Paritosh Mokhasi

Optimized sensor placement for urban flow measurement

In this paper, we discuss a novel approach to the description of atmospheric flows in urban geometries. Our technique is based on the method of proper orthogonal decomposition (POD). We devise a method that enables us to compute the time-varying coefficients of a Karhunen–Loeve expansion of the urban flow field using knowledge of instantaneous velocity data taken at a minimum number of locations simultaneously. Using the POD basis functions and these velocity data, we solve a set of linear equations which gives us an estimate of the exact expansion coefficients. This method allows us to compute estimates for all coefficients thereby enabling us to reconstruct a close approximation to the flow field which is optimal in a certain sense. A quantitative comparison of the approximate coefficients with the coefficients of an exact Karhunen–Loeve expansion shows that the method works very well. Our method provides a practical approach to reconstructing the flow field using a minimum amount of information.

Estimation of Spatio-temporal Evolution of Complex Velocity Fields using Indirect Measurement Models

Understanding and controlling the dispersion of pollutants and contaminants in urban areas has become a major concern in the last decade. This study addresses the need for a better understanding of the dynamics involved in the flow field around complex geometries and the need for novel techniques to sense and predict its temporal evolution. Experimentally, it is not possible yet to track the temporal evolution of three dimensional structures in turbulent flows at high Reynolds number and with good spatial resolution. Typically, sparse measurements are used in wind forecasting models for updating and prediction via variational data assimilation. To improve upon this method, an experimental investigation combining Particle Image Velocimetry and Static Pressure measurements is carried out to study the airflow around wall mounted obstacles in a turbulent boundary-layer. A companion study is performed on a Large Eddy Simulation of the flow around a wall mounted cube in a channel which provides complementary information. The method of Proper Orthogonal Decomposition is used to obtain a reduced order representation of the flow field and is used in the construction of indirect measurement models based on sparse measurements from wall pressure sensors. Ultimately the combination of these with state-space models should provide more robust dynamical models.