Scholarly article on topic 'Developing a High Speed Fiber-optic Endoscopic Technique for Measuring Particle Phase Characteristics in a Spouted Bed'

Developing a High Speed Fiber-optic Endoscopic Technique for Measuring Particle Phase Characteristics in a Spouted Bed Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Procedia Engineering
OECD Field of science
Keywords
{"Fiber-optic endoscope" / "High speed photography" / "Particle phase velocity" / "Statistical analysis"}

Abstract of research paper on Materials engineering, author of scientific article — Long Qian, Yong Lu, Wenqi Zhong, Baosheng Jin

Abstract A novel endoscopic photography technique coupling a high speed CCD camera and an intrusive fiber-optic endoscope has been extended to enable the investigation of particle flow behavior in a 3D spouted bed 1500mm in height and 200mm in diameter. A series of algorithms have been specifically optimized to realize the process of image processing and the calculation of the instantaneous particle velocity data. Then particle phase velocity and granular temperature were obtained respectively by statistical analysis and detailed study of the obtained data. The results showed that instantaneous particle velocities exhibited random complexity in both magnitude and direction. Although little information could be extracted from the raw data directly, the profile of particle phase velocity was able to show the mainstream particle flow patters along bed radius. Also, the computed granular temperature could be used to characterize the local particle fluctuation intensity. All experimental results could be useful to validate the CFD models for gas-solid flow systems.

Academic research paper on topic "Developing a High Speed Fiber-optic Endoscopic Technique for Measuring Particle Phase Characteristics in a Spouted Bed"

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Procedía Engineering 102 (2015) 150-158

Procedía Engineering

www.elsevier.com/locate/procedia

The 7th World Congress on Particle Technology (WCPT7)

Developing a high speed fiber-optic endoscopic technique for Measuring particle phase characteristics in a spouted bed

Long Qian, Yong Lu*, Wenqi Zhong, Baosheng Jin

Key Laboratory of Energy Thermal Conversion and Control of Ministry of Education, School of Energy and Environment, Southeast University,

Nanjing 210096, China

Abstract

A novel endoscopic photography technique coupling a high speed CCD camera and an intrusive fiber-optic endoscope has been extended to enable the investigation of particle flow behavior in a 3D spouted bed 1500 mm in height and 200 mm in diameter. A series of algorithms have been specifically optimized to realize the process of image processing and the calculation of the instantaneous particle velocity data. Then particle phase velocity and granular temperature were obtained respectively by statistical analysis and detailed study of the obtained data. The results showed that instantaneous particle velocities exhibited random complexity in both magnitude and direction. Although little information could be extracted from the raw data directly, the profile of particle phase velocity was able to show the mainstream particle flow patters along bed radius. Also, the computed granular temperature could be used to characterize the local particle fluctuation intensity. All experimental results could be useful to validate the CFD models for gas-solid flow systems.

© 2015The Authors.PublishedbyElsevierLtd.Thisis an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Selectionand peer-reviewunderresponsibility of Chinese Society of Particuology, Institute of Process Engineering, Chinese Academy of Sciences (CAS)

Keywords: Fiber-optic endoscope; High speed photography; Particle phase velocity; Statistical analysis

1. Introduction

The spouted bed originated as a simple low-tech method for drying moist wheat particles, but has by now achieved considerable development and several new industrial applications, such as coal combustion, coating tablets, drying tobacco and materials blending etc[1,2]. However, the real-time behavior of particle flow field in a spouted

* Corresponding author. Tel.: +86-25-83790655; fax: +86-25-83790721. E-mail address: luyong@seu.edu.cn

1877-7058 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Selection and peer-review under responsibility of Chinese Society of Particuology, Institute of Process Engineering, Chinese Academy of Sciences (CAS) doi:10.1016/j.proeng.2015.01.118

bed is still not well understood due to the difficulty in obtaining the flow parameters inside vessels of large sizes. So far, numerous techniques such as impact method, heat balance method, optical fiber probes[3,4] as well as the advanced instruments like Laser Doppler Anemometry (LDA)[5], Particle Image Velocimetry (PIV)[6,7], electrostatic charge measurement^] and various kinds of tomography methods[9,10,11] have been developed to investigate particle behavior in gas-solid flow. A systematical summary of these techniques is given by Werther (1999)[12]. Although these techniques have achieved many successful experimental results in 2D fluidized and spouted beds or dilute phase gas-solid flow, the measurement of local particle velocity inside a 3D spouted bed remains a hot spot under research.

