Improrement of PIV Measurement Accuracy and Identification of Structures for Turbulent Flows

用粒子成像测速仪（PIV）测量了沿1 m宽、4 m高的铜板垂直受热表面诱导的湍流自然对流边界层。用双脉冲Nd: YAG激光系统发射的光片照射边界层流动，并在边界层流动中播撒烟雾。利用科学研究资助项目购买的大容量存储设备直接记录CCD摄像机拍摄的几分钟时间序列视频图像，并采用互相关法和亚像素算法计算瞬时速度矢量和湍流统计量。结果发现，由于获得了足够的时间序列数据，PIV的湍流统计数据与热线和冷线的湍流统计数据大致一致，而这在没有大容量存储设备的情况下是无法实现的。然而，在边界层外围区域，观测到不正确的速度矢量，并且不正确矢量出现的频率增加，因此需要进一步提高测量精度。在实际可视化测量的基础上，对平面上随时间发展的湍流边界层进行直接数值模拟（DNS），建立了可视化图像数据库，验证了PIV测量的准确性。将模拟得到的浓度分布作为种子粒子的亮度分布，生成基准图像，并对其进行处理，计算速度矢量。然后，将计算得到的矢量与直接从DNS得到的矢量进行比较，分析了激光片宽、脉冲注入间隔和图像处理算法对PIV测量的不确定度的影响。通过比较，确定了在自然对流边界的三维湍流行为中，不正确的速度矢量肯定会出现间歇区，激光片宽对计算的速度矢量没有影响，最佳脉冲注入间隔为3 ~ 4秒，应相对扩大探测窗口的大小，以便用互相关法计算准确的速度矢量。这些结果将在近期内提交给有关流量测量和可视化的国际期刊。少

A turbulent natural-convection boundary layer induced along a vertical heated surface made of a 1 m wide and 4 m high copper plate was measured with a particle image velocimetry (PIV). The boundary layer flow was seeded with smoke and illuminated by light sheets discharged from a double pulse Nd : YAG laser system. Time-series video images captured by a CCD camera for several minutes were directly recorded with a mass-storage device purchased by Grant-in AID for Scientific Research, and instantaneous velocity vectors and turbulence statistics were calculated with a cross-correlation method and a sub-pixel algorism. In consequence, it is founded that turbulent statistics with the PIV were approximately consistent with those obtained with a hot-and cold-wires due to the acquirement of adequate time-series data, which had not been realized without a mass-storage device. However, incorrect velocity vectors were observed and frequency appearance of the incorrect vectors increased in the int … More ermittent outer region of the boundary layer, and thus the advanced improvement of measurement accuracy was required.In addition to the actual visualization measurement, database of visualized images were established with a direct numerical simulation (DNS) for a time-developing turbulent boundary layer on a flat plat to confirm the accuracy of PIV measurement. Benchmark images were created by looking upon the concentration distribution derived from the simulation as the luminance distribution of seeded particles, and were processed to calculate velocity vectors. Then, the calculated vectors ware compared with those obtained directly from the DNS and the uncertainty of PIV measurement was accessed for the effects of laser-sheet width, pulse injection interval and image processing algorism. As the result of such comparisons, it is ascertained for the three-dimensional turbulent behavior in the natural-convection boundary later that incorrect velocity vectors certainly apt to appear intermittent region, the laser-sheet width has no effect on the calculated velocity vectors, the optimum pulse injection interval was 3-4 seconds and the size of interrogation windows should be relatively enlarged to calculate accurate velocity vectors with a cross-correlation method. These results will be submitted to the international journal concerning flow measurement and visualization in the near term. Less