3708 Ind. Eng. Chem. Res.

2008, 47, 3708-3719

Electrical Capacitance Tomography - A Perspective

Q. Marashdeh, L.-S. Fan,* B. Du, and W. Warsito
Department of Chemical and Biomolecular Engineering, The Ohio State UniVersity, Columbus, Ohio 43210

This article describes the recent progress in research and development on electrical capacitance tomography
(ECT). Specifically, the article highlights several aspects of ECT including the electrical capacitance volume
tomography (ECVT), 3D sensor design, 3D neural network multicriterion image reconstruction technique
(3D-NN-MOIRT), multimodal imaging based on ECT and ECVT sensors, static-charge effects and the scheme
of their elimination in the ECT image reconstruction, and multiphase flow imaging applications. The multimodal
capability that enables permittivity and conductivity imaging to be simultaneously conducted is illustrated.
The simulation and experimental results are presented to provide quantitative and/or qualitative assessment
of the significance of various ECT techniques. The employment of ECVT in conjunction with using electrical
capacitance based imaging sensors is shown to represent a favorable tool for industrial multiphase flow imaging.

1. Introduction viable noninvasive measurement tool of the dynamics of two-

phase flows based on the variations of the dielectric constants
Multiphase flows and multiphase reactors are commonly in the phase media. ECT was subsequently applied to imaging
encountered in industrial operations for process applications. of flow systems such as oil-gas and oil-water flows in oil
Such applications include riser flows, particle coating and wells,43 gas-solid flows in pneumatic conveyors,44 gas-solid
blending, drying, polymerization, and Fischer-Tropsch syn-
fluidized beds,45 and trickle bed reactors.49 However, expanding
thesis.1,2 Nonintrusive techniques are used for tracking particle
the application of ECT to more complex multiphase flows is
movement, and/or mapping instantaneous or time-averaged,
challenged by its low spatial resolution compared to other
local or cross-sectional averaged, phase hold-ups, and phase
tomography systems and its limitation to nonconductive media
velocities. These techniques include particle image velocimetry,3-5
such as organic liquids.
laser Doppler anemometry,6-9 phase Doppler anemometry,10-11
positron emission tomography12-15, and radioactive particle In recent developments, the accuracy of ECT image recon-
tracking,16-23 ultrasonic tomography,24-26 nuclear magnetic struction has been significantly improved through the employ-
resonance imaging (NMR or MRI),27-32 and electrical imped- ment of optimization reconstruction techniques. Such techniques
ance (resistive or capacitive) tomography.33-38 can be represented by the neural network multicriterion opti-
The inherently complex nature of multiphase flows requires mization image reconstruction technique (NN-MOIRT).35 NN-
a multidimensional measurement technique capable of providing MOIRT is based on optimizing a set of objective functions
real time monitoring of the process dynamics and physical related to both the measured capacitance vector and the
properties. Examples of different physical properties used in reconstructed image. The improvements provided by this
tomography imaging are density, electrical capacitance and reconstruction technique has enabled three-phase imaging to be
resistance, reflection and refraction index, and electric charge.39 conducted on the basis of ECT as well as quasi-3D ECT
The applicability of different imaging modalities to regimes imaging. Another development that has supported the ECT
depends on many factors related to the nature of the regime technique and its applications is 3D direct ECT imaging, noted
and the sensor itself. Among various factors of importance to as electrical capacitance volume tomography (ECVT).36,50,51
the selection of tomography sensors are cost, safety, applicability Unlike traditional 3D imaging, based on stacking 2D images
to different sizes of vessels, and the imaging speed. In this to obtain an interpolated 3D image, the volume tomography
regard, the electrical tomography sensors are attractive when acquires the 3D image directly from the measured capacitance
compared to sensors used in other tomography techniques.40 data. Such imaging was made possible through the 3D ECT
Electrical tomography is based on the measurement of electrical (ECVT) sensor design as well as the additional objective
properties through the utilization of the capacitive, resistive, or functions used in the optimization scheme for 3D reconstruction.
inductive nature of the flow media under investigation. Among The first published work on 3D ECT (ECVT) with the layering
different electrical tomography systems, electrical capacitance concept in sensor design of Warsito and Fan36 shed new light
tomography (ECT) has attracted increased attention in recent on the capability and feasibility of the 3D imaging using ECT.
years as an imaging tool for multiphase flow systems. Subsequent work that contributes to the research and develop-
The development of ECT started in the early 1980s when a ment of 3D ECT imaging was presented by others.52,53 ECVT
group of researchers at the Morgantown Energy Technology has also been used to develop images of high-contrast dielectrics
Center41developed a 2D capacitance tomography system with using cuboid sensors.54 Additional applications of ECVT
its subsequent applications for cross-sectional imaging of technology are yet to be explored. For example, ECVT
fluidized bed systems. Applications of ECT to oil pipelines and technology developed51 was recently recommended for dem-
pneumatic conveyors were explored by researchers at the onstration by the National Aeronautics and Space Administration
University of Manchester Institute of Science and Technology for a space mission for imaging high-contrast dielectrics on
in the early 1990s.42 These earlier works ascertain ECT as a extraterrestrial surfaces.55
The traditional use of the ECT sensor for permittivity imaging
* To whom correspondence should be addressed.Tel.: (614) 688- has been extended to include conductivity imaging as well. In
3262. Fax: (614) 292-3769. E-mail: [email protected]. other recent progress, the ECT sensor has been used for dual
10.1021/ie0713590 CCC: $40.75 © 2008 American Chemical Society
Published on Web 01/31/2008
 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008 3709

Figure 1. Diagram of the ECT system including sensor, data acquisition, and computer for reconstruction.