In the current work, a novel fiber high speed photography measurement system coupling a fiber-optic endoscope with a high speed CCD camera has been applied to investigate the particle phase velocity profile inside a 3D cone-base cylindrical spouted bed. Besides, the measurement uncertainty in the process of calculating particle velocity is evaluated using the root sum-of-squares method and the granular temperature characterizing the mean kinetic fluctuation energy of local particles is also obtained through detailed measurements of the instantaneous particle velocities.

2. Experiment

2.1. Experimental set-up

The experiment was carried out in a 3D spouted bed, as is illustrated schematically in Fig. 1. The spouted bed was a cylindrical column made of 5 mm thick Plexiglas with a conical base. The cylindrical section was 1500 mm in height and 200 mm in diameter. The conical base was 156 mm in height and had an angle of 60°. Two lines of measurement ports were symmetrically arranged along the column wall, beginning near the bottom of the cylindrical section. Closely filtered millet of mean diameter 1.93 mm, density 1330 kg/m3 and voidage 0.4 was used as the bed material. Air at ambient temperature, as spouting gas, was provided by a compressor and measured by rotameters. The spouting gas flow rate could be manually controlled by two valves. The initial bed height was 240 mm and the minimum spouting velocity, Ums, for the bed height was estimated by means of a pressure drop measurement to be 0.324 m/s.

Fig. 1. Sketch of experimental apparatus.

In order to investigate the particle motion behavior inside the 3D spouted bed, a high speed CCD camera with high sensitivity to light was employed in combination with an intrusive fiber-optic endoscope. The fiber-optic

endoscope of diameter 8 mm and 3 mm-® visual range was inserted into the bed column through measurement ports and traversed radially to observe the local flow structure. The high speed CCD camera in the experiments was BASLER A504K, which was capable of acquiring 500 frames per second (fps) at the full resolution of 1280x1024 pixel and up to 1200 fps at lower resolutions. A 150W halogen lamp was used to provide illumination for the measurement space in front of the endoscope lens through a set of guiding fibers. Then particles appearing in the measurement space could be seen via the endoscope and recorded by the camera. A powerful computer equipped with the image grabbing card of Meteorll-CameraLink was employed to monitor the process of video capture and to store the data as AVI format for further off-line processing.

A set of measurements have been carried out at Port #1, which was 280 mm above the bottom line of the conical base, when superficial gas velocity was kept as 0.43 m/s (U=1.3Ums). The measurements were taken at six different radial positions: the tip of the endoscope was located 20, 30, 40, 50, 65, 75mm away from the bed axis (r/R=0.2, 0.3, 0.4, 0.5, 0.65, 0.75). During the experiment, the resolution of the video camera was set as 512x480 pixel and the recording rate was 1000 fps in order to prevent any blurring. The sampling time was 2 s for each measurement and the halogen lamp was set at the maximum brightness for detailed viewing.

2.2. Processing method

As for image processing, custom software based on MATLAB has been programmed to improve the image quality, automatically search for particles and to extract particle properties such as particle area, diameters, centroid coordinates and circularities, which enables the calculation of instantaneous particle velocity.

Image processing mainly includes the following steps: (1) calibration of the endoscopic distortion; (2) image denoising; (3) contrast enhancement; (4) binarization; (5) particle recognition and labeling; (6) properties extraction. Since step(1), (2) and (3) are all aimed at improving the image quality, they are collectively referred to as image restoration. The detailed information about image processing is described in previous work by Zhang et al. (2012)[13] and Qian et al.(2013)[14] Fig. 2 illustrates the image processing procedure for a typical particle frame.

Restoration Binarination Recognition & Labeling

Fig. 2. Image processing procedure.

