Estimating wind velocity and direction using sparse sensors on a cylinder

Using finite pressure measurements on a cylinder, we are able to estimate both the oncoming wind speed and direction of uniform flow over a cylinder at Reynolds numbers 20 000<Re<120 000. While reduced-order methods, such as proper orthogonal decomposition with QR factorization, require at least nine sensors to estimate the oncoming wind speed and direction with <10% error, other methods, such as probabilistic approaches or curve-fitting, can achieve similar results with as few as five sensors. A utility function, based on the Kullback–Leibler divergence, is used to determine the locally optimal location of the sensors to accurately estimate inlet conditions. It was found that sensor arrangement also plays a significant role, with unevenly distributed sensors being preferable than evenly distributed sensors. These techniques, when paired with existing flow field estimation approaches, allow the user to predict the surrounding flow field from any oncoming direction.