Skip to main navigation Skip to search Skip to main content

A three-dimensional tomographic particle image velocimetry approach for cross-interface imaging in confined space based on reverse ray-tracing

  • Chen Ling
  • , Zhiyong Shi
  • , Yue Wu*
  • , Wentao Zheng
  • , Qin Wu
  • , Biao Huang
  • *Corresponding author for this work
  • Beijing Institute of Technology
  • Shanxi Diesel Engine Industry Co.
  • China State Shipbuilding Corporation

Research output: Contribution to journal › Article › peer-review

Abstract

Tomographic particle image velocimetry (Tomo-PIV) is a powerful technique for measuring three-dimensional, three-component velocity fields in complex flows. However, in confined spaces with refractive interfaces (e.g., optical engines), traditional Tomo-PIV faces challenges due to light path deviations. This study proposes a cross-interface tomographic PIV (CI Tomo-PIV) framework to address this. The core innovation is the integration of a physically accurate reverse ray-tracing model with spherical voxels into the reconstruction kernel. Unlike traditional Tomo-PIV models that assume straight-line projection, this method enforces Snell's law at the voxel level, correcting geometric distortions during projection. Benchmarking against polynomial models shows that CI reduces point spread function computation time by ∼70% while maintaining comparable accuracy. Numerical validation with synthetic toroidal vortices and Johns Hopkins turbulence databases turbulent flows demonstrates its accuracy advantages over the uncorrected model. CI Tomo-PIV consistently yields high-fidelity displacement fields with cross correlation coefficients stably within 0.85–0.9, surpassing the uncorrected model (∼0.5). Quantitative error analysis shows that under ideal conditions, the root mean square (RMS) error is below 10%, with bias within 2.5% and random uncertainty ∼9.7%. Even with realistic noise and calibration residuals, RMS error remains 13%–17%. A proof-of-concept experiment in an optical engine cylinder shows that the reconstructed fields resolve the evolution of swirling structures, capturing plausible topological features such as linear velocity decay toward the vortex core. This confirms the need to integrate CI models into reconstruction, extending high-precision PIV beyond the uncorrected model to confined flows with curved refractive boundaries.

Original languageEnglish
Article number095150
JournalPhysics of Fluids
Volume38
Issue number9
DOIs
Publication statusPublished - 1 Sept 2026
Externally publishedYes

Fingerprint

Dive into the research topics of 'A three-dimensional tomographic particle image velocimetry approach for cross-interface imaging in confined space based on reverse ray-tracing'. Together they form a unique fingerprint.

Cite this