Advances in tomographic PIV

Abstract

This research deals with advanced developments in 3D particle image velocimetry based on the tomographic PIV technique (Tomo-PIV). The latter is a relatively recent measurement technique introduced by Elsinga et al. in 2005, which is based on the tomographic reconstruction of particle tracers in three-dimensional space from a small number of its projections obtained with digital cameras. Tomography is widely known in the medical diagnostics (e.g. computerized axial tomography) to inspect the human body. For PIV applications the problem is formulated as that of reconstructing the spatial distribution of sparse emitters (illuminated tracers). The present work initially surveys the state of advancement of the research conducted on this new measurement technique and the main bottlenecks and aspects to be improved are identified. The two major elements covered in this research are the 3D object reconstruction and the advanced analysis of the tracers motion. Concerning the first aspect, one of the recognized limitations is the exponential increase of ghost particles when a higher particle concentration is desired for high-resolution measurements (e.g. in turbulence studies). Although Tomo-PIV already outperforms other volumetric 3D techniques in terms of allowed particle tracers concentration, many efforts are constantly devoted to find ways to further increase the particle density. A novel concept is presented in this work that makes use for the first time of more than a single recording to increase the accuracy of tomographic reconstruction. This method considers that the moving particle field can be regarded as a solid object recorded from a moving imaging system. That is why the author refers to the concept of fluid tomography, whereby the two recordings of the same set of particles are “deformed back” to the same time instant when the particle tracers come to coincide and the ghost particles do not. The Motion Tracking Enhancement reconstruction technique (MTE) is described in Chapter 4 of this thesis. The validity of the MTE working principle is verified both numerically and by experiments. Its application in turbulent shear flows shows that the seeding density can be increased by a factor 4 (ppp=0.2) with respect to that currently practiced (ppp=0.05) without loss of accuracy. The focus is then set on techniques to increase the spatial resolution of velocity fields measured by tomographic PIV. The approach followed is that of locally adaptive interrogation volume, following the concept of non-isotropic resolution in PIV (Scarano 2003). The study shows a novel technique that exploits the additional degrees of freedom when adapting window shape and orientation in a 3D domain based on velocity gradient tensor invariants analysis (Chapter 5). It is shown that the measurement spatial resolution can be increased by a factor 2.5 and 1.5 across shear layers and in the core of a vortex respectively. The present work deals also with the advanced treatment of time-resolved Tomo-PIV experimental data (Chapter 6), where the accurate measurement of the velocity material derivative is of paramount importance to extract the instantaneous flow field pressure (pressure from PIV, van Oudheusden, 2013). The approach investigated here is based on the use of particle-tracking for a time-resolved sequence of 3D particle fields. Adopting a high-order polynomial basis for the particle trajectory reconstruction allows the reconstruction of long trajectories with a strong reduction of random error and nearly the complete elimination of the truncation error. The application to a 3D measurement of a transitional jet demonstrates the higher accuracy obtained for the estimate of fluid parcels acceleration and in turn of the instantaneous pressure field. The work is concluded with a synthesis of the advances obtained in this field, followed by a perspective towards the most significant upcoming developments for the tomographic PIV technique.