Wall Wetting Characterization Using an Image-Based Scattering Technique 2004-01-1953

A simple imaging technique was explored as a means for characterizing in-cylinder wall wetting in GDI engines. For technique development, a GDI fuel injector was directed vertically down on the top of a temperature controlled flat piston within a non-motored research cylinder in an experimental arrangement described previously [3, 4, 6, 7, 14]. A three-factor randomized factorial design of experiments was performed that included laser sheet level (with three treatments including 0 mm, 2 mm, and 5 mm from the piston surface), piston surface temperature (with three treatments including 100 °C, 150 °C, and 200 °C), and time after start of fuel (with five treatments including 1 ms, 2 ms, 3 ms, 4 ms, and 5 ms after start of fuel). The technique for characterizing wall-wetting differences involved subtracting the 8-bit pixel intensity values at every pixel location for one laser-illuminated scattered image from another image. This technique is similar to that described in Wicker and Eaton (2001) used to identify particle number density differences in scattered particle images. In the current technique, when individual pixel intensity values of the first image were greater than the pixel values in the second, that pixel was assigned a value for gray, and when the opposite was true (the second image was greater than the first), the pixel was assigned a value for black. A user-defined threshold was also included that required the absolute value of the pixel intensity differences to be greater than the specified threshold value; otherwise those pixels were assigned the image background intensity value (white). Multi-factor analysis of variance on the resulting number of black and gray pixels obtained in image subtractions determined that each factor (laser sheet level, time after start of fuel, and piston surface temperature) had statistically significant effects on the number of black and gray pixels with P-values for each factor less than 0.05. Fisher's least significant difference procedure was used to determine statistically significant differences between the means of the populations. Based on these results, it appears that this technique shows promise in capturing differences in wall wetting for the quite complex impingement and evaporation environment within GDI engines.