publications

2021

1. Sci Rep

   Deep learning-based automated and universal bubble detection and mask extraction in complex two-phase flows

   Kim, Yewon,  and Park, Hyungmin

   Scientific reports, 2021

   Abs HTML Code

   While investigating multiphase flows experimentally, the spatiotemporal variation in the interfacial shape between different phases must be measured to analyze the transport phenomena. For this, numerous image processing techniques have been proposed, showing good performance. However, they require trial-and-error optimization of thresholding parameters, which are not universal for all experimental conditions; thus, their accuracy is highly dependent on human experience, and the overall processing cost is high. Motivated by the remarkable improvements in deep learning-based image processing, we trained the Mask R-CNN to develop an automated bubble detection and mask extraction tool that works universally in gas–liquid two-phase flows. The training dataset was rigorously optimized to improve the model performance and delay overfitting with a finite amount of data. The range of detectable bubble size (particularly smaller bubbles) could be extended using a customized weighted loss function. Validation with different bubbly flows yields promising results, with AP50 reaching 98%. Even while testing with bubble-swarm flows not included in the training set, the model detects more than 95% of the bubbles, which is equivalent or superior to conventional image processing methods. The pure processing speed for mask extraction is more than twice as fast as conventional approaches, even without counting the time required for tedious threshold parameter tuning. The present bubble detection and mask extraction tool is available online (https://github.com/ywflow/BubMask).

2019

1. Int J Multiph Flow

   Upward bubbly flows in a square pipe with a sudden expansion: Bubble dispersion and reattachment length

   Kim, Yewon,  and Park, Hyungmin

   International Journal of Multiphase Flow, 2019

   Abs HTML

   We investigate the bubble dynamics and subsequent changes in the liquid-phase flow characteristics of upward bubbly flows in a square pipe with a sudden expansion (expansion ratio of 4.0). The experiments are conducted under three liquid-phase Reynolds numbers of 0 (stationary), 420 (laminar) and 6000 (turbulent). The inlet volume void fraction ranges about 0 - 1.0 % and we use a high-speed two-phase particle image velocimetry and shadowgraphy to measure two phases simultaneously. It is observed that after the expansion, smaller bubbles tend to migrate toward the wall and larger bubbles rise in a core region, which is more encouraged by the steeper velocity gradient with increasing the Reynolds number. This bubble distribution is further analyzed by estimating the interfacial forces acting on rising bubbles. Affected by this bubble distribution, the enhanced turbulence in the inlet flow energizes the separating shear layer, resulting in the reduction of reattachment length behind the edge. Finally, we suggest an empirical relation for a two-phase flow reattachment length in terms of Reynolds number and mean void fraction, which explains the contribution of added dispersed phase on the mixing enhancement in the backstep flows.