Validation Study Of Cross-Correlation Based Solids Velocimetries In Gas-Solids Fluidization Systems

Fluidized-beds are widely explored in energy production,chemical synthesis,pharmacy and other industrial processes.In these applications,accurate measurement of hydrodynamic and performance parameters is critical to the design,modeling,scaling up and operations of these gas-solids fluidized bed reactors.One of the most prevalently measured parameters is particle velocity.To date,particle velocity measuring instruments based on cross-correlation algorithm have been widely applied in various multiphase flow systems.However,it is found that the particle velocity measurement quality differs significantly under different fluidization regimes in terms of data repeatability and cross-correlation coefficient.Researchers and industry experts have observed that cross-correlation coefficient obtained by using the solids velocimeter based on the cross-correlation algorithm,is much lower in the low-velocity dense bed than in high-velocity fast-fluidization riser.Consequently,the algorithm produces different particle velocity measurement qualities under different fluidization regimes,which make the measurement results questionable.Although such an instrument is widely used,there are few reports on the quantitative evaluation of its measurement quality.The purpose of this study is to systematically evaluate the measurement quality of the cross-correlation based solids velocimeter in different fluidization systems.Thus,the underlying reasons and mechanisms of these discrepancies are worth exploring.New methods to improve the measurement quality of such instruments are thus essential.The measurement quality of particle velocity was systematically investigated by Labasys?100 fiber optic velocimetry probe under different fluidization regimes.It was found that,compared to fast fluidized bed(FFB),the measured particle velocities in bubbling fluidized bed(BFB)showed lower data repeatability,higher percentage of invalid data and higher sensitivity to the selection of the threshold value of maximum cross-correlation coefficient,which corresponded to much lower measurement qualities in this regime.Comparatively,higher measurement qualities were observed in the FFB based on the same evaluating methods.The highest measurement qualities appear in the core region of the riser and under high superficial gas velocities.A matchup relationship between the measurement quality and the shape of the probability density function(PDF)curves of maximum cross-correlation coefficient was found.Narrower PDF curves corresponded to higher measurement quality and vice versa.In view of the indirect evaluation of measurement quality of particle velocity based on cross-correlation algorithm,the real system lacks an absolutely accurate measurement value.As such,the reliability and scope of application of this method has not been quantified.Therefore,a CFD-based “virtual calibration” method,namely,“virtual error quantification”,was proposed to quantify the reliability of crosscorrelation algorithm in the measurement of solids velocity in two different gas fluidization regimes,i.e.,BFB and FFB.It was found that the cross-correlation algorithm works well in the core region of FFB,but its applications in the annulus region of FFB and in BFB should be considered with great caution.Finally,the findings were delineated by solids convection-mixing mechanism based on the relative standard deviation(RSD)of solids velocity signals and solids Péclet number.In order to further verify the correlation measurement qualities of cross-correlation based solids velocimetry with solids convection-mixing competing mechanism in different gas fluidization regimes.A method was proposed to analyze the local solids residence time distribution(RTD)from the results of Eulerian-Lagrangian simulation,which enabled the calculation of Péclet numbers at local positions.Eulerian-Lagrangian simulations of five different fluidization regimes,i.e.bubbling fluidization,turbulent fluidization,fast fluidization,pneumatic transport and downer,were then performed to analyze the local solids RTD and obtain the local Péclet numbers in those systems.Systematic simulations were then conducted to acquire the local solids RTDs in various gas-solids fluidization regimes.Afterward,the local Péclet numbers were quantified.It was found that the measurement qualities are intimately related to the local Péclet numbers and the solids convection-mixing competing mechanism.Present study also revealed the underlying mechanism of the success and failure of cross-correlation based solids velocimetry,and proved that the cross-correlation algorithm is ideal for solids convection-dominated systems or regions in a system,but caution is needed when solids mixing is important.Although,the root cause of the difference in particle velocity measurement quality in different fluidization systems is given,the problem of lower measurement quality still needs to be resolved.Further,to improve the measurement quality of particle velocity,an improved particle velocity measurement method based on the data-driven algorithm is proposed.Firstly,four commonly used algorithms,namely linear model,random forest,support vector machine and neural network,were used to train the model.By performing training on a group of particle concentration and velocity,a large training error of single data-driven model can be found,where the determination coefficient is below 0.2,indicating a poor generalization ability of the training model.Thus,the hybrid model coupled with artificial neural network and random forest is further proposed.The model results show better data prediction ability and generalization performance.The relative error of the prediction results is reduced from dozens of original method to less than 0.01,which solves the significant deviation of particle velocity calculated by the cross-correlation algorithm based on particle concentration.In conclusion,this study began with the discovery of the difference in the measurement quality of cross-correlation based particle velocity in the experiment,then extended to the quantitative evaluation of the difference in the measurement quality of solids velocimeter based on the cross-correlation algorithm in two fluidization systems.The “virtual calibration” methods not only verified the experimental findings,but also found the root cause of this difference.Finally,an improved method of data-driven based particle velocity measurement was proposed.