Characteristics of large volume tomographic particle image velocimetry using helium filled soap bubbles in forced and thermal convection

Abstract

The characteristics of large-scale tomographic PIV in thick measurement volumes (MVs), which are necessary for measuring large-scale structures in convective air flows, are discussed. In thick MVs the number of illuminated particles outside the reconstructed volume increases under different circumstances. The complete intensity of these particles needs to be back projected in the volume during the reconstruction process. By conducting numerical simulations the influence of the back projected intensity of these particles on the measurement accuracy is investigated. Further, in order to make possible processing of large amounts of data a parallel implementation of the tomographic PIV algorithms was developed. In this course an implementation of the simultaneous multiplicative reconstruction algorithm as recently proposed by Atkinson and Soria (Exp Fluids 47:553-568, 2009) for tomographic PIV is discussed and validated for large-scale tomographic PIV. A strategy of processing these routines on high performance computer clusters is also reviewed. Large-scale tomographic Particle Image Velocimetry (tomographic PIV) using Helium filled soap bubbles as tracer particles has been applied to convective air flow in a long rectangular convection cell. The mean flow fields in forced and mixed convection of air in a convection cell for a Reynolds number of 345, 530 and 800 and for the same Re at a Rayleigh number Ra = 1.65*10^8, respectively are analyzed.