permittivity-conductivity imaging based on the quasistatic 2.1. Image Reconstruction. Tomography reconstruction
analysis of the sensor for a time varying excitation signal.56 characterizes a process in obtaining a map of physical property
This progress provides a design for a global ECT sensor capable distribution from a set of boundary measurements. Tomography
of 3D volume multimodal imaging. Ongoing activities in the systems can generally be classified into hard field and soft field.
area of ECT research include developments of all aspects of In hard-field tomography, the field lines of the interrogating
the ECT systems. The exploration of innovative applications signal are independent of the physical property distribution in
of ECT systems has also gained considerable momentum in the the imaging domain. An example of the hard-field scenario is
areas including detection of flammable liquids,57 imaging of X-ray tomography. In soft-field tomography, the field distribu-
flames in porous media,58 estimation of average particle size,59 tion is dependent on the physical property distribution, and the
and comparative studies of other imaging modalities.60 reconstruction process takes a higher form of complexity.61 ECT
In this article, a perspective on the development of the ECT is of a soft-field category where the electric field lines are
techniques and their applications is presented. In section 2, the dependent on the permittivity distribution. The electric potential
ECT system and reconstruction techniques are briefly reviewed. and permittivity distributions are related according to the Poisson
Recent advances in the ECVT sensor design are illustrated along equation
with the ECT sensor usage for multimodal imaging. Simulation
results are provided in section 3 to allow the capability of the (x,y,z)∇2φ(x,y,z) + ∇(x,y,z)∇φ(x,y,z) ) 0 (1)
ECVT technique to be assessed. The applications of ECVT are
highlighted in section 4 for the imaging of multiphase flow where (x,y,z) is the permittivity distribution and φ(x,y,z) is the
systems including bubble columns, gas-liquid-solid three- potential distribution. The capacitance of the field is obtained
phase fluidized beds, gas-solid fluidized beds, and risers. by finding the accumulated charge on the capacitance plate per
unit voltage according to
2. Review of ECT Techniques
Similar to any other tomography system, an ECT system is Qi ) I (x,y,z)∇φˆndl (2)
composed of three components: (1) sensor, (2) data acquisition Γi
system, and (3) computer for reconstruction and viewing as
depicted in Figure 1. The ECT sensor is composed of N where Qi is the accumulated charge on plate i and Γi is a surface
capacitance plates distributed on the wall of the process vessel enclosing the plate. Equation 1 is a linear partial differential
providing N(N - 1)/2 independent capacitance measurements. equation; however, the nonlinear coefficients pose an additional
Recent developments are geared toward an ECT sensor design complexity to the ECT reconstruction problem.
with 3D features for detecting the capacitance variations due Solving eq 2 to obtain the capacitance from a known
to permittivity perturbations in the imaging volume. The permittivity distribution is referred to as the forward problem,
challenge in achieving it arises from the low level of capacitance whereas the problem of finding the permittivity map from the
change compared to noise and stray capacitance in the system. capacitance measurements is referred to as the inverse problem.
The current ECT systems are capable of providing up to 100 In hard-field tomography, the inverse problem is solved by back
frames per second for a 12-electrode system. Increasing the projecting the boundary measurement vector on a perturbation
number of electrodes plays an important role in enhancing the matrix known as the sensitivity matrix; this process of solving
reconstructed image quality. The ECT reconstruction technique the inverse problem is known as linear back projection (LBP).62
is, however, an integral part of the tomography system. In the LBP has been applied to ECT, and the sensitivity matrix in this
following, ECT reconstruction techniques, sensors, and applica- case is constructed through finding the capacitance response of
tions are discussed. an ECT sensor for permittivity perturbations in different
3710 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008

locations of the imaging domain. The permittivity distribution where X is an M × M smoothness matrix and plays the role of
can be calculated by a low-pass filter to remove the noise of the high-frequency
component, and γ1,γ2, and γ3 are normalization constants
G ) STC (3) between 0 and 1. The NN-MOIRT uses a Hopfield neural
network to optimize the weights of the three objective functions.
where G is a 1 × M image vector, S is an N × M sensitivity Hopfield networks possess the advantage of guaranteed con-
matrix, and C is an N × 1 capacitance vector. LBP has been vergence to local minima and can be realized onto hardware
applied successfully for solving the inverse problem in hard components. As a local minimum does not necessarily cor-
field tomography. The use of LBP for image reconstruction has respond to the most optimal solution, a penalty factor has been
attracted attention due to its simplicity and speed in implement- added to the set of objective functions to assist in the energy
ing the algorithm. However, the severely ill-posed ECT problem function escape from entrapment in a local minimum, thereby
and the limited number of capacitance measurements prompted converging to a global minimum of energy.35 In a new
research in image reconstruction for improved image resolution. development, the NN-MOIRT has been extended to include
In this regard, regularization techniques have been used to volume tomography reconstruction through modification of the
enhance image quality by decreasing the level of ill-posedness. ECT sensor to include axial variation along with the addition
The most commonly used regularization method is the Tikhonov of a new objective function.51 The new objective function, noted
regularization. Tikhonov regularization is based on adding an as the matching function, is used to ensure consistency between
invertible diagonal matrix to the ill-posed matrix to make it 2D and 3D volume images.51 This is achieved by comparing
invertible. This method has been applied to ECT image projections of the volume image to the 2D reconstruction results.
reconstruction.63 However, single-step image reconstruction The matching objective function takes the form
techniques did not provide the required image reconstruction
quality, and iterative image reconstruction techniques have been 1
developed and applied for solving the ECT problem.62 fm(G) ) γ4|H2DG3D - G2D|2 (8)
2
Iterative reconstruction techniques are generally classified
into: (1) algebraic and (2) optimization techniques. In algebraic where H2D is the 2D projection matrix, G2D is the 2D image
reconstruction techniques, the reconstructed image is obtained reconstruction result, and γ4 is the normalization constant
through minimization of the mean square error (MSE) objective between 0 and 1. The four objective functions are optimized
function. One of the most used algebraic techniques is the simultaneously to obtain the most likely volume image. The
iterative linear back projection (ILBP).64 The reconstruction use of Hopfield networks is also motivated by its fast calculation
results from the ILBP are obtained through back projecting the speed, convergence of monotonic decrease, and inherent paral-
error vector on the sensitivity matrix iteratively according to lelism of the network.
the following equation Following eq 1, the potential distribution is a function of
charge distribution in addition to boundary conditions and
Gk + 1 ) Gk + RST(C - SGk) (4) physical property (permittivity) distribution. However, image
reconstruction in ECT is based on the electrostatic free charge
where k is the number of iterations, and R is the relaxation factor.
conditions, characterized by the right side of eq 1 for being
Minimization of MSE refers to minimization of the residual
zero. Thus, in imaging fields where a high level of static-charge
error between the capacitance vector and the forward solution
generation is present, the charge would distort the reconstructed
of the reconstructed image. However, because of the severely
ECT image. A remedy for this distortion can be made by
ill-posed ECT problem and noise contamination in the capaci-
correcting the distorted measured capacitance values by those
tance measurements, the reconstructed image with the least MSE
induced by the electrostatic effect.65 The approach for correction
is not necessarily the best solution. In optimization reconstruc-
can be implemented using the ECT capacitance sensor in its
tion, the most likely image is obtained through optimization of
passive mode that measures the voltage difference between
several objective functions related to both the capacitance vector
plates resulting from free charge distribution in the imaging
and the image itself. The NN-MOIRT35 possesses such proper-
domain. This measured voltage is then used to correct the
ties and has been demonstrated to be an accurate reconstruction
applied voltage for each plate in the capacitance measurement
technique. The objective functions used by NN-MOIRT are the
process. Simulation results for static-charge effect are shown
MSE, entropy, and smoothness functions. The MSE function
in section 3.
used in NN-MOIRT is similar to the MSE function in algebraic
2.2. 3D ECT Sensor. In 2D ECT, the sensor has only a 2D
techniques and is aimed at obtaining an image solution that
field variation in radial directions. The sensor in this case is
matches the measured capacitance vector in the forward solution.
assumed to be infinite in length. However, the limited sensor
The MSE function is defined as
length in reality provided distortion to electric field lines in a
phenomenon known as fringing. The fringing effect in 2D ECT
fmse(g) ) γ1|SG - C|2 (5)
has been always viewed as an undesired component, which adds
The entropy function is used to obtain the image with inaccuracy to the 2D reconstructed images. The volume tomo-
maximum information and is defined as graphy concept, or ECVT, is based on utilizing the fringing
effect to produce a field variation in the axial direction for 3D
N imaging.66 Sensor design for volume tomography is an important
fi(g) ) γ2 ∑ Gj ln(Gj)
j)1
(6) part of the image reconstruction. A desired sensor is to be able
to provide an equal distribution of the electric field in all three
dimensions. Thus, the sensitivity variance and strength along
The smoothness function is used to minimize the noise in the 3D imaging domain can be used as criteria for determining
the reconstructed image and is defined as the suitability of a sensor for volume imaging. In an ideal
scenario, a uniform field variation along all dimensions and a
fs(g) ) γ3(GTXG + GTG) (7) proper extent of difference between the maximum and minimum
 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008 3711

