preview

Phantom Lab Analysis

Decent Essays

The accuracy of the CT measurements has been recently addressed and quantified in our previous publication through scanning Plexiglas® phantom that was consisted of two concentric cylinders (inner cylinder with 3 in. diameter while the outer cylinder with 6 in.). Four different cases of this phantom (i.e., empty phantom, inner cylinder was filled with water while the space between coaxial cylinders was empty, inner cylinder was empty while the space between them was filled with water only, and both cylinders were filled with water) were scanned independently and then the linear attenuation coefficients (μ, cm-1) for these different cases of the phantom were reconstructed by using alternating minimization (AM) algorithm. The experimental …show more content…

As seen in these figures, the cross-sectional gas holdup distributions that are measured and reconstructed for experiment No.1 and No. 2 for either superficial gas velocities of 5 or 30 cm/s are qualitatively identical. Moreover, the azimuthally averaged gas holdup profiles of the experiments No.1 and No. 2 for the same operating conditions (at either superficial gas velocities 5 or 30 cm /s) are close to each other along the diameter of the bubble column, indicating the high precision and the reliability of CT measurements. For instance, the average absolute relative difference (AARD) between two profiles for each superficial gas velocity was calculated by Eq. 2, and it was found to be 2.17% and 3.47% for a superficial gas velocity of 5 and 30 cm, respectively.
AARD=1/N ∑_(i=1)^N▒|(ε_1 (r)-ε_2 (r))/(ε_1 (r))| (2)

where ε_1 (r) and ε_2 (r) represent gas holdup values of experiment No.1 and No.2, respectively at the corresponding dimensionless radius positions while N represents the number of data points along the diameter of the column.
Furthermore, the standard deviation (SD), which represents the devision of the measured values of gas holdup from the mean 〈ε〉 of these values along the diametrical profiles, was also calculated by Eq. 3.
SD=√(1/(N-1) ∑_(i=1)^N▒(ε_i-〈ε〉)^2 ) (3)

It was found that SD values were minimal, within

Get Access