Particle Image Velocimetry (PIV)


Modern PIV calculates the velocity field in a plane by comparing two images containing particles separated by a very short time interval. The flow field of interest is usually illuminated by a thin laser sheet at these two times, with each laser pulse triggering the capture of the particle field using a cross-correlation CCD camera that is placed normal to the plane of the laser sheet. This type of CCD camera is capable of capturing two images in very quick succession (usually 10μs apart). A cross-correlation algorithm is then used to locate intensity peaks in both images for small areas (of the order of 64x64 pixels) and this is repeated across the whole image. The peak in the correlation should correspond to the same particle in both images. The distance each particle moves in terms of pixels (and physical space via a calibration) in the time separating the two images can then be calculated yielding the velocity vector in the plane. Extension of the technique to measuring the out-of-plane velocity component, and therefore the full velocity vector is possible by utilising perspective error. If the CCD camera is placed at 90 degrees to the laser sheet, the angle of view at the edges of the image will decrease from the optimal 90 degrees at the centre of the image, and the motion of particles perpendicular to the laser sheet can be detected. Stereoscopic or 3D PIV requires two cross-correlation CCD cameras placed at approximately 40 degrees to the laser sheet. This obviously causes a problem with focus as different parts of the laser sheet are at different distances from the camera lens. This can be overcome by using a large depth of field which results in a small lens aperture and low light intensity at the CCD. The Scheimpflug condition for focusing solves this problem. It states that if the image plane if the CCD sensor and the lens plane are coincident at the focus plane of the lens, then the particles in the laser sheet will be in focus on the CCD sensor.

The temporal resolution of PIV is governed by the repetition rate of the laser and the CCD camera. In general these are usually 15-30Hz, although high speed PIV at frequencies up to 1MHz is possible using cinematic cameras and high frequency lasers and multiple CCD cameras. Spatial resolution is also limited by the CCD element and the number of pixels and their size. An example of a high end CCD camera is one with 1280x1024 pixels and a pixel length of 6.7μm.

We will determine the pressure distribution, the size of leading edge bubbles, the trailing edge separation points and the skin friction in all tests. The pressure distribution will be measured using Pressure Sensitive Paint (PSP). Skin friction will be measured using oil film interferometry or shear sensitive liquid crystals. Particle Image Velocimetry (PIV) will be performed on a two upwind and two downwind designs to create a detailed picture of the flow past the sail and in the wake. Expertise in all of the measurement techniques described is available at NASA Ames.

We will return to the 7x10 ft windtunnel in late 2003 to test sail shapes for the Maltese Falcon suggested by our optimization algorithm.






Home | Contact | People | Projects | Gallery | Publications | Links | Stanford University