Figure 2. Elements of an ECT sensor that can be used for designing a 3D capacitance volume sensor.

field values are required for volume image reconstruction. the application of ECT sensors to include volume imaging leads
However, the ill-posedness of the ECT sensor as well as the to an increased number of unknowns for the same number of
dependence of the field variation on the permittivity distribution capacitance measurements. Thus, the ECVT reconstruction
represented by nonlinear coefficients in the Poisson equation problem is of a higher level of complexity compared to the 2D
presents a challenge in 3D ECT sensor design. In reality, the case due to the increased number of unknowns, severity of the
electric field variation changes significantly in the imaging 3D ill-posed problem, and decreased absolute value of measured
domain depending on the location of the observation point and capacitance due to a lower value of SNR. All of the above-
the difference between the maximum and minimum field values, mentioned factors contribute to a decrease in the resolution of
which may be of several orders of magnitude. As for the former, the output volume image. Clearly, a 3D sensor is needed to
a nonlinear image reconstruction technique is required to solve provide a better image resolution. Thus, using such sensor
the problem concerning nonuniformity. As for the latter, an ECT
designs as shifted layered rectangular plate sensors and multiple
acquisition hardware of high signal-to-noise ration (SNR) is
or single layered non-rectangular plate sensors as noted above
required to detect capacitance signals of low amplitude (on the
could yield a reasonable axial resolution in imaging.51,66
order of femto Farad) and high variation in its absolute value
(2 orders of magnitude). 2.3. Multimodal Tomography Based on ECT System.
The reconstruction techniques and acquisition hardware are Among different process tomography techniques, the most
topics of active ongoing research. The 3D ECT sensor is an conspicuous are those based on the measurement of electrical
evolving topic, and two components of an ECT sensor are properties through the utilization of the capacitive, conductive,
commonly considered in establishing the 3D field variation: (1) or inductive nature of the flow components under investigation.
number of planes, and (2) shape of plates, as depicted in Figure In most cases, electrical tomography is applied based on
2. Examples of the two different design methods are depicted measurements of a single constitutive property; that is, permit-
in Figures 3 and 4. A combination of both strategies can be tivity for ECT or conductivity for impedance (resistivity)
used by integrating irregular shaped plates in multiple planes.66 tomography. However, the need for real-time imaging of
In Figure 3, a triple plane of a square plate shape is used, complex processes involving multiphase components has in
whereas in Figure 4 a trapezoidal plate is used in a single plane. recent years motivated the development of imaging systems
Single-plane capacitance sensors for volume imaging, similar exploiting multiple electrical properties, that is, multimodal
to the one depicted in Figure 4, are most suited for applications tomography.
with limited sensor height. Single-plane arrangements can also
The implementation of multimodal tomography is generally
be used for focused volume imaging covering a relatively lesser
achieved through the integration of more than single tomography
height of the imaging domain. However, it is noted that the
three-plane shifted sensor design depicted in Figure 3 was hardware into one system, using a reconstruction algorithm
among several most viable sensor designs experimentally tested capable of differentiating between different components and
over hundreds of designs conducted in the authors’ lab. Many phases based on a single sensor and signal, or implementing an
of the ECVT applications to multiphase flow systems reported inherently multimodal tomography sensor capable of capturing
by the authors67-68 were based on such a sensor design. The a signal related to different physical properties. An example of
merit behind this particular design is that it increases the axial the first approach is demonstrated in a published work69 where
resolution while reducing its impact on the radial resolution. an ECT system is combined with an ERT system for dual
This design has also been used by other researchers53 in 3D imaging of permittivity and conductivity. Considering the cost,
imaging. complexity, and speed of different multimodal strategies, the
In 2D ECT reconstruction, the sensor is completely utilized multimodal system could be the most versatile. However, not
for providing the maximum 2D resolution. However, extending all tomography sensors possess the capability of multimodal
3712 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008

Figure 3. Sensitivity distributions of selective pairs for a three-layer sensor. Volume enclosed by blue isosurface represents volume covered by indicated
pair of sensors for reconstruction.

Figure 4. Sensitivity distributions of selective pairs of a trapezoidal single-layer sensor. Volume enclosed by blue isosurface represents volume covered by
indicated pair of sensors for reconstruction.

imaging. For the ECT sensor, it has been traditionally viewed approximation can be employed to describe the field behavior.
as a single modal tomography sensor based on electrostatic The ECT analysis reported in the literature is carried out by
analysis. assuming a static approximation for the electric field distribution
In a typical ECT system, the frequency of the excitation signal (modulated by the time variation). In situations related to time
is about 1-10 MHz,70 and the sensor size is less than a few varying fields, the electric and magnetic fields at the observation
meters in either dimension. As a result, the wavelength is much point change values some time after a change at the source.
larger than the size of the sensor, and a static or quasistatic The time (propagation) delay depends on the distance between
 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008 3713

Figure 5. Permittivity and conductivity reconstruction based on quasistatic analysis.

