If the particle density is high enough, seeded flows can be used similarly to dyed flows, which works well for boundary marking techniques. But when particle concentrations are low enough to image individual particles, other flow visualization techniques become available: particle pathlines, (a.k.a. particle tracks) and even quantitative methods — particle image velocimetry (PIV) and particle tracking velocimetry (PTV). We will not discuss quantitative techniques here — like many aspects of technology, the AI revolution will create new analysis methods that will make any guidance presented here obsolete within a few years, if not a few months. Perhaps the second edition of this page will discuss PIV and PTV.
Qualitative flow visualization with particle tracking can show flow structures within a flow, not just the boundary between seeded and unseeded areas. Generally particles are uniformly but sparsely distributed in the fluid and imaged over a period of time, recording the particle paths — a type of deliberate motion blur, as shown in Figure 1. Small air bubbles entrained into the flow can be seen swirling around the central vortex.
The particles can be illuminated throughout the volume and imaged onto a 2D sensor, resulting in line-of-sight integrated information, similar to schlieren and shadowgraphy methods. This works well for flows that are approximately two dimensional but can be confusing in three-dimensional flows. For example, in Figure 1, does the apparent vertical motion of the particles represent the real flow, or is it an artifact of perspective as the particles orbit the central vortex?
Typically, researchers only image and present a single slice through a three-dimensional volume. Ideally, the slice is oriented so that the flow stays within the thickness of the slice. One way to achieve this is to limit the particles to a flat interface between two fluids, such as an air-water interface; it’s easy to float pine pollen on a water surface and image its motion. However, surface tension and the Cheerios effect may dominate the motion of floating particles.
More commonly, the slice is achieved by limiting the light to a sheet. Thin (1 – 5 mm thick) sheets of light are easily made by expanding a laser beam through a cylindrical lens. A glass stirring rod can serve as an inexpensive cylindrical lens, although the resulting sheet will not be uniform; alternatively, a decent plano-convex anti-reflective coated lens costs $50 to $150. The laser power required for a sheet of 0.25 m2 is around 0.5 Watts continuous, putting it in the Class III or IV safety category. While not at all expensive — $25 or so — serious precautions must be taken to protect users’ eyes. For larger fields of view, a linear array of LED lights can be used, although the minimum sheet thickness is about a centimeter. The light from the array needs to be guided with ‘barn doors’ to form a non-diverging sheet .
To image particle motion throughout a truly three-dimensional flow, consider alternate imaging technologies: holographic , plenoptic or stereoscopic methods. Of these, the stereoscopic approach is currently the best developed and least expensive. Stereoscopy uses two cameras; one with a point of view from the left, and the other from the right, analogous to human eye perspectives. Stereoscopic drawings actually predate the invention of photography . Stereo photography was fairly popular in the 1950’s and ’60s, and you may have seen 3-D movies where you wear special glasses (red-blue or polarized) that direct the correct image to each eye, or virtual reality headsets that do the same for computer-generated imagery. However, there are no true glasses-free viewing technologies that anyone can use.
A final consideration: particle path images are inherently not time-resolved. The longer the shutter is open, the longer the pathline will be. You can estimate the particle velocity if you know the shutter time and the length of the path in space — another good reason to take a photo of a ruler in your setup, so you know the conversion from inches to pixels.
