LAUSR

Abstract Although the characterization and identification of solid particle-wall collision behaviours within oil-conveying flows are important, these issues, which are closely associated with the optimization of particulate systems in the chemical and petroleum industries, are rarely discussed in the literature. In this work, a series of multi-scale vibration analysis methods were proposed for the quantitative characterization and identification of sand/glass features in an oil-conveying pipe flow. Dilute particles were identified within an oil-conveying flow in the 2-D time-frequency plane and then verified by the coherence method with dual sensors. The particle-wall collision features were analysed at multiple resolutions by empirical/variational mode decomposition (EMD/VMD) in combination with statistics for each intrinsic mode function (IMF) component. Corresponding verification experiments were performed, and good agreement was found between the test conditions (oil viscosity of 10-50 mPa·s, velocity of 1.8-3.0 m/s, and particle size of 160-270 μm) and multi-scale vibration energy (SE) responses at multiple bandwidths (the macro-scale bandwidths of 30-50 kHz and 40-50 kHz, meso-scale bandwidth of 45-50 kHz, and micro-scale bandwidth of 40-42 kHz). Particularly, a solid content model was established with variations in the oil viscosity h(μ), flow velocity G(v) and particle size p(s), expressed as S o l i d content = A  · p s  · [ G ( v )  · S E ] B h μ , and the model was verified for each scale. Moreover, considering the signal-to-noise ratio (SNR) performance, the best particle characterization scales were the meso-scale and micro-scale, with good error rates of 4.92 % and 5.10 %. Therefore, this study complements existing particle characterization methods and deepens the current understanding of particle-wall collisions in multi-phase flows.