the source and the observation points and on the intervening the inverse problem for this multimodal imaging system. In
media. In the cases where the physical dimensions of the Figure 5, a reconstruction result obtained using the known ILBP
problem are significantly smaller than the wavelength of the reconstruction technique is used to demonstrate the capability
excitation, this propagation delay can be neglected and the static of an ECT sensor for dual permittivity and conductivity imaging.
laws can be applied to time varying fields. Applying a quasistatic The capacitance and power measurements were given using
approximation in Maxwell’s equations, the electric field distri- computer simulations.
bution obeys the following equation
3. Reconstruction and Validation
∇(σ + jω)∇φ ) 0 (9) The effectiveness of NN-MOIRT reconstruction technique
relative to other widely known techniques, such as LBP, ILBP,
where φ(r b) is the electric potential, B b) ) -∇φ is the electric
E(r and simultaneous iterative reconstruction technique (SIRT) can
b) is the conductiv-
field intensity, ω is the angular frequency, σ(r be shown in figures published in earlier work.35 The first and
b) is the permittivity, and b
ity, (r r is the position vector. The second columns in the figures indicate the model images
mutual capacitance Cij between any two pair of electrodes i and (permittivity distributions in the two-phase system). The fol-
j (source and detector) is given by lowing columns show, respectively, the reconstructed images
1 using LBP, ILBP, SIRT, and NN-MOIRT where dark red refers
Cij ) I ∇φijn̂dl (10) to a normalized pixel value of 1 and dark blue refers to that of
∆Vij j
0. The reconstructed images are of 32 × 32 pixel resolution.
The reconstruction process is set to stop when all of the neuron
where ∆Vij is the potential difference, Γj is a closed surface (or
update values are less than 10-4. The results yield an effective
path in 2D) enclosing the detecting electrode, and n̂ is the unit
and noise-immune NN-MOIRT technique as compared to other
normal vector to Γj. Taking under consideration the quasistatic
reconstruction techniques (LBP, ILBP, and SIRT). Similar
analysis, the interrogating signal of the alternating field dis-
results have been obtained in the case of volume reconstruction.
sipates electric power in the presence of materials with nonzero
As depicted in an earlier work,51 the 3D-NN-MOIRT provides
conductivity. Measuring the power dissipated through excitation
better reconstruction results when compared to LBP and
of different ECT pairs of plates provides additional boundary
Landweber reconstruction techniques. The Landweber technique
measurements that can be used for the reconstruction of
is a form of ILBP.64 A sensor similar to the one in Figure 3
conductivity maps. The rms power dissipated by a conductive
was used in this case.
object in the domain of interest given the potential distribution
In Figure 5, a reconstruction result obtained using the known
φij due to the source electrode i at voltage Vi and detector
ILBP reconstruction technique is used to show an ECT sensor
electrode j at ground voltage is given by
that is capable of performing dual permittivity and conductivity
imaging. The capacitance and power measurements were
Pij )
1
2
∫∫Ω σ|∇φij|2 dS (11) obtained using computer simulations. Figure 6 depicts the
reconstructed images of spheres in a bent conduit.51 Utilizing
Equations 10 and 11 relate the permittivity and conductivity the 3D change in sensitivity variations has enabled imaging
distributions to (global) measurements of capacitance and power, complex conduits and objects (unlike the limited traditional
respectively. The solution for Cij and Pij given (r b) and σ(r
b) ECT). Although the reconstructed results in Figure 6 are
constitutes the forward problem. The process of obtaining (r b) elongated at the edges, the results are still reasonably accurate.
and σ(rb) from capacitance and power measurements constitutes Part A of Figure 7 shows the simulation of the effect of 10-8 C
3714 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008

Figure 6. Volume imaging of flow in a bent pipe. (a) Sensing domain, (b) 12-electrode sensor system, (c) reconstructed result of a sphere in the center of
the pipe, a cross-section of the volume image is depicted above, (d) reconstructed result of one and a half spheres located near the sensing edge, a cross-
section of the volume image depicted above. The color bars represent relative permittivity for the cross-section images.

Figure 7. Reconstruction results for the capacitance data in the left image, results of charge elimination are depicted in the lower right of the right image.

charge placed at the center of the domain on capacitance domain on reconstruction results is depicted in the lower-left
measurement of a centered dielectric cylinder. As shown in the image. It is clear from this image that the static charge distorts
figure, the capacitance values, which reflect the permittivity the reconstruction result. The static-charge elimination method
distribution, are contaminated by free charge placement. Part discussed earlier is shown to remove the charge contamination
B of Figure 7 presents the reconstruction results for the in the capacitance vector, yielding desirable reconstruction
capacitance vector depicted in part A of Figure 7. The upper- results as given in the lower right part of the figure. The results
left image refers to the original permittivity distribution, and indicate the effectiveness of the charge-elimination method in
the upper-right image is for the reconstruction without static- providing accurate reconstructed images in situations when static
charge contamination. The effect of free charges in the imaging charges are involved.
 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008 3715

Figure 8. Snapshots of 3D volume images of gas-liquid flow in the bubble column at Ug ) 2 cm/s obtained by ECVT.

4. Experiments and Imaging Results dielectric constant ) 3.8) are used as the gas, liquid, and solids
phases, respectively. The gas velocity ranges are from 2.5 to
In this section, sample results obtained by the ECVT for the 15 cm/s. The ECVT sensor is located 18 cm above the
multiphase flow systems are presented. The systems considered distributor, and the total height of the measuring plane is 10
include the bubble column and gas-liquid-solid three-phase cm. The initial bed height for the bubble-column experiment is
fluidized bed,35,71-72 gas-solid fluidized bed,73-77 and circulat- set at 35.5 cm above the distributor. The images and data
ing fluidized bed riser.78-81 presented by the ECVT represent those in a real 3D space
The ECVT that used the sensor design given in Figure 3 confined by the height of the sensor plane and the diameter of
comprises a 12-electrode rectangular sensor arranged in triple the column.
planes. Its performances for 3D volume imaging are examined. The image reconstruction algorithm generates the permittivity
The choice of the electrode number is based on the data map determined on or voxels (ECVT) corresponding to the
acquisition system available for experiments, which are 12 apparent permittivity of the mixed two or three phases of the
channels. The electrode number significantly greater than 12, system imaged. To convert the permittivity map to the phase
depending on the availability of the data acquisition system, distribution, a combined series and parallel capacitance models
can be used, which would appreciably increase the spatial can be used.82 The model relates the permittivity map into the
resolution of the images obtained. The electrical field intensity phase concentration in the mixed media through series and
with the electrode arrangement given in Figure 3 can be parallel capacitance connectivity with the same probability.
distributed reasonably uniformly in axial and radial directions. Figure 8 shows a snapshot of the ECVT image (3D gas
Such sensor design is equivalent to the rectangular sensor concentration distribution) of the gas-liquid system at a gas
arrangement of eight-electrode sensors per plane. The length velocity of 2 cm/s. The tomography volume image is constructed
of the sensing domain is 10 cm. The volume images are from permittivity voxel values in 4D matrix components, that
reconstructed at 20 × 20 × 20 resolution based on the algorithm is, three space components with a spatial resolution of 5 × 5 ×
described above. There are 66 combinations of independent 5 mm3 and a one-time component with a temporal resolution
capacitance measurements between electrode pairs from 12 of 12.5 ms. The first two figures in the top row are slice cut
electrodes. The results are obtained from a 12-channel data images of the planes defined by the coordinate system in the
acquisition system (DAM200-TP-G, PTL Company, U.K.), and bottom-right of the figure. The first and second figures in the
the image capture speed is of 80 frames per second. The bottom row are, respectively, a 3D volume image that is partly
reconstruction process and data postprocessing are obtained from cutoff to display the inside of the 3D representation and a 3D
a 3 GHz Pentium 4 machine, with a memory of 2 GB. isosurface image that displays the 3D boundary (surface) of the
4.1. Bubble Column. The imaging results for a bubble bubble swarm image. The cutoff boundary value was set at 10%
column are obtained from a column size of 0.1 m ID by 1.0 m of the gas holdup value. There is no specific criterion in setting
height. The top of the column is the enlargement section with this cutoff boundary; it is arbitrary and is used to provide some
a diameter of 0.2 m. The gas distributor is a single nozzle with sense of distinction of the boundary of high-concentration bubble
a diameter of 0.5 cm. The experiments are conducted in swarm from the surrounding low gas concentration region. For
semibatch mode without inlet liquid flow. Air (dielectric comparisons with the tomography images, a photograph of the
constant ) 1), Norpar 15 (paraffin, density ) 773 kg/m3, two-phase flow taken using a high-speed digital video camera
viscosity ) 0.253 m Pa/s, dielectric constant ) 2.2) and glass system under the same condition is displayed in the right-hand
beads (density ) 2500 kg/m3, average diameter ) 200 µm, side of the figure.
3716 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008

