Large Signal-to-Noise Technique for Unsteady Pressure Measurements

Abstract

I order to obtain mean and fluctuating pressure measurements, many investigators perform their experiments while operating at the maximum speed of the wind tunnel. Hunt and Fernholz,' for example, point out that the speed should be in excess of 10 m/s in order to maximize the pressure signal and to minimize sound pressure contamination. With the I.I.T. Environmental Wind Tunnel, as with many tunnels, this may not be the most efficient operating condition. For example, if the sound level increases with velocity at a power larger than 2, or if the flow conditions in the tunnel are less desirable at high velocities, the operation at a low or moderate freestream speed may be more appropriate. One of the earlier and most valuable discussions of sound contamination in wind tunnels was presented by Batchelor. In wind tunnels that are not carefully designed to eliminate sound contamination, one is either restricted to mean pressure measurements or must accept the additional component in the signal. This contamination is especially critical in wind engineering applications where the frequencies of interest lie in that region where generated noise due to the fan, turbulence manipulators, and the tunnel geometry are most prominent. In these cases, a significant percentage of the unsteady pressure signal being measured will most likely be due to this additional component. Some wind engineers use this as an additional safety factor in predicting unsteady wind loads on buildings and structures. The technique presented here has been successfully used to remove this ever-present undesirable sound contamination from unsteady, low-level pressure signals. At I.I.T., we often utilize a freestream velocity of 3.8 m/s in the low-speed section of the Environmental Wind Tunnel for studying atmospheric flows around simulated structures and building models. With this comparatively low velocity, the mean and unsteady differential pressures on the model are of the order of 10 ~ 4 psid. The sound pressure level inside the tunnel is of the same order of magnitude, but because of its frequency content it dominates over the pressure fluctuations occurring on a model placed in the simulated atmospheric surface layer. This sound-saturated pressure signal results in erroneous rms readings which are insensitive to changes in the monitored port, model orientation, or test boundary layer. The percentage of the sound component present in the signal was determined through the overall rms of the signal and by the peak-to-peak value of its correlation function. A comparison of these for several sound conditions in the tunnel indicated that at least 96% of the sound pressure fluctuations were entering the system through the model ports. Thus, the most any sound insulation of the transducer and tubing could provide would be a 4% reduction. Narrow-band filtering of the sound component is not a satisfactory solution because it would also remove a substantial and relevant part of the pressure signal from the model. Therefore, a subtraction technique was adopted to remove the pressure signal due to the contaminant sound from the output of the transducer.