scholarly journals Computation of turbulent flow in a square duct: aspects of the secondary flow and its origin

1994 ◽  
Vol 23 (1) ◽  
pp. 157-176 ◽  
Author(s):  
F.M. Wang ◽  
Y.T. Chew ◽  
B.C. Khoo ◽  
K.S. Yeo
2007 ◽  
Vol 588 ◽  
pp. 153-162 ◽  
Author(s):  
MARKUS UHLMANN ◽  
ALFREDO PINELLI ◽  
GENTA KAWAHARA ◽  
ATSUSHI SEKIMOTO

A direct numerical simulation of turbulent flow in a straight square duct was performed in order to determine the minimal requirements for self-sustaining turbulence. It was found that turbulence can be maintained for values of the bulk Reynolds number above approximately 1100, corresponding to a friction-velocity-based Reynolds number of 80. The minimum value for the streamwise period of the computational domain is around 190 wall units, roughly independently of the Reynolds number. We present a characterization of the flow state at marginal Reynolds numbers which substantially differs from the fully turbulent one: the marginal state exhibits a four-vortex secondary flow structure alternating in time whereas the fully turbulent one presents the usual eight-vortex pattern. It is shown that in the regime of marginal Reynolds numbers buffer-layer coherent structures play a crucial role in the appearance of secondary flow of Prandtl's second kind.


1964 ◽  
Vol 19 (3) ◽  
pp. 375-394 ◽  
Author(s):  
E. Brundrett ◽  
W. D. Baines

Secondary flows in non-circular ducts are accompanied by a longitudinal component of vorticity. The equation of motion defining this component in a turbulent flow is composed of three terms giving the rates of production, diffusion and convection. Since the expression for production is the second derivative of Reynolds strees components, longitudinal vorticity cannot exist in laminar flow. For turbulent flow in a square duct the Reynolds stress tensor is examined in detail. Symmetry requirements alone provide relationships showing that the production is zero along all lines of symmetry. General characteristics of flow in circular pipes are sufficient to indicate where the production must be greatest. Experimental measurements verify this result and define the point density of production, diffusion and convection of vorticity. Data also indicate that the basic pattern of secondary flow is independent of Reynolds number, but that with increasing values of Reynolds number the flows penetrate the corners and approach the walls. A similar experimental investigation of a rectangular duct shows that the corner bisectors separate independent secondary flow circulation zones. Production of vorticity is again associated with the region near the bisector. However, there is some evidence that the secondary flow pattern is not so complex as inferred from the distortion of the main longitudinal flow.


1987 ◽  
Vol 53 (492) ◽  
pp. 2370-2376 ◽  
Author(s):  
Hideomi FUJITA ◽  
Hajime YOKOSAWA ◽  
Masafumi HIROTA ◽  
Satoru NISHIGAKI

1987 ◽  
Vol 109 (1) ◽  
pp. 43-48 ◽  
Author(s):  
J. W. Baughn ◽  
H. Iacovides ◽  
D. C. Jackson ◽  
B. E. Launder

The paper reports extensive connective heat transfer data for turbulent flow of air around a U-bend with a ratio of bend radius:pipe diameter of 3.375:1. Experiments cover Reynolds numbers from 2 × 104 to 1.1 × 105. Measurements of local heat transfer coefficient are made at six stations and at five circumferential positions at each station. At Re = 6 × 104 a detailed mapping of the temperature field within the air is made at the same stations. The experiment duplicates the flow configuration for which Azzola and Humphrey [3] have recently reported laser-Doppler measurements of the mean and turbulent velocity field. The measurements show a strong augmentation of heat transfer coefficient on the outside of the bend and relatively low levels on the inside associated with the combined effects of secondary flow and the amplification/suppression of turbulent mixing by streamline curvature. The peak level of Nu occurs halfway around the bend at which position the heat transfer coefficient on the outside is about three times that on the inside. Another feature of interest is that a strongly nonuniform Nu persists six diameters downstream of the bend even though secondary flow and streamline curvature are negligible there. At the entry to the bend there are signs of partial laminarization on the inside of the bend, an effect that is more pronounced at lower Reynolds numbers.


Author(s):  
Iztok Tiselj ◽  
Mohit Pramod Sharma ◽  
Boštjan Zajec ◽  
Nikola Veljanovski ◽  
Jan Kren ◽  
...  
Keyword(s):  

2014 ◽  
Vol 747 ◽  
pp. 186-217 ◽  
Author(s):  
S. Türk ◽  
G. Daschiel ◽  
A. Stroh ◽  
Y. Hasegawa ◽  
B. Frohnapfel

AbstractWe investigate the effects of superhydrophobic surfaces (SHS) carrying streamwise grooves on the flow dynamics and the resultant drag reduction in a fully developed turbulent channel flow. The SHS is modelled as a flat boundary with alternating no-slip and free-slip conditions, and a series of direct numerical simulations is performed with systematically changing the spanwise periodicity of the streamwise grooves. In all computations, a constant pressure gradient condition is employed, so that the drag reduction effect is manifested by an increase of the bulk mean velocity. To capture the flow properties that are induced by the non-homogeneous boundary conditions the instantaneous turbulent flow is decomposed into the spatial-mean, coherent and random components. It is observed that the alternating no-slip and free-slip boundary conditions lead to the generation of Prandtl’s second kind of secondary flow characterized by coherent streamwise vortices. A mathematical relationship between the bulk mean velocity and different dynamical contributions, i.e. the effective slip length and additional turbulent losses over slip surfaces, reveals that the increase of the bulk mean velocity is mainly governed by the effective slip length. For a small spanwise periodicity of the streamwise grooves, the effective slip length in a turbulent flow agrees well with the analytical solution for laminar flows. Once the spanwise width of the free-slip area becomes larger than approximately 20 wall units, however, the effective slip length is significantly reduced from the laminar value due to the mixing caused by the underlying turbulence and secondary flow. Based on these results, we develop a simple model that allows estimating the gain due to a SHS in turbulent flows at practically high Reynolds numbers.


Sign in / Sign up

Export Citation Format

Share Document