As the measurement space is located in the fountain of the spouted bed, which is classified as a dilute phase flow structure, an improved Particle Tracking Velocimetry (PTV) method based on the similarity of particle area and circularity is adopted to realize the tracking of matching particles in consecutive frames. Here it is assumed that the sampling rate of the CCD camera is sufficiently high so that particles move in straight lines and are not accelerated. Then the centroid coordinates of the matching particles in different frames are extracted and each instantaneous particle velocity is computed over at least three sequential frames to improve accuracy. Fig. 3 shows the tracking procedure of four typical matching particles in the video recorded at z=280 mm, r/R=0.65, U=1.3Ums.

Horizontal Position /pixel 50 100 150 200 250

Fig. 3. Particle tracking procedure.

2.3. Measurement uncertainty

The measurement uncertainty for the high speed fiber-optic endoscopic technique is evaluated by combining the measurement uncertainties of the primary measurements aiming to acquire the independent parameters which are used to calculate the instantaneous particle velocity[15]. The root sum-of-squares method (RSS) is applied here. An instantaneous particle velocity Up is calculated by measuring the displacement of a particle between consecutive sampling frames:

Up = = ^ (1)

p n-AT n-AT

Where S2 and S1 are the particle centroid coordinates as the particle moves through a series of n frames, AT is the sampling interval[16,17].

Therefore, the RSS equation in terms of the independent parameters in this study, namely the particle displacement and sampling interval, is

uu = (-)2 + (-UpuAT)2 =. uiS)2 + (--uTT)2 (2)

Up VdAS iS SAT V n-AT iS n-AT2 iT

Where uAS and uAT represent the measurement uncertainties in particle displacement and sampling interval respectively.

The measurement uncertainty in AS happens throughout the whole process of the image acquiring and processing. To be specific, it arises from the pixel resolution of the camera sensor, the uncertainty of the method employed to calibrate the lens distortion, and the particle recognition algorithm used to determine the centroid coordinates of a particle[18]. Thus, the measurement uncertainty in AS can be evaluated according to the following equation:

- V ure.soluton ucalibration urecognition (3)

Where uresolution, ucalibra,ion and urecognUion stand for the measurement uncertainties caused by the pixel resolution, calibration method and recognition algorithm respectively.

When a CCD camera is used to capture particle images, the error in the sampling of particle displacement is 42 times the particle centroid error, which is dependent on the particle image radius, pixel area and the Gaussian peak location on the CCD sensor, etc[19]. In this study, the particle centroid error of the CCD camera for a typical PTV object is estimated to be about ±1/3 pixel, which leads to a sampling error in particle displacement of ± >/2 /3 pixel. Since particle velocities are computed over at least three consecutive frames (n ^2) and the corresponding particle displacements during that time are at least 20 pixels, the measurement uncertainty due to pixel resolution of the CCD camera (uresolution) can be calculated as: uresolution = (yf2/ 3) / 20 x 100% = ±2.36%.

Another measurement uncertainty in AS arises from the distortion calibration of the endoscope lens. Although the barrel distortion is significantly reduced after the calibration, the residual distortion should be investigated in terms of its influence on the uncertainty. In the evaluation process, an image of the matrix sample of identical dots with the same interval is captured and calibrated. Then the centroid coordinates of dots both in the original and calibrated images are extracted and the L2-norm of these coordinates is calculated to estimate the mean error resulted from the calibration. The estimation results are listed in Tab. 1.

Tab 1. Error analysis of calibration.

L2-norm Number of dots Mean error /pixel 37.2 56 0.66

As the typical particle displacement is at least 20 pixels, the measurement uncertainty due to calibration (ucalibration) can be evaluated as:

uciMmti0n=0.66/20x 100%=±3.3%.

As for the measurement uncertainty caused by particle recognition algorithm, it happens when an image is binarized with certain threshold. During the process, the inaccuracy may reach ±0.5 pixel. Considering the particle displacement, the measurement uncertainty due to recognition algorithm (urecognition) can be estimated as: urecognition=0.5/20^100%=±2.5%. According to Eq. (3), the measurement uncertainty in AS is calculated as:

uS =Ju 2 + u b o2 + u " = V2.362 + 3.32 + 2.52% = ±4.765%

AS \ resolution calibration recognition *

The measurement uncertainty in AT is produced when the internal clock of the camera is not well calibrated. In this study, the error in camera parameters is quite low, on the order of nanosecond. Therefore, it is reasonable to neglect the measurement uncertainty in AT, which gives uAT the value of 0.

