Industrial Process Tomography - Platform II grant funded by EPSRC

FACULTY OF ENGINEERING

 

Typical EIT sensors




Adjacent Straitegy
ECT Sensor


Sensors

Sensor design is an important part of EIT as it plays a significant role in terms of the overall quality of data following image reconstruction.

Traditionally, electrode configuration in EIT comprises numerous sensors (typically 16) arranged equidistantly around the periphery of a vessel or pipe. The sensors in EIT systems must be in continuous electrical contact with the electrolyte inside the process vessel. In addition, the sensors must be more conductive than the electrolyte in order to obtain reliable measurements. For the majority of process applications, the electrodes are metallic and can be fabricated from stainless steel, silver, gold, platinum, silver palladium or any suitable material exhibiting a number of properties. The most favourable properties, in no inferred order of priority, are low-cost, ease of fabrication and installation, good electrical conduction and resistance to abrasion and corrosion. The electrodes are located equi-distantly around the process vessel to map resistivity changes across the plane or planes of interest.

Data Collection Strategies

Different data collection strategies are available for extracting a full dataset of measurements (based on conductivity distribution) at the boundary walls of the vessel.

a) Adjacent Strategy

Figure 1
Adjacent Electrode Pair Strategy

In this method, popularised by [1] current is applied through an adjacent pair of electrodes and voltage measurement is made through successive pairs of neighbouring electrodes. Current is then switched to an alternate pair of electrodes and voltage measurements are made at a second set of electrodes. This is repeated until all independent measurements have been completed. As voltage measurements are not taken from current carrying electrodes, the number of independent measurements, M = N(N-3)/2. Where N (total number of electrodes) in an array is 16, the number of independent measurements is 104. This method is quick in terms of image reconstruction and is undemanding of computational memory. It also requires a minimum amount of hardware to implement [2]. However, the current density in the centre of the vessel will be lower, as the current remains close to the electrodes; the method is susceptible to measurement error and noise interference [3]

b) Diagonal Strategy

Current is injected at electrodes separated by large distances. For a 16-electrode configuration, a current reference is traditionally set at electrode one and voltage measurement at electrode two. Current is then injected at electrodes three, five, seven and so on up to electrode 15. Voltages from all electrodes, apart from current carrying electrodes, are measured with reference to electrode two. Current reference is then switched to electrode four and voltage reference is similarly switched to electrode three. Current is again applied to electrodes 6, 8, 10, 12, 14, 16 and 2, with voltage measurements taken on all electrodes apart from the current injecting one. For each pair of current electrodes, 13 voltage measurements are recorded and seven different current electrode pairs selected; 7 x 13 = 91 data points (a further 91 data points obtained by altering the current and voltage reference electrodes). A 16-electrode system yields a maximum of 182 data points (104 of which are independent). This method does not have high sensitivity in the periphery compared with adjacent method, however, it is not as sensitive to measurement error and therefore often results in a better quality image [3].

c) Opposite Strategy

Current is applied to electrodes on opposite sides of the vessel/pipeline. The electrode adjacent to current injecting electrode is used for voltage measurement. Current is then switched to the next pair of opposing electrodes (in a clockwise direction), with voltage measurement electrodes also moved around so as to remain in formation with the current injecting electrode. This method is less sensitive to conductivity change (as in adjacent strategy) as current flows through the centre point of the cross-section. This gives rise to even distribution of currents, leading to good image characterisation [3]. The number of independent current projections however is lower than for the adjacent strategy (for same number of electrodes, N) [4]. The number of independent measurements, M = N/4 x (3N/3 – 1). For N = 16, M = 92. Image resolution will be decreased by as much as 23% (compared to adjacent strategy) for the same number of electrodes [5].

d) Conducting Boundary Strategy

Designed by [6] for use on pipes/process vessels with conducting boundaries (metallic), this method uses only two electrodes. Conducting strategy has lower common-mode voltage (CMV) than the adjacent strategy as the large surface area of the conducting boundary acts as the current sink. However, measured voltage amplitudes are lower (by a factor of seven) for conducting boundary strategy compared to adjacent strategy for identically shaped process vessels. It has been suggested [6] that “higher CMV component compensates the lower amplitude measurements”, leading to the two strategies having similar sensitivity.

Adjacent Collection strategy
a.

Adjacent Collection strategy
b.
Adjacent Collection strategy
c.

Adjacent Collection strategy


d.

Figure 2. a. Adjacent Collection strategy, b. Diagonal measurement strategy, c. Opposite measurement strategy, d. Conducting boundary measurement

Reference:

[1] W.Q. Yang and L. Peng, “Image reconstruction algorithms for electrical capacitance tomography,” Meas. Sci. Technol., vol. 14, pp. 1-13, 2003.
[2] W.Q. Yang, D.M. Spink, T.A. York and H. McCann, “An image-reconstruction algorithm based on landweber’s iteration method for electrical-capacitance tomography,” Meas. Sci. Technol., vol. 10, pp. 1065-1069, 1999.
[3] W.R. Breckon and M.K. Pidcock, “Mathematical aspects of impedance imaging,” Clin. Phys. Physiol. Meas. A., vol. 8, pp. 77-84, 1985.
[4] T.J. Yorkey, “Comparing reconstruction algorithms for electrical impedance tomography, PhD Thesis,University of Wisconsin, 1986.
[5] M. Wang, W. Yin and N. Holliday, IoP Meas. Sci. Technol., vol. 13, pp. 1884-1889, 2002.
[6] M. Wang, “Inverse solutions for electrical impedance tomography based on conjugate gradients methods,” Meas. Sci. Technol., vol. 13, pp. 101-117, 2002.