## Flow visualization and fluid-structure interaction of tornado-like vortices

##### Abstract

A Ward-type tornado simulator has been built using a configuration of 16 slotted jets instead of a rotating screen to create the required far field circulation needed to produce a tornado4ike vortex. Flow visualization data, velocity data and pressure data were all obtained using the simulator. The produced vortices observed ranged from a laminar, rope-like, single-celled vortex to a turbulent, much larger diameter, two-celled vortex. Helium bubbles were used to visualize the vortices in the convergent region of the tornado simulator. At a=0.5, the low swirl ratios (the ratio of the tangential flow rate to the updraft flow rate) calculated were s=2.23 and at a=1 s=1.51. The high swirl ratios calculated were s=8.03 at a=0.5 and s=6.72 at a=1. The swirl ratios calculated are unique to the TTU TVS II.
The initial vortex configuration in the TTU TVS II was that of a single-celled vortex. During flow visualization, as the swirl ratio was Increased, a breakdown bubble was observed moving down the vortex core region toward the surface of the simulator. Once the breakdown bubble has traversed the vortex core to the surface of the simulator, the vortex is defined as two-celled. The TTU TVS II was capable of producing single-celled and two-celled vortices.
Pressure data was obtained on cubical and cylindrical models that were positioned at various radial locations within the simulator. The models were also subjected to moving tests through the TTU TVS II In order to compare the stationary data to the moving data. Using the pressure data, non-dimensional force coefficients were calculated and contour plots of the force coefficients on the cube and cylinder were generated for the stationary tests while, for the moving tests, specific points on the models were chosen, and the force coefficients at these points were plotted as a function of position In the TTU TVS II. The stationary tests show that both the cube and the cylinder models experience flow regimes at different points In the TTU TVS II similar in pattern to those induced by boundary layer-type flows, but mainly towards the outer regions of the simulator in low swirl cases (s=2.23 and s=1.51). Also for the cylinder, the contour plots indicate a horseshoe vortex forms around the cylinder. At the center of the simulator, both the cylinder and the cube disrupt the flow field significantly, and at this point, the flow field is very complex and at the present time the experimental equipment and data are not sufficient to quantify the flow In this region. The leading and trailing edges of the roof as well as the leading and trailing sides of each model were chosen and force coefficients were calculated
and plotted as a function of radial position In the TTU TVS II for the moving tests. Each of these moving tests had approximately the same trends for the leading edge and side and the trailing edge and side with a few exceptions. The stationary test data followed the trends of the moving test data In most cases tested. This would mean that less significance could be placed on the much more complicated moving tests and more significance on the less complicated stationary tests In future testing
Limited statistical analysis was also performed on the obtained data sets. This showed that standard deviation for all cases Is very small, so the distribution should be concentrated towards the center of the normal distribution. Most of the skewness values are negative Indicating the normal distribution is skewed to the right of the centerllne and slow, Infrequent variations In pressure below the mean. Very high kurtosis values like the ones shown for the center of the roof of the cylinder at the center of the simulator indicate an Increase In the high-frequency content of the fluctuating pressure signals read.