A Bayesian approach to characterising multi-phase flows using magnetic resonance: Application to bubble flows

• Published in: Journal of magnetic resonance (1997) 2011-03, Vol.209 (1), p.83-87
• Main Authors: Holland, D.J., Blake, A., Tayler, A.B., Sederman, A.J., Gladden, L.F.
• Format: Article
• Language: English
• Subjects: Algorithms; Bayes Theorem; Bayesian analysis; Bubble size distributions; Bubbles; Gases - analysis; Image acquisition; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Imaging; Magnetic resonance; Magnetic Resonance Imaging - methods; Microspheres; Multi-phase flows; Optical measurement; Pattern Recognition, Automated - methods; Reproducibility of Results; Rheology - methods; Sensitivity and Specificity; Size distribution; Sparse k-space
• ISSN: 1090-7807; 1096-0856
• DOI: 10.1016/j.jmr.2010.12.003

Photograph of a bubble swarm and comparison of the corresponding magnetic resonance and optical measurements of the bubble size distribution. [Display omitted] ► Bayesian approach developed to characterise image features directly in k-Space. ► Enables measurements of transient systems that could not be studied conventionally. ► Approach validated by numerical simulations. ► Demonstrated experimentally on bubble sizing in a gas–liquid flow. ► Applicable to low magnetic field and poor signal-to-noise ratio systems. Magnetic Resonance (MR) imaging is difficult to apply to multi-phase flows due to both the inherently short T 2 ∗ characterising such systems and the relatively long time taken to acquire the data. We develop a Bayesian MR approach for analysing data in k-space that eliminates the need for image acquisition, thereby significantly extending the range of systems that can be studied. We demonstrate the technique by measuring bubble size distributions in gas–liquid flows. The MR approach is compared with an optical technique at a low gas fraction (∼2%), before being applied to a system where the gas fraction is too high for optical measurements (∼15%).