Figure 9. Snapshot of volume image of bubble in gas-solid fluidized bed using 200 µm glass beads. (a) 3D volumetric (partly chopped) distribution of
solids concentration obtained by ECVT, (b) isosurface image of bubble (Ug ) 0.1 m/s), (c) (d) X-ray photograph taken by Rowe (1971).

4.2. Gas-Solid Flows. 4.2.1. Gas-Solid Fluidized Bed. the riser and exit at the top through a right-angled bend into a
Three fluidized beds with diameters of 0.05, 0.1, and 0.3 m ID horizontal tube connected to the separator and secondary cyclone
are examined for the scale effect on ECVT measurements. A where the particles are separated from the gas. Subsequently,
two-stage cyclone separates gas and particles for each fluidized the particles are fed back to the bottom of the riser through the
bed and is installed in the freeboard of the bed. A porous plate nonmechanical L-valve. The solids’ circulation rate, which is
with a pore size of 20 µm and a fractional free area of 60% is controlled by adjusting the air aeration rate at the injection points
employed as a distributor for all fluidized beds. The fluidized of the L-valve, is measured by timing the falling distance of
particles are FCC catalyst with a mean diameter of 60 µm and tracer particles in the standpipe. The compressed air is intro-
a particle density of 1400 kg/m3. The measuring sensor for the duced into the riser through an oil filter, humidifier, pressure
ECVT is located 18 cm above the distributor, and the total height manometer, and flow meter. The air humidity is controlled by
of the measuring plane is 10 cm. The static bed height is 0.5 m the water level and water temperature in the humidifier. The
for all fluidized beds. relative humidity and temperature are measured by means of a
Figure 9 shows the snapshot of the tomography volume image humidity probe inserted into the air stream. The superficial gas
of a 0.1 m ID gas-solid fluidized bed with 200 µm glass beads velocity is measured by the flow meter adjusted by the
at a gas velocity of 0.2 m/s. The 3D solids concentration temperature and pressure of the airflow. The fluidized particles
distribution in the bed is obtained by the ECVT, as shown in employed in this study are FCC catalysts with a mean diameter
part a of Figure 9. The color image from blue to red represents of 60 µm and particle density of 1400 kg/m3 and sand particles
the solids’ concentration spanning from low (empty bed) to high with a mean diameter of 240 µm and particle density of 2200
(packed bed). A bubble with a spherical cap shape is clearly kg/m3. The measuring sensor for the ECVT is located at 0.48
observed in part b of Figure 9, in which the 3D isosurface image m above the distributor of the riser.
is obtained by setting the cutoff boundary of 25% (solids Figure 10 compares the quasi-3D flow structure obtained by
concentration) in the solids concentration distribution profile combining a sequence of ECT images to form a quasi-3D image,
shown in part a of Figure 9. The 3D bubble shape obtained by and the real 3D flow structure obtained by the ECVT in a 0.1
the ECVT is consistent with the quasi-3D bubble obtained by m ID circulating fluidized bed with 200 µm sand particles (group
the ECT technique.74 Parts c and d of Figure 9 show the X-ray B particles). The quasi-3D imaging provides quasi-3D images
photography of the bubble in the gas-solid fluidized bed with from combining 2D cross-sectional ECT images obtained from
crushed coal particles (dp ) 530 µm, Fp ) 1250 kg/m3) and averaging over an axial distance of 5 cm, representing the
magnetite particles (dp ) 2800 µm, Fp ) 2930 kg/m3) thickness of the electrodes. In this kind of imaging, time serves
,respectively.82-83 The similarity of the images affirms the as the third dimension. A color bar from blue to red represents
ECVT technique. the variation of the solids’ concentration from 0 to 0.3. The 3D
4.2.2. Circulating Fluidized Bed. The circulating fluidized flow structure or the volume image is constructed from
bed consists of a 0.1 m ID riser with a height of 6.32 m, a permittivity voxel values in 3D matrix components, that is, three
separator and secondary cyclone system, a large-volume particle space components with spatial resolution of 5 × 5 × 5 mm3
storage hopper and an L-valve. Particles are carried upward in obtained by ECVT. Under the operating conditions of Ug )
 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008 3717

Figure 10. 2D and 3D flow structure variation during the choking transition in a 0.1 m ID circulating fluidized bed with 200 µm glass beads at Ug ) 2.4
m/s.

