Development of hybrid particle tracking algorithms and their applications in airflow measurement within an aircraft cabin mock-up

Obtaining reliable experimental airflow data within an indoor environment is a challenging task and critical in studying and solving indoor air quality problems. The Hybrid Particle Tracking Velocimetry (HPTV) system is aimed at fulfilling this need. It was developed based on existing Particle Tracking Velocimety (PTV) and Volumetric Particle Tracking Velocimetry (VPTV) techniques. The HPTV system requires three charge-coupled device (CCD) cameras to view the illuminated flow field and capture the trajectories of the seeded particles. By adopting the hybrid spatial matching and object tracking algorithms, this system can acquire the 3-Dimensional velocity components within a large volume with relatively high spatial and temporal resolution. Synthetic images were employed to validate the performance of three components of the system: image processing, camera calibration and 3D velocity reconstruction. These three components are also the main error sources. The accuracy of the whole algorithm was analyzed and discussed through a back projection approach. The results showed that the algorithms performed effectively and accurately. The reconstructed 3D trajectories and streaks agreed well with the simulated streamline of the particles. As an overall testing and application of the system, HPTV was applied to measure the airflow pattern within a full-scale, five-row section of a Boeing 767-300 aircraft cabin mockup. A complete experimental procedure was developed and strictly followed throughout the experiment. Both global flow field at the whole cabin scale and the local flow field at the breathing zone of one passenger were studied. Each test case was also simulated numerically using a commercial computational fluid dynamic (CFD) package. Through comparison between the results from the numerical simulation and the experimental measurement, the potential model validation capability of the system was demonstrated. Possible reasons explaining the difference between experimental measurement and model prediction were discussed