing model for closure and the performance of two micro-mixing models is investigated. The Interaction by Exchange with the Mean (IEM) micro mixing model is the most commonly adopted model. The IEM model was developed for the scalar PDF method and does not depend on velocity statistics. A physically more sound extension of the IEM is the Interaction by Exchange with the Conditional Mean (IECM) which involves mixing of the scalars towards mean values conditional on the velocity. Both models are applied in this work and it is shown that the IECM model does perform significantly better than the simple IEM model.

Unified RANS-LES Analysis of Turbulent Jets Covering Several Swirl Number Regimes

Many turbulent flow simulations require the use of hybrid methods because large eddy simulation (LES) methods are computationally too expensive and Reynolds-averaged NavierStokes (RANS) methods are not sufficiently accurate. However, a large variety of hybrid RANS-LES methods are currently in use, leading to the question of which method should be preferred. Desired properties of a theoretically optimal hybrid RANS-LES method are formulated here. It is shown that a unified RANS-LES model derived from a stochastic turbulence model has all the properties of a theoretically optimal hybrid RANS-LES method. The unified RANS-LES model is applied to turbulent swirling and nonswirling jet flow simulations to illustrate the difference to other commonly applied hybrid RANSLES methods. The unified RANS-LES results are shown to be in a very good agreement with experimental studies of nonswirling and mildly swirling jet flows. The unified RANSLES model does not suffer from the ’modeled-stress depletion’ problem that is observed in detached eddy simulation (DES) and improved delayed DES (IDDES) simulations of nonswirling jet flows. Regarding the simulation of high swirl number flows it is shown that the unified RANS-LES model performs better than segregated RANS-LES models, which suffer from their inability to correctly simulate the central vortex core, the precessing vortex core, and the mixing efficiency of passive scalars. The results presented contribute to a better understanding of swirl flows by the explanation of conditions for the onset and the mechanism of vortex breakdown.