CHARACTERISTICS OF FLOW OVER A CIRCULAR CYLINDER AT Red=140,000

Abstract

In the present study, we conduct both numerical and experimental studies to investigate the characteristics of flow over a circular cylinder at Red = 140,000, where Red is the Reynolds number based on the cylinder diameter d and free-stream velocity. Large eddy simulation (LES) is conducted for numerical investigation, and direct force and particle image velocimetry (PIV) measurements are conducted for experimental investigation. The drag and base pressure coefficients from present numerical and experimental studies agree well with each other. The instantaneous flow over a circular cylinder exhibits various flow structures such as laminar separation at the cylinder surface, shear-layer after the separation, vivid small-scale vortices and vortex shedding in the wake. In the near wake of circular cylinder, it is observed that a vortical structure distorted in spanwise direction and weak vortex shedding intermittently occurs, inducing relatively low drag and lift fluctuations. INTRODUCTION A circular cylinder is a representative bluff body, found in the structures such as wind generator, lamp post, etc. The flow over a circular cylinder at subcritical Reynolds number exhibits various flow phenomena including boundary layer, separation, shear layer evolution and vortex shedding in the wake. Therefore, there have been numerous studies to understand the flow over a circular cylinder experimentally and numerically. However, the experimental results from several researchers show non-negligible variations among themselves (Bearman, 1969; Achenbach and Heinecke, 1981; West and Apelt, 1982; Farell and Blessmann, 1983; Cantwell & Coles, 1983; Szepessy and Bearman, 1992). Furthermore, some attempts were also made to predict the flow around a circular cylinder using large eddy simulation (LES; Fröhlich et al., 1998; Breuer, 2000), but those studies only dealt with the effects of numerics such as grid system, resolution and subgrid-scale models and could not evaluate the prediction performance from numerical simulation due to the scatter in the experimental results. Therefore, in the present study, we investigate the flow over a circular cylinder at a subcritical Reynolds number (Red = 140,000) from both the experimental and numerical approaches. EXPERIMENTAL & COMPUTATIONAL DETAILS August 28 30, 2013 Poitiers, France WAK1A 2 Figure 1. Experimental set-up for (a) force and surface pressure measurements; (b) PIV measurement. Experimental setup The present experiment is conducted in a closed-type wind tunnel whose size of test section is 900 mm × 900 mm. The turbulence intensity is lower than 0.3% at the free-stream velocity of 20 m/s. Figure 1a shows the schematic diagram of the present experimental set-up for force and surface pressure measurements, consisting of a circular cylinder, end plate, load cell, pressure holes, scannivalve, and manometer. The cylinder is made of ABS resin with the diameter d = 70 mm. The aspect ratio of the cylinder is 11.4. Boundary-layer thickness on the tunnel-wall is about 25 mm at 20 m/s. The end plate, suggested by Stansby (1974), is installed with the distance 35 mm from the top and bottom of the test section to remove the effect of the boundary layer. The blockage ratio of present setup is 7.8%. Drag force on the cylinder is measured directly using two load cells installed at the both end of cylinder and averaged for 80 seconds. There are nine holes for pressure measurement in the spanwise direction. The distance between pressure holes is 70 mm and each hole is connected directly to the scannivalve with Teflon tube. The free-stream velocity varies from 20 to 55 m/s, corresponding to the Reynolds numbers based on the freeFigure 2. (a) Schemtaic diagram of the computational domain and boundary conditions; (b) typical mesh near circular cylinder. Every 4th grid is shown. stream velocity and the cylinder diameter, Red = 90,000 – 260,000. Figure 1b shows the schematic diagram for particle image velocimetry (PIV) measurement. The PIV system consists of a Nd:Yag laser of 120 mJ, a CCD camera of 2048 × 2048 pixels resolution and a delay generator. The velocity measurement is conducted in xy-plane at the center of the cylinder span. A 60 mm lens mounted on a digital camera is used to provide a field of view, whose size is 130 mm × 130 mm. Computational details In the present study, LES of the flow over a circular cylinder is conducted at Red = 140,000. The governing equations of an unsteady incompressible viscous flow for LES are the filtered continuity and Navier-Stokes equations. For the time integration, a fully implicit method based on the Crank-Nicolson method is used. For the spatial discretization, we use a hybrid scheme (Yun et al. 2006): a third-order QUICK scheme is used in laminar flow region before separation to prevent the dispersion error caused by the central difference scheme, and the second-order central difference scheme is used elsewhere. The no-slip boundary condition on the cylinder surface is realized by the immersed boundary method by Kim et al. (2001) in the Cartesian coordinate system. The subgridscale stress for LES is modelled using the dynamic global model (Park et al., 2006; Lee et al., 2010). Figure 2a shows the schematic diagram of computational domain, coordinate system and boundary conditions used in this study. The computational domain is -15≤x/d≤15, 25≤y/d≤25, and 0≤z/d≤π, where x, y, and z denote the streamwise, transve rse, and spanwise directions, respectiAugust 28 30, 2013 Poitiers, France WAK1A 3 Table 1. Flow statistics at Red = 140,000.