In accordance with Eq. (2), the total measurement uncertainty in particle velocities (uUp) is given by:

(—— ^)2 + (--^r)2 = J(— x 4.765)2 + 0% = ±2.383%

n -Al n -Al V 2 >

It is shown that the measurement uncertainty in instantaneous particle velocities is ±2.383% and the major uncertainty is introduced in the process of endoscopic distortion calibration.

3. Results and discussion

In order to gain a deeper insight into the complex internal flow phenomena in the spouted bed, two parameters from different perspectives (i.e. particle phase velocity and granular temperature) are defined and analyzed through detailed measurement of instantaneous particle velocities. To be specific, particle phase velocity focusing on the mainstream particle flow behavior is obtained through statistical analysis while granular temperature representing the local kinetic fluctuation energy is acquired through the method proposed by Tartan et al (2004)[20].

3.1. Particle phase velocity

In the case of z=280 mm, r/R=0.65, U=1.3Ums, the instantaneous particle velocities were obtained using the high speed fiber-optic endoscopic technique and the processing method described above. These instantaneous particle velocities were then decomposed along the axial and radial directions respectively. Fig. 4 shows the scatter diagram of axial component velocities and the corresponding probability distribution.

0 200 400 600 800 1000 -1.5 -1.0 -0.5 0.0 0.5 1.0

Particle Serial Number Axial Component Velocity /m s-1

Fig. 4. Instantaneous axial component velocity: (a) scatter diagram; (b) probability distribution.

According to Fig. 4(a), particles within the measurement space exhibit complex flow behavior along the axial direction: there are a large number of particles moving upwards and downwards simultaneously and little relation is found between time and the variation in velocity magnitude or direction. Fig. 4(b) indicates that the axial component velocities fall into a narrow range around a certain value and this value may be adopted to characterize the flow behavior of the majority of particles in the present measurement. Here, it can be determined statistically that more than 80% particles move within -0.43~0.22 m/s. Therefore, the mean value of these velocities is defined as the particle phase axial velocity, which is calculated to be -0.10 m/s in this case. The minus denotes that the majority of particles flow downwards axially.

' Boundary _

' \ Fountain Core Periphery .

0.4 0.5 0.6

Radial Position r/R

Fig. 5. Profile of particle phase axial velocity

The radial profile of particle phase axial velocity in the case of z=280 mm, U=1.3Ums is shown in Fig. 5. It is revealed that particle phase axial velocity reaches its peak around the bed axis and decreases to 0 along the radius in the fountain core. Then it changes direction around r/R=0.45 and gradually increases to a constant value towards the wall. The variation of particle phase axial velocity is consistent with the internal circulating regime of the spouted

bed: particles in the fountain core travel upwards at high velocity with spouting gas and rain back onto the bed surface in the periphery. The radial position where particle phase axial velocity turns 0 denotes the boundary between fountain core and periphery. The results agree with those obtained experimentally by He et al(1994)[3].

3.2. Granular temperature

The similarity of granular flows to thermal fluids has led to the extension of concepts used in the analysis of gassolid flow[21]. An analog to the thermodynamic temperature, granular temperature 0, can be defined in terms of the mean kinetic fluctuation energy of particles in Eq. (4).

6(f,x) = 3{CaCa) +1 (CC) + 3{C'C) (4)

Where 0 is the granular temperature and it varies with time t and position x; {CfC^j is the particle stress per unit bulk density over a given frame (Specifically, when subscripts i and j are the same, {CC^ denotes the particle normal stress per unit bulk density); Subscripts a and r are for axial and radial components of normal stress respectively.

Since the particle flow in the spouted bed is nearly axisymmetric radially, Eq. (4) can be simplified to:

6(t, x) = 3 (CrC) + 3 (CaCa) (5)

As for particle normal stress per unit bulk density, it can be computed from the hydrodynamic velocity vi and the corresponding particle peculiar velocities. The calculation procedure resembles that provided by Jung et al (2005)[22], as is given by Eq. (6).

