Usage of body-fitted windows in PIV image processing

Flow close to the boundaries of bodies is often difficult to measure using particle image velocimetry (PIV); factors such as glare, velocity gradients and the body itself all present challenges in obtaining good quality results. One common problem in conventional PIV algorithms is that they are based on square grids, which for most applications is not aligned with the shape of the body. This means the body will clip part of the window resulting in fewer particles and erroneously placed vectors. Being able to align the interrogation window shape to the body boundary would remove these sources of error. This study investigates the effect that applying non-square windows has, compared to conventional square windows for three test cases: a free-field, a circular body and an airfoil shape. For each of these a number of interrogation methods are tested. For the circular body, conventional and rotated square windows were tested along with a body fitted mesh. The image was also deformed into R-θ coordinates to enable conventional square window processing, before de-warping the vector field. Square, rotated square and body fitted techniques were also tested on an airfoil shape. It was found that the body-fitted and warping methods both showed significant improvements in the boundary layer over the square meshes; although the warping process added computational expense. The meshing technique was found to have little impact on the free field as there was no body or velocity gradient. Utilising this interrogation method in conjunction with developed methods of automated edge detection and more mature processing algorithms will result in better measurements close to bodies than is possible with conventional square windows. An example application of this technique on flow around a sphere is also demonstrated.