UW-Madison | College of Engineering | Search

Particle Image Velocimetry


Particle Image Velocimetry (PIV) provides a spatially resolved measurement of the velocity field at an instant in time. The technique is deceiving in its simplicity. Small particles are added to the flow at a sufficient density to fill the field, and two images, separated by a known time, are acquired. The local velocity is determined by interrogating a small portion of the image to determine the particle displacement between exposures. Using the known time separation and calculated particle displacements, the local velocity can be determined. The process is repeated throughout the image in order to provide the full flow field.

The implementation is slightly more complex. Two lasers, separated in time and formed into a coincident sheet, are used to illuminate the flow field in order to restrict the measurement to a well-defined plane. In digital PIV, a special CCD camera is employed to separately record the two images, referred to as ‘Frames’. A single-camera system would be used to measure the two components of velocity that lie in the plane of the laser light sheet, while a dual camera, or stereoscopic, PIV system would be used to measure all three components of the velocity field. Stereoscopic PIV uses two cameras viewing the same region of the flow from different angles. Similar to the way in which viewing with two eyes allows us to evaluate the depth of objects, using two cameras allows calculation of displacement into and out of the plane of the laser light sheet. This is possible because the light sheet is not truly a plane, but has a typical thickness of around 1 mm. By knowing which particle image originated from the first laser pulse, the absolute direction of the flow is known (without this there is an ambiguity of 180 degrees).

The analysis of the images would be very time consuming if individual particle image pairs had to be manually identified. Instead, a statistical approach is employed. The small interrogation region from the first laser pulse is cross correlated with the image from the second laser pulse. This results in the identification of a “most probable separation distance” between the particles. This process is computationally efficient and requires minimal user input. Tracking of individual particles would be computationally very difficult as well, since individual particles may move in and out of the light sheet between the two laser pulses, and also since the particles look very similar to each other when viewed individually. Interrogation regions contain multiple particles that create a distinct pattern, or fingerprint, which simplifies the displacement calculation and minimizes the effects of particles moving into or out of the light sheet between laser pulses.

Samples from each stage of the image interrogation process are shown below for a single camera, two component PIV measurement. For a stereoscopic measurement of all three velocity components, each camera would capture two separate frames and the third component of velocity is calculated from the respective displacement fields.

1.) A flow seeded with small particles is illuminated by a laser light sheet. Two laser pulses are fired with a short time separation, and the camera captures two separate images of the illuminated particles from each laser pulse.


2.) The particle images are divided into small grid blocks referred to as ‘interrogation regions’. Statistical methods are used to determine the displacement of each interrogation region in going from Frame 1 to Frame 2, yielding a single velocity vector for each interrogation region. The ‘most probable displacement’ is identified by the peak in the cross correlation function.


3.) This process is repeated for each interrogation region, yielding the global velocity field within the imaging area.


Images provided by Tyson Strand at TSI, Inc. (www.tsi.com)