Spray formation from diesel fuel injection through a realistic heavy-duty multi-hole common rail injector is studied in a newly developed high pressure, high temperature cell, using digital high speed shadowgraphy at 4500 frames per second. Care is taken to establish accurate synchronisation between camera and injection system and because of the relatively large exposure time, an effective camera image time is calculated for every frame. Further emphasis is given to determining the actual start of fuel mass injection by comparing (for each injection) a predetermined, rail pressure dependent needle relaxation distance to the actual needle lift signal. The spatiotemporal evolution of the spray is found to reproduce well in general, but often sprays suffer from short-lived, small, laterally moving anomalies, which influence axial motion and the spray cone angle. High speed shadowgraphy allows this to be observed and taken into account. After an overview of methods found in the literature, an algorithm for geometrical analysis is presented, which is based on an extension of a combination of those methods. In this algorithm, a local spray angle theta(i)(x) is determined from lateral cross-sections at 80% of the shadow level in order to encompass most of the spray without being too sensitive to background noise. The macroscopic cone angle theta(cone) is derived from the approximate constancy of theta(i)(x) over a relatively long axial distance. Spray penetration is obtained by lateral integration of the spray shadow. A procedure for accurate correlation of spray growth with time shows that the growth is proportional to t(b) with b = 0.57 +/- 0.02 for a common rail pressure of 150 MPa and a gas density 33 kg/m(3) (N-2 at room temperature). The exact value of b is very sensitive to uncertainties in synchronisation and the start of injection determination. The spray cone angle theta(cone) is not constant, but varies with time during an injection, mainly as a result of spray shape changes. (c) 2006 Elsevier Ltd. All rights reserved.
