Assessing the Ability of the DDES Turbulence Modeling Approach to Simulate the Wake of a Bluff Body

Abstract

Abstract:A detailed numerical investigation of the flow behind a square cylinder at a Reynolds number of 21,400 is conducted to assess the ability of the delayed detached-eddy simulation (DDES) modeling approach to accurately predict the velocity recovery in the wake of a bluff body. Three-dimensional unsteady Reynolds-averaged Navier-Stokes (URANS) and DDES simulations making use of the Spalart-Allmaras turbulence model are carried out using the open-source computational fluid dynamics (CFD) toolbox OpenFOAM-2.1.x, and are compared with available experimental velocity measurements. It is found that the DDES simulation tends to overestimate the averaged streamwise velocity component, especially in the near wake, but a better agreement with the experimental data is observed further downstream of the body. The velocity fluctuations also match reasonably well with the experimental data. Moreover, it is found that the spanwise domain length has a significant impact on the flow, especially regarding the fluctuations of the drag coefficient. Nonetheless, for both the averaged and fluctuating velocity components, the DDES approach is shown to be superior to the URANS approach. Therefore, for engineering purposes, it is found that the DDES approach is a suitable choice to simulate and characterize the velocity recovery in a wake.

Numerical Capture and Validation of a Massively Separated Bluff-Body Wake

“…13. The same order of magnitude in this over-prediction of total effective viscosities is also reported [16]. In contrast to the experimental data, the DDES results predicts the fluctuation peak slightly farther downstream.…”

“…At higher Reynolds numbers ( > 1.0 × 10 4 ), the wake shed by a bluff body can be described by several characteristic quantities. Benchmark conditions [8,[16][17][18][19][20][21] at = 21 400 showed that the time-averaged drag coefficient ranges approximately 2.1 to 2.4, with Strouhal numbers ranging 0.12 to 1.4. This shedding frequency is correlated to the vortex shedding pair that forms due to the roll-up of the shear-layer instabilities at the leading edges.…”

“…As they are of engineering importance, Unsteady-RANS (URANS) methods become unsuitable as fluctuations can be lost through the process of Reynolds decomposition ( = Φ + ). Wake behaviour can therefore become numerically damped, which is attributed to an excess of turbulent viscosity production [16]. LES techniques are among those that have been increasingly used for resolving the turbulent flows [17,18].…”

“…However, the two coarser Δ / = [0.05, 0.1] refinement grids produced similar results, with an approximate 0.1 / ∞ over-prediction in wake velocity recovery. A similar 0.05Δ/ grid resolution in the wake was also used with spatial (upwind-based) schemes that are second-order accurate, although with comparatively slightly better results [16]. Both examples successfully employed the hybrid method, by demonstrating reasonable accuracy in representing the time-averaged velocity recovery and its root-mean-square statistics for the wake centre.…”

Special Issue “Computational Aerodynamic Modeling of Aerospace Vehicles”

“…Active and passive flow control methods were studied to improve the intake performance. Finally, the article by Boudreau et al [13] investigates the use of large eddy simulations in predicting the flow behind a square cylinder at a Reynolds number of 21,400.…”