2.4 m/s and Gs ) 14.86 kg/m2s for group B particles, the bed the reconstruction algorithm on hardware devices, given the
undergoes the choking transition from the three-region structure already attractive low sensor construction cost and its safe
with solids blobs at the central region to the formation of wall applications.
slugs as shown in the quasi-3D diagram in Figure 10. The
similar bed density variation during the choking transition is Literature Cited
observed on the basis of the real 3D flow structure with the
existence of solid blobs at the center of the column before the (1) Fan, L.-S. Gas-Liquid-Solid Fluidization Engineering; Butter-
worths: Markham, ON, Canada, 1989.
choking transition and the formation of wall slugs after the
(2) Fan, L.-S.; Zhu, C. Principles of Gas-Solid Flows; Cambridge
choking transition. The similarity between the quasi-3D by ECT University Press: New York, 1998.
and the real 3D flow structure by ECVT further substantiates (3) Chen, R. C.; Fan, L.-S. Particle Image Velocimetry for Characterizing
the inherent characteristics of the bed behavior and the choking the Flow Structures in Three-Dimensional Gas-Liquid-Solid Fluidized Beds.
phenomena in the gas-solid circulating fluidized bed. Chem. Eng. Sci. 1992, 47, 3615-3622.
(4) Reese, J.; Chen, R. C.; Fan, L.-S. Three-Dimensional Particle Image
Velocimetry for Use in Three-Phase Fluidization Systems. Exp. Fluids 1995,
5. Concluding Remarks 19, 367-378.
(5) Reese, J.; Jiang, P.; Fan, L.-S. Bubble Characteristics in Three-Phase
This study presents a perspective of the development of the Systems Used for Pulp and Paper Processing. Chem. Eng. Sci. 1996, 51,
electrical capacitance tomography imaging techniques, particu- 2501-2510.
larly with respect to their recent evolution of the techniques (6) Vial, C.; Laine, R.; Poncin, S.; Midoux, N.; Wild, G. Influence of
from 2D to 3D imaging. It describes the early ECT imaging Gas Distribution and Regime Transitions on Liquid Velocity and Turbulence
in a 3-D Bubble Column. Chem. Eng. Sci. 2001, 56, 1085-1093.
efforts that were restricted to 2D, averaged along sensor length. (7) Kulkarni, A.; Joshi, J.; Ramkrishna, D. Determination of Bubble
The technique was extended to the 3D domain through the Size Distribution in Bubble Columns Using LDA. AIChE J. 2004, 50, 3068-
stacking of 2D images obtained in different times or planes in 3084.
what is known as quasi-3D imaging. Most recently, direct 3D (8) Cui, H.; and Chaouki, J. Effects of Temperature on Local Two-
imaging was realized through the introduction of the volume Phase Flow Structure in Bubbling and Turbulent Fluidized Beds of FCC
Particles. Chem. Eng. Sci. 2004, 59, 3413-3422.
tomography concept in what is noted as ECVT. The ECVT
(9) Cui, Z.; Fan, L.-S. Energy Spectra for Interactive Turbulence Fields
development was made possible through advances in sensor in a Bubble Column. Ind. Eng. Chem. Res. 2005, 44, 1150-1159.
design and 3D image reconstruction. In principle, the spatial (10) Braeske, H.; Brenn, G.; Domnick, J.; Durst, F.; Melling, A.; Ziema,
resolution in volume imaging can be significantly improved with M. Extended Phase-Doppler Anemometry for Measurements in Three-Phase
increased number of sensors augmented with innovative sensor Flows. Chem. Eng. Technol. 1998, 21, 415-420.
design and image reconstruction techniques. (11) Brenn, G.; Braeske, H.; Durst, F. Investigation of the Unsteady
Two-Phase Flow with Small Bubbles in a Model Bubble Column Using
The use of volume imaging concept allows the ECVT to be Phase-Doppler Anemometry. Chem. Eng. Sci. 2002, 57, 5143-5159.
applied to qualify and quantify the 3D dynamics of multiphase (12) Bemrose, C. R.; Fowles, P.; Hawkesworth, M. R.; O’Dwyer, M.
flow systems, which are of considerable interest to the mul- A. Application of Positron Emission Tomography to Particulate Flow
tiphase flow and other fields. The multiphase flow systems of Measurement in Chemical Engineering Processes. Nucl. Instrum. Methods
which the ECVT has been applied include bubble columns, Phys. Res., Sect. A 1988, A273, 874-880.
(13) Stein, M.; Ding, Y. L.; Seville, J. P. K.; Parker, D. J. Solids Motion
gas-solid fluidized beds, and risers. These are systems that are
in Bubbling Gas Fluidized Beds. Chem. Eng. Sci. 2000, 55, 5291-5300.
widely practiced in industry. The potential for ECVT usage in (14) Dechsiri, C.; Ghione, A.; van de Wiel, F.; Dehling, H. G.; Paans,
industry can be enhanced by the further improved accuracy and A. M. J.; Hoffmann, A. C. Positron Emission Tomography Applied to
image reconstruction speed and the added capability of realizing Fluidization Engineering. Can. J. Chem. Eng. 2005, 83, 88-96.
3718 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008

