Global POD Modes From PIV Asynchronous Patches

Abstract

A method is proposed to obtain full-domain spatial modes based on Proper
Orthogonal Decomposition (POD) of Particle Image Velocimetry (PIV)
measurements performed at different (overlapping) spatial locations.
When performing robotic volumetric Particle Image Velocimetry (PIV) (Jux
et al., 2018), the large-scale mean velocity fields are estimated
merging several measurements performed at different (adjacent) locations
covered by the robot sequence. The proposed methodology leverages the
definition of POD modes as eigenvectors of the spatial correlation
matrix to obtain also large-scale modes. Performing measurements over
overlapping (50-75%) regions allows to approximate the correlation
matrix in the two adjacent domains. When applied over a sequence of
views, this method has the potential to deliver full-domain POD modes
spanning the volume covered by the robot sequence, even if different
regions are covered asynchronously. This methodology is particularly
well-suited for applications that seek to investigate large-scale flow
structures, whenever the dynamic spatial range (DSR) of the measurement
system does not allow to capture the whole domain at once. The
methodology is validated using a 2D experimental dataset of a turbulent
boundary layer, where patches are artificially created from splitting
the PIV measurements, later used as ground truth to assess the results.
Furthermore, we apply the technique to a 3D robotic volumetric PIV
experiment of the flow around a wall-mounted cube.