Implicit large eddy simulation of ship airwakes

Drikakis

Youngs

et al. 2012

Journal of Turbulence

Self Cite

This paper presents a numerical study of a single-shocked turbulent mixing layer using high-order accurate implicit large-eddy simulations (ILES) for low-and high-amplitude (linear and non-linear) perturbations and for a reshocked turbulent layer. It investigates the differences in flow physics between these three cases, examining a recent proposition that single-shocked high-amplitude initial perturbations can be employed to model a reshocked turbulent layer. At early times, the shocked high-amplitude perturbation has high levels of mixing; however, at later times, it grows in an almost identical manner to the low-amplitude case. Despite exploration of several choices of scaling to map the reshocked results to the single-shocked study, a satisfactory agreement could not be reached and mixing parameters remain disparate. The conclusion is that a single-shock interaction with a high-amplitude perturbation is not a good representation of reshock of a turbulent mixing layer.

Section: Methodsmentioning

confidence: 99%

Physics of the single-shocked and reshocked Richtmyer–Meshkov instability

Drikakis

Youngs

et al. 2012

Journal of Turbulence

Self Cite

“…The solvers will be referred to by the abbreviations in the first column of Table 1. ) for the turbulent kinetic energy, k, for which closure is obtained through the algebraic expression for the turbulent eddy viscosity, µ t , in Equation 6.…”

Section: Governing Equations and Numerical Methodsmentioning

confidence: 99%

“…4 It has been established that Reynolds-Averaged Navier Stokes (RANS) simulations are insufficient for capturing the unsteady flow features that impact on pilot workload, 5 and may be insufficient for capturing the time-averaged flow gradients that affect the control margins. 1 Hence, much recent work has used ILES 6,7 and Detached Eddy Simulation (DES) 8,9 for simulation of the ship airwake due to their inherent ability to simulate unsteady flows.…”

Section: Introductionmentioning

confidence: 99%

A Study of LES Methods for Simulation of Ship Airwakes

Linton

46th AIAA Fluid Dynamics Conference

Widjaja

2016

The ship airwake is one of the most important components of a ship-helicopter dynamic interface simulation as the airwake determines many of the operating limits during dynamic interface operations. As part of the development of a dynamic interface simulation we have compared different Large Eddy Simulation (LES) methods for simulating ship airwakes using two test cases: a cube in a channel and a simplified frigate shape. The cube case was used to compare a compressible Implicit Large Eddy Simulation (ILES) solver (Flamenco) with compressible and incompressible standard LES solvers (OpenFOAM). The simplified frigate shape case compares LES with and without wall modelling with ILES and provides validation over a range of wind-over-deck angles. The results are presented as comparisons of time-averaged velocity and pressure profiles with experimental data and flow field visualizations. The comparisons between LES techniques may be used to inform simulation design choices for flow around ships and other sharp-edged bluff-bodies.

“…This algorithm has been validated extensively in a wide range of flows from applied aerodynamics through to turbulent mixing problems. 7,15,[15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30] …”

Section: Governing Equations and Numerical Methodsmentioning

confidence: 99%

Impact of Computational Domain Size in Simulations of Homogeneous Decaying Turbulence and Mixing Layers

22nd AIAA Computational Fluid Dynamics Conference

2015

An effective simulation of homogeneous decaying turbulence (HDT) or a turbulent mixing layer (TML) requires a computational domain which is larger than the integral length scale such that the turbulent length scales are not constrained. This paper investigates for two fundamental flows the point at which the size of the computational domain impacts the accuracy of the simulation. It is shown through simulations from 64 3 − 1024 3 that for HDT the prediction of integral lengths becomes inaccurate as it approaches 10% of the box size, however kinetic energy is reasonably predicted with even just one or two modes at low wavenumbers. For the TML, the key errors arise from the lack of statistical averaging and truncation of the layer rather than a direct impact of the box size on the flow physics.