Abstract
We perform high-resolution stereo P.I.V. at 15 Hz in a cross-stream plane in the lower half of a smooth-bed open channel flow at $Re_\tau=u_\tau h/v=300$. This paper describes technical details of our system, concentrating on questions of calibration and validation. We demonstrate theoretically and experimentally that the pinhole model may be applied without modification to a Scheimpflug camera, and present a calibration target designed for Willert's “front-and-rear” stereo camera arrangement. A cross-check on the two horizontal components of instantaneous velocity vectors, using a third camera viewing from below with beam illumination (rather than sheet illumination), gives good agreement with results from stereo P.I.V.
Introduction
Stereo Particle Image Velocimetry (PIV) is a cornerstone technique for capturing three-dimensional velocity components in fluid dynamics. However, its effectiveness is often limited by the complexity of calibration. Traditional methods typically rely on planar calibration grids that must be translated with extreme precision—often within $10^{-4}$ radians—to avoid significant reconstruction errors. In the context of open-channel flow, where capturing small-scale structures like streamwise vortices requires high spatial resolution, these calibration challenges are particularly acute.
This study presents a robust framework for stereo PIV calibration using a "front-rear" camera arrangement. By leveraging geometric symmetry and a novel 3D calibration target, we simplify the experimental setup while maintaining high precision in velocity measurements.
Experimental Methodology
The experiments were conducted in a 7.5m long, 50cm wide smooth-bed open channel. The flow was seeded with 10-15 $\mu$m hollow glass spheres and illuminated by a double Nd:YAG laser (~70 mJ/pulse) forming a cross-stream laser sheet. Images were captured using high-resolution Kodak ES1.0 cameras (1K x 1K pixels) at 30 f.p.s. (15 Hz pulse pairs).
Front-Rear Camera Arrangement
Unlike standard setups where cameras are placed on the same side of the laser sheet, this system utilizes a front-rear configuration. This arrangement offers two primary advantages:
- Enhanced Illumination: It maximizes the capture of forward-scattered light, leading to brighter and more balanced particle images.
- Symmetric Distortion: Image distortion becomes symmetric across the views, which facilitates the cancellation of certain registration errors during the reconstruction process.
3D "Finger" Calibration Target
To bypass the difficulties of precise grid translation, we developed a stationary 3D calibration target. The target consists of five vertical "fingers" machined to a precision of +/- 5 $\mu$m. Each finger contains reference points at known locations. The fingers are beveled, allowing one camera to "see through" the gaps to the points on the fingers behind. This design enables a complete volume calibration from a single pair of images, significantly reducing setup time and human error.
Theoretical Insights: The Pinhole Model and Scheimpflug Cameras
A common concern in stereo PIV is whether the standard pinhole camera model remains valid when cameras are tilted to satisfy the Scheimpflug condition (tilting the lens to maintain focus across an angled light sheet). We demonstrate that the 11-constant pinhole mapping function works without modification for Scheimpflug cameras. While tilting the lens shifts the plane of focus, it does not alter the fundamental geometric projection of the image centroid. Consequently, the mapping remains linear in projective space.
Results and Validation
The accuracy of the calibration was verified by comparing known physical distances between target reference points (12 mm) with the distances reconstructed by the PIV software. The results showed exceptional precision, with a standard deviation of error of 0.00139 mm in the vertical (z) direction and 0.00362 mm in the spanwise (x) direction.
Independent validation was performed using a third camera positioned beneath the glass-bottomed channel. This camera provided a 2D PIV check on the horizontal velocity components. The stereo PIV results showed good agreement with this independent check, with overall velocity profiles for streamwise and spanwise components matching well, despite minor local discrepancies of up to 0.5 mm/sec.
Conclusion and Future Work
The combination of a front-rear camera arrangement and a precision 3D target provides a reliable and efficient method for stereo PIV calibration in open-channel flows. While the current setup is highly effective for small fields of view (approx. 5cm x 4.5cm), future work will focus on scaling the 3D target for larger domains and developing theoretical models to better account for anisotropic refraction at the air-water interface.
Summary updated