The direct imaging technique in fluid spray diagnosis is a simple approach which can be achieved by the direct photography of a spray jet using a digital camera and a flash light source. The pulse duration of a typical electronic flashgun is around 50 µs, in comparison with a few nanoseconds for a laser pulse. The relatively long illumination time of the flash-‐ based systems restricts their ability of providing information about high speed flows in the time domain. Furthermore, imaging fluid droplets at the microscopic scale requires a high-‐ power highly-‐directional light beam, which is only available in lasers. However, direct imaging techniques have been used in the visualisation of spray jets, and in the extraction of general macroscopic characteristics such as spray cone-‐angle, spray symmetry and jet penetration.
Global spray photography can be achieved in two ways depending on the method of illumination. Light scattering images are produced using front or side illumination. Shadowgraph images are produced when backlighting is used. Although no significant differences in results between the two methods have been reported in literature, shadowgraphs were found to (be able to) distinguish the spray borders from the
surrounding gas much easier [Ochoterena et al. (2010)]. The light scattering technique,
alternatively, can provide statistical information about the distribution of the fluid volume over the spray pattern, due to the connection between volume and scattered light intensity.
4.3.1.Related Work
A qualitative study on diesel spray was presented by Shao and Yan [Shao & Yan (2006, 2009)] using a direct imaging technique. Images were captured by a CCD camera and a flash light source. The images were processed for the extraction of the macroscopic characteristics of the spray jet, including the tip penetration, the near-‐field angle, the far-‐ field angle, and the average spray-‐tip velocity, at an injection pressure of 600-‐1400 bar. Their experiments were made on a common rail fuel injection system in a pressurised non-‐ evaporating environment, discussing the effect of pressure on the spray macroscopic characteristics. A variation in luminous intensity from one image to another was observed, which could lead to errors in extracting the spray contours. The fluctuation in the light intensity is expected when using long exposure periods, as pixels can be saturated at the high density regions, and this becomes less controllable in highly turbulent flows. Shao and Yan results showed that diesel sprays penetrate faster, deeper, and with wider angles at higher pressure values, which agrees with the previous discussion (chapter.2).
The common rail diesel system was also investigated by Seneschal et al., [Seneschal, et
al. (2003)], using direct illumination from a set of halogen lights located around a multi-‐hole diesel injector operating at 800 bar fluid pressure. The shortest imaging time in this case was 0.5 µs which is the minimum exposure time of the employed camera. A similar system
was tested by Hwang et al. [Hwang, et al. (2003)], for Dimethyl Ether (DME)1 sprays. In this
case, the spray pattern was determined by averaging out 30 shadowgraph images for each experimental condition. A threshold of 80% was used for extracting the spray image, although a higher threshold value can be used when the contrast between the background
1 “Dimethyl Ether (DME) is an alternative fuel that provides lower particulate matter (PM) than diesel fuel
and the spray in the foreground is higher. Their results showed that increasing the differential pressure increases the spray angle and tip penetration (formula 2-‐10); further increasing the nozzle size in this case has increased the tip penetration.
The low temporal resolution and the very low number of samples used in each condition can lead to errors in the spray progression estimation, or in any other time related characteristic. Only 6 images per test condition were used by Shao and Yan [Shao & Yan
(2006, 2009)], and 30 images by Hwang et al. [Hwang, et al. (2003)]. The number of samples
is important in the image processing of fluid sprays for reducing the margin of error, but it is limited by the memory size, the computing capacity and the system speed.
Another research on diesel sprays at high fluid pressure1 was conducted by Morgan et
al. [Morgan et al. (2001)] and Kennaird et al. [Kennaird et al. (2002)] of the University of
Brighton (UK). A back-‐lighting technique was applied using an argon flash lamp with a
lighting duration of about 3 μs. Kodak 400 digital camera was used for still imaging. Another
video CCD system (Kodak Ektapro HS Motion Analyzer) was employed for high speed
imaging, with a maximum frame rate of 4500 fps at the full resolution (256 by 256 pixels).
The tip penetration and the cone angle of the sprays were calculated directly from the processed images. The images in the background were removed using a manually defined threshold. The tip of the nozzle and the primary spray were out of the camera field of view, which may lead to an error in angle calculation. The effect of the injection pressure on
penetration has been carried out by Karimi K. et al. [Karimi K.; et al. (2006)] using a Phantom
V7.1 high-‐speed camera. High-‐speed shadowgraphy is a well known approach in spray
progression analysis, investigated by recent studies such as Ochoterena, et al. of Chalmers
University, Sweden [Ochoterena R., et al. (2010) ] and Klein-‐Douwel, et al. of Eindhoven
University of Technology [ Klein-‐Douwel, et al.(2007)].
-‐a-‐ -‐b-‐
Figure 4.4 : a. Raw and binary images of the diesel spray (penetration study) [Morgan et al. (2001)]; b. Setup of a high-‐speed recording with backlight illumination [Kennaird et al.(2002)].
Figure 4.5: Image of diesel spray taken by Klein-‐Douwel, et al. at 1500 bar fluid pressure behind the nozzle. The high-‐speed imaging system has a temporal resolution of 222 µs [Klein-‐Douwel, et al. (2007)].
4.3.2.Discussion
The direct imaging technique is a popular low-‐cost choice for liquid sprays visualisation. Most of the available resources in literature are focused on the diesel grades at high injection pressures (400-‐1600 Bar). Both the scattering and the shadowgraph images can be used in this case. Spray contours can be extracted by binarising the raw images via a suitable threshold. Since producing (thin) light sheets is not possible in the conventional flash light
sources, this method can only provide an overall view of the spray pattern rather than detailed information about the fluid mass distribution in the gaseous phase. The direct imaging can be very useful in the extraction of spray angle, general spray pattern, and spray tip penetration, when a large number of samples are used. However, the temporal analysis of the spray formation process needs a higher temporal resolution and a far shorter illumination period than that of flash lights, as well as the need for a higher optical energy for the droplet size investigation at the microscopic scale (1-‐10 µm), which emphasises the importance of using laser beams in similar tests.