Study On In-line Image Method For Measuring Particle Characteristics Distribution In A Multiphase Reactor

Multiphase reactors are widely used in process industries.The heterogeneous characteristics(the inhomogeneity of the distribution of multiphase fluids in time and space)such as particle size distribution and phase holdup in multiphase reactors is an important information for reactor optimization and scaling-up design.It can also provide basic data for the numerical simulation verification and the model key parameters determination of multiphase reactors.The dispersed phase holdup of industrial multiphase reactors is generally higher than negligible.In theory,the intrusive in-line imaging technique can measure the system with a high concentration of the dispersed phase due to the feature of a short optical path,but in practice,the maximum phase holdup is also closely influenced by the particle size.In addition,the phenomenon of particle image overlapping in high phase holdup systems is very common,so the automatic and accurate selection of target particles is very important for the accuracy of image-based measurement.In this thesis,the in-situ particle images of solid particles(subsphaeroidal alumina and silver-plated glass beads,irregular quartz sand)were acquired in a liquid-solid stirred tank by using the telecentric photographic probe developed by our group.Firstly,the selection strategy for particle images of different characteristic was developed based on quartz sand and alumina particle images,and the particle selection method was optimized by comparing with the manual marking.The conclusions are as follows:(1)For regular particles,the circle detection on original particle images causes a large error.The target particles can be easily detected after edge enhanced using the Canny operator,but the edges of non-target particles will be also possibly enhanced,leading to more particles being misidentified.The error of circle detection after the threshold segmentation is the smallest,and the optimal threshold is about 60%of the threshold calculated by the Maximum Between-Class Variance Method(Otsu).(2)For irregular particles,the ellipse fitting method suffered to a large error.Contour detection based on the threshold segmentation will lead to a larger Sauter diameter.The relative errors of the detection results by using the watershed segmentation are small on the phase holdup and the Sauter diameter measurements.The optimal threshold is also about 60%of the threshold calculated by OtsuBy utilizing the optimal image processing strategy,the measurement results of alumina were compared with vernier calipers,and the measured particle sizes of quartz sand and alumina were compared with the measurements by a laser particle size analyzer.The change of depth of field(DOF)of silver-plated glass bead images was also verified and calibrated by measurement at high concentration.All the particles used in the experiment were obtained through sieving.The analysis suggested conclusions are as follows:(1)The measurement result of regular particle size by the present imaging method is close to that measured by the vernier caliper.(2)The results of particle size distribution obtained by the telecentric photographic probe and the laser particle size analyzer are quite different.The measurement results of irregular particle size by the telecentric photographic probe are more concentrated in the sieve partition.(3)The calculated depth of field(DOFC)decreases linearly with the increase of phase holdup in the measurement of high phase holdup of silver-plated glass bead.