Volumetrische Strömungsvermessung: Eine Implementierung von Shake-The-Box

Abstract

This thesis deals with volumetric flow field measurements using particle imaging with multiple cameras. Focus of interest is the 30 Lagrangian-Particle-Tracking (LPT) technique Shake-The-Box (STB) and 30 Partic/e-/mage-Velocimetry (30-PTV). Since the introduction of the Shake-The-Box algorithm by Schanz et al. in 2013 LPT and 30-PTV experienced a new revival. While the community focused on tomographic Partic/e-/mage-Velocimetry (Tomo-PIV} in the past and believed 30PTV to be unfeasible at high particle image densities, STB overcame this limitation with a clever combination of novel 30 triangulation a/gorithms (namely the lterative­Partic/e-Reconstruction /PR) and usage of temporal information. STB shows a range of advantages over Tomo-PIV. Especially, it allows ve/ocity and acceleration measurements with high precision and spatial resolution. STB also shows much /ower demands on the computing hardware and a faster processing. Since its proposal STB is actively developed at DLR in Göttingen. However, in recent years it did not experience huge progresses but underwent a lot of detai/ improvements. In this master thesis the completely new framework PySTB for 30 partic/e tracking is developed and implemented. Ouring the development nearly all associated a/gorithms were extensively revised, rewritten and adapted for higher flexibility, improved robustness and faster processing. PySTB is written in C++ and Python and shows the best of both worlds; C++ provides high performance and Python makes the deve/opment of new a/gorithms and techniques really simple and fast. For that reason the STB working group at OLR already migrated to PySTB. In this thesis PySTB is used to reimplement several former algorithms and also develop new ones. The reimplementation of /PR and STB showed to be more robust and faster on the evaluation of synthetic and experimental data. The new algorithm Two-Pulse STB allows to use 30-PTV techniques for the evaluation of two-pulse measurements what was reserved to Tomo-PIVin the past. Thereby, Two-Pulse STB shows significantimprovements over the results that were achieved with previous reconstruction methods (tested at the synthetic Case C of the 4th internationalP/V challenge and an experimental data set). Further, the Online Volume-Self­Calibration is presented, which enab/es to simultaneously calibrate the camera system and reconstruct the particle positions and intensities, promising the possibility to use STB on measurements under tough circumstances, like strong vibrations. Also, some new techniques for flow-field estimation from noisy point clouds are presented (in addition, or as an alternative to the Partic/e-Space-Correlation). Many of the proposed modifications and the new a/gorithms will be discussed in this thesis.