Turbulence occurs at high flow speeds and increases the stickiness of the flow. In turbulent flows the fluid undergoes small circular motions called eddies, which leads to loss of energy. But due to the stickiness, turbulence can be used to carry away heat from hot surfaces like gas turbine blades. The blades have cooling passages with ribs to enhance the heat transfer rate. In my thesis I have presented the effects of rotation (Coriolis acceleration) and ribs on turbulence in ribbed channel flows, where a channel is basically a rectangular duct with a very high width-to-height ratio. The ribs are placed on the channel walls transverse to the flow, i.e., like speed-breakers to the traffic on the roads. Navier-Stokes equations govern the fluid flow including the turbulence phenomenon and I solved these equations numerically without using any empirical models for turbulence.
In the first part of my study, I consider how the height of the ribs affects turbulence in non-rotating flows. In the second part I consider rotating flows at moderate rotation rates. Rotation axis is along the rib length and the effect of Coriolis force is to transfer turbulence energy between the other two directions—along-stream and normal-to-wall directions. Hence with increasing rotation rates, the Coriolis force makes one channel side more and more turbulent (pressure side) and the other side less and less turbulent (suction side). At moderate rotation rates the eddies on the pressure side form along-stream bands, which vanish at further higher rotation rates. At these higher rotation rates, I witnessed a travelling wave on the suction side, which I discuss in the third part of my study. I found that ribs are necessary on the suction-side wall for the wave to appear, irrespective of the pressure-side condition. These waves can counteract the Coriolis effects and slightly increase the heat-transfer rate on the suction side.
Direct numerical simulation (DNS) technique used in this study is currently feasible for simple flow configurations and for relatively lower fluid speeds. However, the technique is highly accurate for studying the nature of turbulence at all the length scales (various eddy sizes) and therefore the results obtained from a DNS study are generally used to benchmark turbulence models used in industrial-scale simulations.