(15) Hoffmann, A. C.; Dechsiri, C.; Van de Wiel, F.; Dehling, H. G. (40) Scott, D. M.; McCan, H. Process Imaging For Automatic Control;
PET Investigation of a Fluidized Particle: Spatial and Temporal Resolution Taylor & Francis Group: Boca Raton, FL, 2005.
and Short Term Motion. Meas. Sci. Technol. 2005, 16, 851-858. (41) Fasching, G. E.; Smith, N. S. High Resolution Capacitance Imaging
(16) Chen, M.; M., Lin, J. S.; Chao, B. T. Computer-Aided Particle System; DOE/METC-88/4083; U. S. Dept. of Energy, 1988; p 37.
Tracking, a Technique for Studying Solid Particle Dynamics in Gas (42) Tomographic Techniques for Process Design and Operation
Fluidized Beds, AIChE Annual Meeting, New Orleans, November 8-12, (Computational Mechanics: Southampton, U. K.) 1993. Process Tomo-
1981; paper 65e. graphy - A Strategy for Industrial Exploitation (UMIST: Manchester, U.
(17) Cassanello, M.; Larachi, F.; Marie, M. N.; Guy, C.; Chaouki, J. K.) 1993; Beck M. S., Williams, R. A., Campogrande, E., Morris, M. A.,
Experimental Characterization of the Solid Phase Chaotic Dynamics in Waterfall, R. C., Eds.
Three-Phase Fluidization. Ind. Eng. Chem. Res. 1995, 34, 2971-80. (43) Yang, W. Q.; Stott, A. L.; Beck, M. S.; Xie, C. G. Development
(18) Larachi, F.; Cassanello, M.; Chaouki, J.; Guy, C. Flow Structure of Capacitance Tomographic Imaging Systems for Oil Pipeline Measure-
of the Solids in a 3-D Gas-Liquid-Solid Fluidized Bed. AIChE J. 1996, 42, ments. Meas. Sci. Technol. 1996, 22, 144-145.
2439-2452. (44) Dyakowski, T.; Luke, S P; Ostrowski, K. L.; Williams, R. A. On-
(19) Larachi, F.; Chaouki, J.; Kennedy, G.; Dudukovic, M. P. Radioac- Line Monitoring of Dense Phase Flow Using Real Time Dielectric Imaging.
tive Particle Tracking in Multiphase Reactors: Principles and Applications. Powder Technol. 1999, 104, 287-295.
In: Non-InVasiVe Monitoring of Multiphase Flows; Chaouki, J., Larachi,
(45) Dyakowski, T.; Edwards, R B; Xie, C. G.; Williams, R. A.
F., Dudukovic, M. P., Eds.; Elsevier: Amsterdam, 1997; 335-406.
Application of Capacitance Tomography to Gas-Solid Flows. Chem. Eng.
(20) Chaouki, J.; Larachi, F.; Dudukovic, M. P.; Noninvasive Tomo- Sci., 1997, 52, 2099-2110.
graphic and Velocimetric Monitoring of Multiphase Flows. Ind. Eng. Chem.
(46) Halow, J. S.; Nicoletti, P. Observation of Fluidized Bed Coalescence
Res. 1997, 36, 4476-4503.
Using Capacitance Imaging. Powder Technol. 1992, 69, 255-277.
(21) Chen, J.; Kemoun, A.; Al-Dahhan, M. H.; Dudukovic, M. P.; Lee,
D. J.; Fan, L.-S. Comparative Hydrodynamics Study in a Bubble Column (47) Chaplin, G.; Pugsley, T.; Lee, L.; Kantzas, A.; Winters, C. The
Using Computer-Automated Radioactive Particle Tracking (CARPT)/ Dynamic Calibration of an Electrical Capacitance Tomography Sensor
Computed Tomography (CT) and Particle Image Velocimetry (PIV). Chem. Applied to the Fluidized Bed Drying of Pharmaceutical Granule. Meas.
Eng. Sci. 1999, 54, 2199-2207. Sci. Technol. 2005, 22, 1281-1290.
(22) Kiared, K.; Larachi, F.; Chaouki, J.; Guy, C. Mean and Turbulent (48) Pugsley T. S. Application of Electrical Capacitance Tomography
Particle Velocity in the Fully Developed Region of a Three-Phase Fluidized to Fluidized Beds Operating in Different Regimes of Fluidization; Arena,
Bed. Chem. Eng. Technol. 1999, 22, 683-689. U., Chirone R., Miccio, M., Salatino, P., Eds.; Fluidization XI, Engineering
(23) Nedeltchev, S.; Kumar, S. B.; Dudukovic, M. P. Flow Regime Conferences International publication, 2004, 107-114.
Identification in a Bubble Column Based on Both Kolmogorov Entropy (49) Reinecke, N.; Mewes, D. Investigation of the Two-Phase Flow in
and Quality of Mixedness Derived from CARPT Data. Can. J. Chem. Eng. Trickle-Bed Reactors Using Capacitance Tomography. Chem. Eng. Sci.
2003, 81, 367-374. 1997, 52, 2111-2127.
(24) Wolf, J. Investigation of Bubbly Flow by Ultrasonic Tomography. (50) Warsito, W; Marashdeh, Q.; Fan, L.-S. Electrical Capacitance
Particle & Particle Systems Characterization; 1988, 5, 170-173. Volume Tomography (ECVT): Sensor Design and Image Reconstruction,
(25) Xu, L.; Han, Y.; Xu, L. A.; Yang, J. Application of Ultrasonic Proceedings of the 4th World Congress on Industrial Process Tomography,
Tomography to Monitoring Gas/Liquid Flow. Chem. Eng. Sci. 1997, 52, Aizu, Japan, 2005; p 827.
2171-2183. (51) Warsito, W.; Marashdeh, Q.; Fan, L.-S. Electrical Capacitance
(26) Utomo, M. B.; Warsito, W.; Sakai, T.; Uchida, S. Analysis of Volume Tomography (ECVT). IEEE Sens. J. 2007, 7, 525-535.
Distributions of Gas and TiO2 Particles in Slurry Bubble Column Using (52) Mazurkiewicz, L.; Banasiak, R.; Wajman, R.; Dyakowski, T.;
Ultrasonic Computed Tomography. Chem. Eng. Sci. 2001, 56, 6073-6079. Sankowski D. Towards 3D Capacitance Tomography, Proceedings of the
(27) Gladden, L. F. Nuclear Magnetic Resonance in Chemical Engineer- 4th World Congress on Industrial Process Tomography, Aizu, Japan, 2005;
ing: Principles and Applications. Chem. Eng. Sci. 1994, 49, 3339-3408. 546-551.
(28) Leblond, J.; Javelot, S.; Lebrun, D.; Lebon, L. Two-Phase Flow (53) Wajman, R.; Banasiak, R.; Mazurkiewicz, L.; Dyakowski T.;
Characterization by Nuclear Magnetic Resonance. Nucl. Eng. Des. 1998, Sankowski, D. Spatial Imaging with 3D Capacitance Measurements. Meas.
184, 229-237. Sci. Technol. 2006, 17, 2113-2118.
(29) Le Gall, F.; Pascal-Ribot, S.; Leblond, J. Nuclear Magnetic (54) Nurge, M. Electrical Capacitance Volume Tomography with High
Resonance Measurements of Fluctuations in Air-Water Two-Phase Flow: Contrast Dielectrics Using a Cuboid Sensor Geometry. Meas. Sci. Technol.
Pipe Flow with and without “Disturbing” Section. Phys. Fluids 2001, 13, 2007, 18, 1511-1520.
1118-1129. (55) Deshpande, M. personal communication, 2007.
(30) Gladden, L. F. Recent Advances in MRI Studies of Chemical (56) Marashdeh, Q.; Warsito, W.; Fan, L.-S.; Teixeira, F. A Multimodal
Reactors: Ultrafast Imaging of Multiphase Flows. Top. Catal. 2003, 24, Tomography System Based on ECT Sensors: IEEE Sens. 2007, 7, 426-
19-28. 433.
(31) Lim, M. H. M.; Sederman, A. J.; Gladden, L. F.; Stitt, E. H. New
(57) Korjenevsky, A. Detection of Flammable and ExplosiVe Liquids
Insights to Trickle and Pulse Flow Hydrodynamics in Trickle-Bed Reactors
Inside Sealed Vessels Using Quasistatic Electrical Tomography, Proceedings
Using MRI. Chem. Eng. Sci. 2004, 59, 5403-5410.
of the 5th World Congress on Industrial Process Tomography, Bergen,
(32) Sederman, A. J.; Gladden, L. F. Transition to Pulsing Flow in
Norway, 2007; 1002-1006.
Trickle-Bed Reactors Studied Using MRI. AIChE J. 2005, 51, 615-621.
(33) George, D. L.; Shollenberger, K. A.; Torczynski, J. R.; O’Hern, T. (58) Liu, S.; Chen, Q.; Xiong, X.; Zhang, Z.; Liu, X. K. Keynote:
J.; Ceccio, S. L. Three-Phase Material Distribution Measurements in a Preliminary Study on ECT Imaging of Flames in Porous Media, Proceedings
Vertical Flow Using Gamma-Densitometry Tomography and Electrical- of the 5th World Congress on Industrial Process Tomography, Bergen,
Impedance Tomography. Int. J. Multiphase Flow 2001, 27, 1903-1930. Norway, 2007; 365-372.
(34) West, R. M.; Tapp, H. S.; Spink, D. M.; Bennett M. A.; Williams, (59) Datta, U.; Mylvaganam, S. AVerage Particle Size and Instantaneous
R. A. Application-Specific Optimization of Regularization for Electrical Mass Flow Rate Estimation in Dilute Phase Pneumatic ConVeying Systems
Impedance Tomography. Meas. Sci. Technol. 2001, 12, 1050-1054. using ECT, Proceedings of the 5th World Congress on Industrial Process
(35) Warsito, W.; Fan, L. S. Measurement of Real-Time Flow Structures Tomography, Bergen, Norway, 2007; 784-792.
in Gas-Liquid and Gas-Liquid-Solid Flow Systems Using Electrical (60) Maad, R.; Sankowski, D.; Grudzein, K.; Wajman, R.; Mosorov,
Capacitance Tomography (ECT). Chem. Eng. Sci. 2001, 56, 6455-6462. V.; Johansen, G. A. Comparing Analysis of Image Visualisation Accuracy
(36) Warsito, W.; Fan, L.-S. DeVelopment of Three-Dimensional Electri- of Electrical Capacitance Tomography and Gamma Tomography, Proceed-
cal Capacitance Tomography, In Proceedings of the 3rd World Congress ings of the 5th World Congress on Industrial Process Tomography, Bergen,
on Industrial Process Tomography, Banff, Canada, 2003; 391-396. Norway; 2007; 1090-1098.
(37) Kim, M. C.; Kim, K. Y.; Kim, S. Improvement of Impedance (61) Marashdeh, Q.; Warsito, W.; Fan, L.-S.; Teixeira, F. Nonlinear
Imaging for Two-Phase Systems with Boundary Estimation Approach in Forward Problem Solution for Electrical Capacitance Tomography Using
Electrical Impedance Tomography. Can. J. Chem. Eng. 2005, 83, 55-63. Feed-Forward Neural Network: IEEE Sens. 2006, 6, 441-449.
(38) Ceccio, S. L. QuantitatiVe Measurements Using Electrical Imped- (62) Yang, W Q; Peng, L. Image Reconstruction Algorithms for
ance Tomography, 6th International Conference on Multiphase Flow, Electrical Capacitance Tomography. Meas. Sci. Technol. 2003, 14, R1-
Leipzig, Germany, July 9-13, 2007; paper KN13. R13.
(39) Williams, R. A.; Beck, M. S. Process Tomography: Principles, (63) Peng, L H; Merkus, H; Scarlett, B Using Regularization Methods
Techniques, and Applications; Butterworth & Heinemann Ltd.: Oxford, for Image Reconstruction of Electrical Capacitance Tomography. Part. Part.
U. K., 1995. Syst. Charact., 2000, 17, 96-104.
 Ind. Eng. Chem. Res., Vol. 47, No. 10, 2008 3719

