Abstract:

Gas-liquid two-phase flow is prevalent in various industries. The measurement and statistical analysis of bubbles characteristics are useful for the investigation of the bubbles motion and generation. Therefore, it can provide the fundamental data for physical mechanism and process control research. In this paper, bubble images were captured by a high-speed camera system at different liquid flow rates and aeration volumes in the circulating water tank. Given that bubble coalescence and breakup during motion could lead to overlapping and sticking issues in the images, these phenomena would cause significant measurement errors in bubble characterization parameters. To address these issues, this paper improved the marker extraction method of the watershed algorithm. The method obtained foreground markers by utilizing the extreme points of the distance transformed image for suppression and fusion, thereby improving the accuracy of the bubbles segmentation. Least squares ellipse fitting was conducted on the segmented bubbles to reconstruct the bubbles profile and obtain their parameters. Experimental results showed that the proposed method have obvious improvement on segmenting the adherent bubbles and extracting the bubble features. Compared to the methods of erosion operation and threshold segmentation for extracting markers, the proposed method increased the accuracy by 22.7% and 13.6%, respectively. Moreover, statistical analysis of the obtained data showed that the number of bubbles increases slightly with increased ventilation volume, while the average bubble size increased significantly. It was found that the ventilation volume mainly affects the size of the bubbles. On the other hand, the bubbles motion was affected by the liquid flow rate. It was noted that the increased liquid flow rate would generate more bubbles with small size due to bubbles breakup.

Key words: gas-liquid flow, overlapping bubble recognition, bubbles characterization statistics, high-speed imaging

摘要：

气液两相流广泛应用于多种工业领域当中，测量气泡特征参数并进行统计分析可以探究气泡的运动及生长规律，这对相关物理规律和过程控制研究具有重要意义。然而气泡在运动中会发生融合、破碎等现象，导致图像中气泡的重叠粘连，以及拍摄背景的不均匀性均会对气泡特征数据提取造成较大误差。本文采用高速摄像技术获得了不同液体流速与通气量下的气泡图像，通过对距离变换图像的极值点进行抑制与融合，改进了分水岭算法的前景标记提取方法，实现了气泡的准确分割，进而对分割气泡进行最小二乘椭圆拟合重构，获取了气泡参数。对比分析发现，相较于腐蚀运算和阈值分割提取标记的方法，本文方法准确率分别提高了22.7%与13.6%。通过对气泡特征进行统计分析，发现随着通气量的增加，气泡数量略微增多而平均尺寸显著增大，表明通气量对气泡尺寸有着重要影响。通气量相同，随着来流速度的增加，气泡破碎，气泡的数量增加且平均尺寸减小。

关键词: 气液两相流, 重叠气泡识别, 气泡特征统计, 高速摄像