Visualization and Image Processing of Compressible Flow in a Supersonic Gaseous Ejector

M.V. Srisha Rao, G. Jagadeesh


A supersonic ejector is an aerodynamic device used as a
compressor/vacuum generator/gas mixer, simple in construction, having varied applications. It is, however, complex in its fl ow structure. Flow visualization techniques are essential to understand fl ow dynamics in general, and with modern image processing techniques their utility has been enhanced to evaluate fi eld data of the fl ow. To understand the gas dynamics in an ejector multiple visualization tools that compliment each other have been implemented including time resolved schlieren and laser scattering from acetone particles seeded in the primary fl ow. While schlieren emphasizes fl ow features by density gradient created due to the fl ow, the need to clearly distinguish the primary and secondary fl ow from one another necessitates the use of seeding and laser scattering. The primary objective is to use two fl ow visualization techniques to clearly capture the fl ow within a supersonic ejector followed by analysis of the images by digital image processing tools to compute the non-mixed length. Following a brief review of fl ow visualizations and the supersonic ejector, experimental techniques and details of the visualization procedures are described in the paper. Instantaneous fl ow images at two operating conditions of primary stagnation pressure at 9.69 bar and 5.69 bar are presented, and their qualitative inferences are described. The image processing tools of MATLAB have been used to develop algorithms that enhance the quality of raw images and through the defi nition of appropriate criteria by using intensity profi le along the height of the ejector, quantitative inferences on non-mixed length is made from both the schlieren and laser scattering images. For the case with primary stagnation pressure of 9.69 bar the length of non-mixed region is evaluated as 4.65 L/D from the laser scattering and 4.95 L/D from schlieren techniques. Given the sources of errors and uncertainity of 6% both the methods of visualizations give results in agreement with each other.

Full Text: PDF


  • There are currently no refbacks.