(64) Yang, W. Q.; Spink, D. M.; York, T. A.; McCann, H. An Image- (74) Du, B.; Warsito, W.; Fan, L.-S. Bed Non-Homogeneity in Turbulent
Reconstruction Algorithm Based on Landweber’s Iteration Method for Gas-Solid Fluidization. AIChE J. 2003, 49, 1109-1126.
Electrical-Capacitance Tomography. Meas. Sci. Technol. 1999, 10, 1065- (75) Du, B.; Warsito, W.; Fan, L.-S. ECT Studies of Gas-Solid Fluidized
1069. Beds of Different Diameters. Ind. Eng. Chem. Res. 2005, 44, 5020-5030.
(65) Fan, L. S.; Park, A. H.; Marashdeh, Q.; Du, B.; Warsito, W. (76) Du, B.; Warsito, W.; Fan, L.-S. Transient 3-Dimensional BehaVior
Electrostatic Tomography for Multiphase Process Imaging, Proceedings of Gas-Solid Fluidization Measured Using Electrical Capacitance Volume
of the 5th World Congress on Particle Technology, Orlando, Florida, 2006. Tomography (ECVT), AIChE Annual Meeting, Cincinnati, OH, October,
(66) Warsito, W.; Marashdeh, Q.; Fan, L.-S. 3D and Real-Time Electrical 2005.
Capacitance Volume-Tomography: Sensor Design and Image Reconstruc- (77) Du, B.; Warsito, W.; Fan, L.-S. Flow Dynamics of Gas-Solid
tion, International Patent PCT/US2006/010352, disclosure date: Dec. 29, Fluidized Beds with Evaporative Liquid Injection. China Particuology 2006,
2004, filing date: March 22, 2005, and publication date: September 28, 4, 1-8.
2006. (78) Du, B.; Warsito, W.; Fan, L.-S. ECT Studies of the Choking
(67) Warsito, W.; Du, B.; Fan, L.-S. Real Time, 3-D ECT Imaging of Phenomenon in a Gas-Solid Circulating Fluidized Bed. AIChE J. 2004, 50,
Multiphase Flow Systems, AIChE Annual Meeting, San Francisco, CA, 1386-1406.
November 16-21, 2003; Paper 509c. (79) Du, B.; Fan, L.-S. Characteristics of Choking Behavior in Circulat-
(68) Marashdeh, Q.; Warsito, W.; Fan, L.-S.; Teixeira, F. Electrical ing Fluidized Beds for Group B Particles. Ind. Eng. Chem. Res. 2004, 43,
Capacitance Sensor Design for 3D Applications, IEEE 2005 Proceedings 5507-5520.
of International Symposium on Antennas and Propagation (URSI), Wash- (80) Du, B.; Warsito, W.; Fan, L.-S. Behavior of the Dense-Phase
ington D. C., June 3-8, 2005. Transportation Regime in a Circulating Fluidized Bed. Ind. Eng. Chem.
(69) Jiang, P.; Peng L.; Lu G.; Xiao D.; Wang, H. Modeling of an ECT/ Res. 2006, 45, 3741-3751.
ERT Dual-modality Tomography Sensor for Oil/Gas/Water Three- (81) Du, B.; Warsito, W.; Fan, L.-S. Imaging the Choking Transition
component Flow Measuring, Proceedings of the 5th World Congress on in Gas-Solid Risers Using Electrical Capacitance Tomography. Ind. Eng.
Industrial Process Tomography, Bergen, Norway, 2007; 592-598. Chem. Res., 2006, 45, 5384-5395.
(70) York, T. Status of Electrical Tomography in Industrial Applications. (82) Rowe, P. N. Experimental Properties of Bubbles. In Fluidization;
J. Electron. Imag. 2001, 10, 608-619. Davidson, J. F., Harrison, D., Eds.; Academic Press: New York, 1971.
(71) Warsito, W.; Fan, L.-S. Dynamics of Spiral Bubble Plume Motion (83) Rowe, P. N. Prediction of Bubble Size in a Gas Fluidized Bed.
in the Entrance Region of Bubble Columns and Three-Phase Fluidized Beds Chem. Eng. Sci. 1976, 31, 285-288.
Using 3D ECT. Chem. Eng. Sci. 2005, 60, 6073-6084.
(72) Warsito, W.; Fan, L.-S. ECT Imaging of Three-Phase Fluidized ReceiVed for reView October 9, 2007
Bed Based on Three-Phase Capacitance Model. Chem. Eng. Sci. 2003, 58, ReVised manuscript receiVed December 3, 2007
823-832. Accepted December 4, 2007
(73) Du, B.; Fan, L.-S.; Wei, F.; Warsito, W. Gas and Solids Mixing in
a Turbulent Fluidized Bed. AIChE J. 2002, 48, 1896-1909. IE0713590