(CC) (t, x) =—£ c - v)2 (6)

The hydrodynamic velocity vi at the position x is defined as:

vl(t, x) = N Z ck (^ x) (7)

Where represents a and r directions and N is the number of particles in each frame. The particle peculiar velocity is given by the instantaneous particle velocity ci minus the hydrodynamic velocity vi.

The granular temperature in the case of z=280 mm, r/R=0.65, U=1.3Ums was calculated according to Eq. (5). During the calculation procedure, it was found that the particle normal stresses per unit bulk density along the a and r directions both fluctuated with time and their values of magnitude were of the same order, which suggested significant particle fluctuations along both directions. It was also indicated that the fluctuation in particle normal stress along the axial direction was slightly larger than that along the radial direction, possibly resulting from the fiercer momentum transport between particles and gas phase axially.

A typical time series of granular temperature in the case of z=280 mm, r/R=0.65, U=1.3Ums is presented in Fig. 6. According to the radial profile of particle phase axial velocity, the majority of particles flow downwards axially around r/R=0.65, leading to an increase in the negative particle phase velocity. In fact, the falling particles in this area also cause the particle random fluctuation to intensify both axially and radially, which results in an increase in granular temperature. It is shown in Fig. 6 that the variation in granular temperature with time directly relates to the changes in random particle fluctuation within the measurement space over the sampling time. The time-averaged granular temperature over 500 sequential frames is calculated to be 0.0162m2/s2, which is higher than that in the annulus but lower than that in the spout. It is mainly because that particle concentration in fountain is much lower than that in annulus while the particle velocity is not as high as that in spout. This result agrees with that reported by Lu et al (2004)[23] using numerical simulations. In addition, it is found that the particle fluctuations are significant

along both directions, indicating obvious momentum transport between particles and gas phase radially around r/R=0.65 in fountain.

Fig. 6. Granular temperature time series.

4. Conclusion

In this work , a high speed fiber-optic endoscopic technique has been developed for measurements of instantaneous particle velocities inside a 3D spouted bed. The particle phase velocity and granular temperature were determined by image processing and statistical analysis. Besides, the measurement uncertainty for this measurement technique was evaluated by the RSS method. Some notable findings are presented as follows:

(1) The measurement technique described is capable of acquiring the particle phase velocity and granular temperature in the spouted bed, which indicates that the technique is a promising tool in the gas-solid flow research.

(2) The measurement uncertainty in instantaneous particle velocities is evaluated to be ±2.383% and the major uncertainty is introduced in the process of endoscopic distortion calibration.

(3) The profile of particle phase velocity shows that particles exhibit different flow patters along the bed radius. The random particle fluctuation energy can be characterized by the parameter of granular temperature calculated in the experiment.

Acknowledgements

Financial support from the National Natural Sciences Foundation of China (No. 51376048) is acknowledged with gratitude.

Nomenclature

(CC^ particle stress per unit bulk density, m2/s2

r/R radial coordinate

AS particle displacement, mm

AT sampling interval, ms

u measurement uncertainty, %

U superficial gas velocity, m/s

Ums minimum spouting velocity, m/s

instantaneous particle velocity, m/s i hydrodynamic velocity, m/s

z vertical distance from bottom line of the base, mm

0 granular temperature, m2/s2

References

[1] K. B. Mathur, N. Epstein, Spouted Beds, Academic Press, New York,1974.

[2] N. Epstein, J. R. Grace, Spouted and Spout-Fluid Beds: Fundamentals and Applications, Cambridge University Press, Cambridge,2010.

[3] Y. L. He, S. Z. Qin, C. J. Lim, J. R. Grace, Particle Velocity Profiles and Solid Flow Patterns in Spouted Beds, Can. J. Chem. Eng. 72 (1994) 561-568.

[4] M. Olazar, M. J. San José, S. Alvarez, A. Morales, J. Bilbao, Measurement of Particle Velocities in Conical Spouted Beds Using an Optical Fiber Probe, Ind. Eng. Chem. Res. 37 (1998) 4520-4527.

[5] M. I. Boulos, B. Waldie, High Resolution Measurement of Particle Velocities in a Spouted Bed Using Laser - Doppler Anemometry, Can. J. Chem. Eng. 64 (1986) 939-943.

