pared to a conventional nozzle, at least in the early Pickett and Siebers [1,2] showed that the combustion phase. A possible reason for the beneamount of soot decreases with decreasing orifice fit in soot formation of the cluster nozzles is the diameter because of higher air entrainment. Ma- higher difference between the penetration length of tsumoto et al. [3] investigated the effects of re- the liquid phase and the lift-off length, so this is a duced nozzle hole diameters for multihole nozzles. serious motivation to measure both quantities siThey reported that smaller nozzle hole diameters multaneously in this study. cause a slightly shorter spray tip penetration, but For the investigated nozzles each hole of a produce leaner and more homogeneous fuel-air conventional nozzle is replaced with two smaller mixture. The effect of smaller droplets and en- holes in order to reduce soot formation. The dihanced evaporation, observed using nozzles with ameter of the smaller holes is chosen so that the equally spaced micro-orifices, can be realized us- flow rate of all nozzles should be equal. The basic ing clustered nozzle holes [4]. A basic idea of the strategy of these cluster nozzles is to provide a Cluster Configuration (CC) nozzles is to prevent a better primary break up and therefore a better mixfuel rich area in the center of the flame where most ture formation caused by the smaller nozzle holes, of the soot is produced, and to minimize the overall while keeping the penetration length of the vapor soot formation in this way. Adomeit et al. [5] inves- phase comparable to the base line of the conventigated divergent and convergent Group Hole noz- tional nozzle. zles compared to a baseline nozzle in a Diesel engine. They found out that Group Hole nozzles Experimental setup can significantly reduce the emissions. The con- Pressurized Chamber - The ambient conditions vergent nozzle showed advantages over the diver- in the pressurized vessel for the investigation have gent one, especially at higher part load. On the been set at a temperature of 800 K and a pressure other hand, in [6] it is shown that convergent noz- of 50 bar. zle holes can produce larger SMD and higher liq- Investigated Nozzles - Table 1 documents the uid phase fraction in the region of the spray tip. nozzles that have been investigated in detail. For Gao et al. [7] compared Group Hole nozzles with comparability all nozzles have identical flow numstandard nozzles using measurements in an opti- bers of 210, additionally a nozzle with half of the cal Diesel engine without EGR. For this application flow rate as a second reference was used. The a divergent Group Hole nozzle with a 10° angle cluster nozzle configuration consists of 6 orifices was found to be best. They pointed out that larger distributed in 3 groups. The included angle bedivergent angles enhance the evaporation, but the tween the sprays is positive causing the sprays to penetration of the liquid and vapor phase is signifi- diverge. The nozzles have been investigated using cantly reduced caused by smaller air entrainment. an energizing time of 850 μs for investigated rail Pawlowski et al. [8] presented a detailed inves- pressures of 600 bar, 1100 bar and 1600 bar. The tigation of cluster nozzles in a combustion vessel nozzles are installed on a state-of-the-art Bosch using visualization techniques and PDA. It has piezo injector. All measurements are conducted for been shown that the liquid phase from cluster noz- a wide range of taei (time after energizing the inzles penetrates significantly slower than a spray jector). from a conventional nozzle. Also a strong interaction between the two clustered sprays was found. This effect diminishes for larger cluster angles. The combustion and soot formation of the same cluster nozzles is investigated in [9] under comparable conditions. There it is demonstrated that the used cluster nozzles generate less total soot mass com-
3 3 6 6 6 6 6 6
Angle 0° 2.5° 5° 7.5° 10° 15° Number of OCpluesntienrg Designation Flow Rate Orifices
Visualization - Mie Scattering, using a light from a defocused laser light sheet has been employed to detect the envelope of the liquid phase. Quasisimultaneously a modified Schlieren technique has been used to detect the envelope of the vapor phase. A CCD camera from LaVision GmbH is used as dectector. Typically, 20 images per time step are recorded and the penetration length is extracted from the images. The spray propagation is oriented upwards.
exposure time of the camera is set to 100 μs and the delay of the two frames is 50 μs.
The penetration lengths of the liquid and the vapor phase (Fig. 1) as well as the lift-off length (Fig. 2) and the ignition delay are determined. The ignition delay t is defined as the taei when the ignition probability has reached 50%. Lift-off length and ignition delay are not determined in the case of the 7.5° CC nozzle, because the concerning measurements were strongly affected by unwanted temperature variations in the vessel. The influence of the cluster configuration and the cluster angle on the penetration and the combustion are analyzed and discussed.
Penetration length In the case of the cluster nozzles the penetration length depends strongly on the opening angle between two sprays. The influence of this parameter decreases with increasing of the opening angle.
