press room
 

CFD vortex modeling around pillars in rivers

 
 

Hydraulic structures installed in rivers are often supported on pillars, which under certain conditions generate vortex systems in the flow. Structural problems can occur if local flows cause systematic movement of sediment near the foundations of a pillar. A combination of physical tests and 3D CFD modeling is required to provide engineers with sufficient details of the flow to optimize their structural designs.

Generally, depending on the shape of the pillar and the Reynolds number, separation occurs in the stream and two types of vortex systems are formed. For pillars with blunt leading edges, a “horseshoe” vortex is generated around the pillar. If the Reynolds number of the flow is between 140 and 300,000 a “wake” vortex system is established with separating vortices generated on alternate sides of the pillar (Von Karman vortex street).

We compared STAR-CD calculations with physical model testing. For the latter we used a flow channel (800 cm long and 134 cm wide) with a blunt leading-edge pillar (width 28 cm, length 126 cm) installed on an asymmetric flow bed. The flow rate was a constant 108 l/s at a water depth at the pillar of 32 cm.

 

 
 

 

Under steady-state conditions the flow velocities were measured at points parallel to the pillar (measuring section 1) and a transverse to the pillar (measuring section 2) 10 cm below the surface level, using a 3D-flow acoustic doppler velocimeter probe.

The corresponding STAR-CD simulation was performed using a 3D mesh of 11,000 cells, with an edge length of about 15 cm. The free surface was modeled with STAR-CD’s Volume of Fluid (VOF) method, and a High Reynolds Number k-epsilon model for the turbulence effects, figure 1.

For measuring section 1, the STAR-CD calculations lay within the ± 6 % standard deviation of the measurements. For section 2, the STAR-CD calculations again lay well within the experimental uncertainty. However, the latter had increased to ± 25 % due to the greater influence of the bottom geometry, figure 2. Qualitatively, both the horseshoe vortex and the wake vortex systems were correctly predicted, as shown in figure 3.

In summary, STAR-CD was able to accurately predict all aspects of these flows, and in at least asmuch detail as the experiment. It is therefore proving its worth as a valuable tool for structural engineers working in riverine industries. For further information please contact either tobias.huettebraeucker@gmx.de or tobilin@web.de

 
 
OpenCube Drop Down Menu (www.opencube.com)