气泡羽流运动是自然界和工程领域普遍存在且十分重要的现象，对气泡羽流内部结构的实验观测有助于提升对气泡多相流运动规律的理解和预测，具有重要的科学意义和应用价值。数字图像处理技术的发展为气泡羽流试验观测提供了巨大的发展空间。本文构建了多相浮射流试验研究平台，开发了基于预估校正的气泡阴影图像处理方法，并开展了气泡羽流试验，得到了较高时空分辨率的连续相PIV图像和离散相阴影图像，基于图像处理和统计分析研究了气泡尺寸对气泡羽流内部结构的影响。论文构建的多相浮射流试验平台能够产生不同的气泡羽流，可以实现多组分的同步测量，可以有效识别和持续追踪不同组分共存的离散相阴影图像。为了更准确地测量气泡阴影图像中高度重叠的气泡的大小和速度，论文开发了基于预估校正的气泡阴影图像处理方法，该方法根据气泡运动的连续性假设，利用已知气泡的精确信息对相邻帧中该气泡的形态和运动信息进行预估和校正。对于人工合成气泡图像序列，成功地捕获了所有气泡，包括单帧图像中无法识别的完全重叠气泡；对于随机抽样的气泡羽流试验图像，气泡的有效识别率也超过95%。验证结果表明本方法可以识别传统单帧方法无法识别的完全重叠气泡，可以有效降低气泡检测的漏检率，提高气泡大小和位置的准确性，对研究不同尺寸气泡的微观运动，测量气泡数密度、气泡大小、气体体积分数等统计参数的分布具有重要意义。论文应用多相浮射流试验平台和气泡阴影图像处理方法进行了静止无分层环境下的气泡羽流试验研究，设计了两组不同气泡尺寸分布的气泡羽流试验，研究了不同尺寸气泡在羽流中的分布规律和运动特征，分析了气泡尺寸分布对羽流运动参数时均值的影响。试验结果给出了不同尺寸气泡的分布和运动差异，气泡羽流中气泡的横向扩散不仅跟羽流的流动结构有关，还会受到气泡横向振荡和剪切流场中气泡升力的影响。多重因素作用下大气泡主要聚集在羽流中心区域，中等气泡的横向扩散范围最大，小气泡的横向扩散范围介于两者之间。研究表明不同尺寸分布组成的气泡羽流有不一样的内部结构，气泡尺寸分布相对不均匀的气泡羽流包含更多的大气泡，中心线气泡体积分数更大，中心线连续相流体速度更大，气泡和羽流的特征宽度更小。
Bubble plume is an important flow phenomenon and widely exists in both nature and industry. Experimental observation of the structure of bubble plume is helpful to improve the understanding and prediction of the bubble multiphase flow, and the study is of scientific and practical importance. The development of digital image processing algorithm promotes the experimental observation of bubble plume. The paper constructs a multiphase plume platform, develops an image processing algorithm using predictor-corrector method and carries out a set of experiments. Shadow images of bubbles and PIV images of high resolution are recorded, and the structure of bubble plume is studied based on image processing and statistical analysis.Different bubble plumes are produced by the multiphase plume experiment platform. The continuous phase and dispersed phase are simultaneously measured and the shadow images are identified and continuously tracked. To measure bubbles in highly overlapped shadow images more accurately, an image processing algorithm using predictor-corrector method is proposed. The method is based on the motion continuity of a bubble, and the information of the known frame is used to estimate and correct the adjacent frame. For the artificial bubble images, the proposed algorithm successfully captures all bubbles including inner bubbles which cannot be detected in a single frame. For the experiment of bubble plume, the effectiveness of the proposed method also reaches 95% by random sampling. The method can capture inner bubbles which cannot be detected in a single frame, which reduces the missing rate of bubble detection and improves the accuracy of bubble size and position. It is of great significance to study the mesoscopic movement of bubbles of different sizes. And it is of great significance to measure the distribution of number density, bubble size, void fraction and other statistical parameters. The multiphase experiment plume platform and the image processing algorithm are applied to study bubble plumes in unstratified stagnant water. A set of bubble plumes with different bubble size distribution are designed, the distribution and movement of bubbles of different sizes are studied, and the impact of bubble size distribution on mean plume characteristics is studied. The experimental results show the difference in the distribution and movement of bubbles of different sizes. The transverse diffusion of bubbles in a plume is not only related to the flow structure of the plume, but also affected by the oscillation of bubbles and the lift force in shear flow field. Under the action of multiple factors, large bubbles mainly concentrate in the central area of plume, the transverse diffusion range of medium bubbles is the largest, and that of small bubbles is next. The results show that bubble plumes with different size distribution have different structures, for the bubble plumes with more large bubbles, the centerline void fraction and centerline liquid velocity are larger, and the characteristic widths of bubble core and plume are smaller.