[6] J. A. Laverman, I. Roghair, M. V. S. Annaland, H. Kuipers, Investigation Into the Hydrodynamics of Gas - Solid Fluidized Beds Using Particle Image Velocimetry Coupled with Digital Image Analysis, Can. J. Chem. Eng. 86 (2008) 523-535.

[7] H. Shi, Experimental Research of Flow Structure in a Gas-Solid Circulating Fluidized Bed Riser by PIV, J. Hydrodynamics. 19 (2007) 712719.

[8] A. Sowinski, F. Salama, P. Mehrani, New Technique for Electrostatic Charge Measurement in Gas - Solid Fluidized Beds, J. Electrostat. 67 (2009) 568-573.

[9] S. M. Rao, K. Zhu, C. Wang, S. Sundaresan, Electrical Capacitance Tomography Measurements On the Pneumatic Conveying of Solids, Ind. Eng. Chem. Res. 40 (2001) 4216-4226.

[10] S. Malcus, G. Chaplin, T. Pugsley, The Hydrodynamics of the High-Density Bottom Zone in a CFB Riser Analyzed by Means of Electrical Capacitance Tomography (ECT), Chem. Eng. Sci. 55 (2000) 4129-4138.

[11] S. Wang, T. Dyakowski, C. G. Xie, R. A. Williams, M. S. Beck, Tomographic Imaging of Phase Distribution and Spatial Correlation of the Fluidization Dynamics in a Fluidized Bed, Part. Part. Syst. Char. 13 (1996) 333-338.

[12] J. Werther, Measurement Techniques in Fluidized Beds, Powder Technol. 102 (1999) 15-36.

[13] J. Zhang, Y. Lu, W. Zhong, X. Chen, Y. Chen, B. Ren, B. Jin, Measurement method of particle phase motion parameter based on fiber high speed photography. J. Southeast Univ. 2012, 42(5): 910-914.

[14] L. Qian, Y. Lu, W. Zhong, X. Chen, B. Ren, B. Jin, Developing a Novel Fibre High Speed Photography Method for Investigating Solid Volume Fraction in a 3D Spouted Bed, Can. J. Chem. Eng. 91 (2013) 1793-1799.

[15] Y. Fei, Error Theory and Data Processing, fifth ed., Machinery Industry Press, Beijing, 2000.

[16] B. Gopalan, F. Shaffer, A New Method for Decomposition of High Speed Particle Image Velocimetry Data, Powder Technol. 220 (2012) 164-171.

[17] F. Shaffer, B. Gopalan, R. W. Breault, R. Cocco, S. B. Karri, R. Hays, T. Knowlton, High Speed Imaging of Particle Flow Fields in CFB Risers, Powder Technol. 242 (2013) 86-99.

[18] K. Miyazaki, G. Chen, F. Yamamoto, J. Ohta, Y. Murai, K. Horii, PIV Measurement of Particle Motion in Spiral Gas - Solid Two-Phase Flow, Exp. Therm. Fluid Sci. 19 (1999) 194-203.

[ 19] P. Veber, J. Dahl, R. Hermansson, Study of the Phenomena Affecting the Accuracy of a Video-Based Particle Tracking Velocimetry Technique, Exp. Fluids. 22 (1997) 482-488.

[20] M. Tartan, D. Gidaspow, Measurement of Granular Temperature and Stresses in Risers, AIChE J. 50 (2004) 1760-1775.

[21] R. D. Wildman, J. M. Huntley, D. J. Parker, Granular Temperature Profiles in Three-Dimensional Vibrofuidized Granular Beds, Phys. Rev. E. 63 (2001) 61311.

[22] J. Jung, D. Gidaspow, I. K. Gamwo, Measurement of Two Kinds of Granular Temperatures, Stresses, and Dispersion in Bubbling Beds, Ind. Eng. Chem. Res. 44 (2005) 1329-1341.

[23] H. Lu, Y. He, W. Liu, J. Ding, D. Gidaspow, J. Bouillard, Computer Simulations of Gas-solid Flow in Spouted Beds Using Kinetic-frictional Stress Model of Granular Flow, Chem. Ing. Sci. 59 (2004) 865-878.