However this behaviour does not present a linear function of the opening angle. Therefore, the following classification can be made: Group 1: Conventional nozzle behavior - (refer10 mm ence nozzle 2, cluster nozzles 0° and 2.5°)
The 0° nozzle and the 2.5° nozzle yield similar results compared to the reference nozzle with identical flow number. Slightly decreasing penetra
Nozzle Tip Position tion of the liquid phase for the 2.5° nozzle related Fig. 1: Sample images of the sprays from a to the Reference nozzle 2 was detected, however 1m5il°niqcaultuiidosntpehorafnspoeeznzaelnetdr,ap(tRriaoiilng=hl et6)n0Sg0tchbh,ali(erL,reetafnte)fiMo=rie1v-0aS0pc0oartµtpesh.riaDnsgeetf.eorr- iTnhtehseegtawsopchluassteernnoomzzalejosr rdeifpfreerseennctesanwienrteerfeosutnindg. option to replace conventional nozzles in terms of
OH* measurements set-up - For the measure- penetration length. ment of the OH* chemiluminescence a double- Group 2: Transition behavior - (cluster nozzles frame ICCD camera is positioned in front of the 5° and 7.5°) chamber, next to the CCD camera for the The 5° nozzle and the 7.5° nozzle show different Mie/Schlieren images. The light passing through tendencies of penetration between both phases. In the quartz window is divided into a VIS and a UV the liquid phase these two cluster nozzles prepart. The VIS part is transmitted through the dich- sented higher values of penetration length comroic beam splitter and collected by the lens of the pare to the reference nozzle with half of the flow CCD camera for the Mie/Schlieren images. The UV number. However, in the vapor phase the same part is reflected and imaged by an UV-lens (B. range of penetration from both cluster nozzles Halle, f = 100 mm, f# = 2). Additionally, a filter compared to the reference nozzle with half of the combination (BP313, MSO Jena and DUG11X, flow number was detected. The reason for this Schott) with a maximum transmission around behavior is most likely due to transition between 313 nm is used to isolate the OH* chemilumines- the behavior of one hole from a conventional nozcence in the range of 290 nm – 325 nm from the zle and two smaller holes corresponding to a clusemissions of the soot and other species [9]. The ter nozzle.
Group 3: Low interaction behavior - (reference nozzle 1, cluster nozzles 10° and 15°) The 10° cluster nozzle and the 5° cluster nozzle presented similar results compared to the reference nozzle 1. The 10° opening angle corresponds to the largest angle where the merging of the clustered sprays was visually determinated. The liquid phase penetration of the 10° cluster nozzle spray is slightly higher than one of the reference nozzle.
However, the vapor phase penetration is clearly inferior. The penetration length from the 15° cluster nozzle is, for both phases, slightly inferior to the reference nozzle behavior. For this group, the results show that after the separation of the clustered spray this design does not present advantages in terms of penetration length.
The lift-off length (lol), its fluctuations (cov) and the time-resolved ignition probability are determined and discussed for all nozzles (except the 7.5° CC nozzle) and three different injection pressures. The following conclusions can be stated:
As expected, the cluster nozzles produce a broader flame with increasing cluster angle.
However, all of the nozzles essentially generate a
single ensemble-averaged high-temperature
reaction zone.
The quasi-steady lol (around taei = 3ms) of all
nozzles with prail = 600bar is very similar (
28mm), although the hole diameter of reference
nozzle #2 is much bigger compared to the other
nozzles. This unexpected behavior can be
explained by separated ignition spots which are
detected only a few mm downstream of the orifices
of the reference nozzle #2. Accordingly, the
quasi-steady cov of the lol is about twice as high (
27%) as for the other nozzles ( 13%).
The CC nozzles generate a quasi-steady lol of
about 34mm at prail = 1100bar, and about 39mm
at 1600bar, respectively. The lol tends to
increase slightly with decreasing cluster angle.
Both reference nozzles generate larger lols
than the CC nozzles with prail = 1100bar and
1600bar (Ref#1: 38mm and 43mm, respectively;
Ref#2 42mm and 48mm, respectively). The
behavior of reference nozzle #2 compared to all
other nozzles could be expected [
The ignition probability increases from 0% to 100% within a transition time of about 0.4 ms for all nozzles and all prail.
A correlation of t and lol is observed when the ambient temperature is slightly changed as expected. But t also changes and it is generally not correlated to the lol when different nozzles and prail are compared (at 800K). This indicates that t also depends on mixture formation. t of the reference nozzle #1 strongly depends on prail ( t =2.4, 2.2, 2.0 ms for prail = 600, 1100, 1600 bar respectively), whereas t is rather independent of prail for the other nozzles (ref. nozzle #2: t = 2.55, 2.5, 2.55ms for 600, 1100, 1600bar; CC nozzles: t is in the range 2.35 0.15, 2.25 0.1, 2.25 0.5 for 600, 1100, 1600bar, respectively). t of the CC nozzles increases slightly with increasing cluster angle, in particular for lower prail.
There is always a gap x between the
quasisteady penetration length of the liquid phase and
the lol in the case of the reference nozzles
(except for reference nozzle #2 with prail = 600 bar,
due to ignition spots close to the nozzle as
mentioned above). x increases with increasing prail
as expected [
The gap x is always smaller for the CC nozzles than in the case of the reference nozzles with prail = 1100 and 1600 bar. x is even vanishing, i.e., x 0, for the CC nozzles at prail = 600bar. This may lead to higher soot formation by the CC nozzles in comparison to the reference nozzles.
This work was supported by the General Motors Corp. (CRL Aachen) and the Deutsche Forschungsgemeinschaft (Cluster of Excellence “Tailor-made Fuels from Biomass”). [6] Gao, J., Y. Matsumoto and K. Nishida, “Effects of Group-hole nozzle Specifications on Fuel Atomization and Evaporation of Direct Injection Diesel Sprays,” SAE Technical Paper, 2007-01-1889, (2007). [7] Gao, J., Y. Matsumoto, M. Namba and K. Nishida, “Group-Hole Nozzle Effects on Mixture Formation and in-cylinder Combustion Processes in Direct-Injection Diesel Engines,” SAE
Technical Paper, 2007-01-4050, (2007). [8] Pawlowski, A., R. Kneer, A. M. Lippert and S. E.
Parrish, “Investiagtion of the Interaction of Sprays from Clustered Orifices under Ambient Conditions Relevant for Diesel Engines,” SAE Technical Paper, 2008-01-0928, (2008). [9] Hottenbach, P., T. Brands and G. Grünefeld, “An Experimental Investigation of Combustion and Soot Formation of Sprays from Cluster Nozzles for DI Diesel Engines,” SAE Technical Paper, 2009-010855, (2009).