Identification of flow regime in a bubble column reactor with a combination of optical probe data and machine learning technique

Prediction of Bubble Sizes in Bubble Columns with Machine Learning Methods

“…With Δ z depending on the frame rate f and bubble velocity u B while ε i,j,k and Δ x , Δ y being the phase fraction and radial distances of the WMS, respectively. In their work, Theßeling and Grünewald 3 calculated u B with a correlation provided by Fan et al 13. The bubble diameter was approximated by …”

“…An overview of ML applications in process engineering divided according to algorithm types has been provided by Ge et al [11], whereas Yan et al present applications of ML methods in multiphase flow measurement [12]. Regarding the investigation and prediction of multiphase flow characteristics, ML algorithms have recently shown their capability and efficiency, i.e., in the identification of flow regimes in bubble column reactors [13] and horizontal pipes [14]. In this study, further developed and optimized ML algorithms based on the work of Theßeling et al [8] are presented, showing clear advantages in terms of prediction accuracy regarding bubble size determination based on multiphase flow measurement with WMS data compared to previous approaches.…”

Passive acoustic identification of bubble flow regime based on synchrosqueezing wavelet transform and deep learning

“…The transition flow is a regime observed between homogeneous and heterogeneous flow where the bubbles begin to coalesce and break up, not well defined as the other two 31 . Actually, there is one more flow regime called slug flow, where the bubbles exist in size nearly equivalent to the length scale of the container, 11 but it will not be discussed for not being observed in the experiment.…”

“…To make up for the shortcomings of subjective methods, lots of objective methods have been developed to better identify bubble flow regimes so as to meet industrial needs. The objective methods generally have four steps: signal acquisition, feature extraction, classifier definition, and regime identification, and can be of many kinds depending on the signal types used, such as differential pressure methods, [5][6][7] electrical properties methods (conductivity, resistivity, capacity), [8][9][10] optical properties methods, 11 acoustic properties methods, 1,4,12 high speed imaging methods, 13,14 and radiation attenuation methods. [15][16][17] Among these methods, acoustic methods are gaining more and more attention lately.…”

“…Experimental setup [Color figure can be viewed at wileyonlinelibrary.com] T A B L E 1 Parameters of the pipes Outer diameter (mm) Inner diameter (mm) No. of pipes Array shape (line/circle) Spacing (mm) I G U R E 2 (a) Layout of the five pipes; (b) Layout of the hydrophones; (c) Commonly observed bubble flow regimes in bubble column33 [Color figure can be viewed at wileyonlinelibrary.com] regime called slug flow, where the bubbles exist in size nearly equivalent to the length scale of the container,11 but it will not be discussed for not being observed in the experiment.As can be seen from Figure2cand known from existing researches, the bubble flow regime generally changes as a result of gas flow rate change. So, in the experiment, we obtained all three bubble flow regimes by changing air flow rate while keeping the air pressure and temperature constant.…”