Experimental investigation of flow around 3D square cylinder using five-hole probe and neural network

In the present study, wake structure downstream a 3D square cylinder is experimentally investigated. The contours of mean velocity, total pressure, static pressure and vorticity are presented for the wake region. Five-hole probe is used for extracting the required wake properties and determining the flow structure. A novel neural network method is used for calibrating the five hole prob. The calibration map obtained by this method is compared with the conventional linear and 5th order polynomial surface fit algorithms. Based on the statistical parameters of calibration data, it is shown the neural network algorithm is more accurate and works faster than the other methods. Experimental results showed that with increasing the distance from the cylinder, tip and base vortices are weakened, the height of wake region decreases while its width increases simultaneously. The effects of free end shear layer on the flow structure are clarified in the results. Downwash flow from free end of cylinder and upwash flow from the bottom wall decrease the width of the wake region near top and bottom of cylinder in comparison with the mid-height region. The effect of free end shape of a finite square cylinder is also investigated. The shear layer separated from the free end plays an important role.