| dc.description.abstract | Multiple jets appear in many engineering applications such as engines of launch vehicles
and aircraft, fuel injectors and other mixing devices. Twin jets are models to understand
the differences from single jets that can aid the design of multiple jet configurations.
This thesis is a study of two canonical configurations, round and plane twin jets.
Flow are subsonic and turbulent. The study investigates the flow-field and interactions
of the twin jets with different flow parameters. The investigation was carried out
using a computational tool, specifically large eddy simulation (LES), which has been
demonstrated to be sufficient/adequate to capture the development of such flows and
scaling that depend on large scale dynamics. The LES employs explicit filtering as its
subgrid scale model. Although there is a lack of authoritative and comprehensive studies
of such flow interactions, especially subsonic twin round jets, in the existing literature,
suitable experiments exist for twin round jets and twin plane jets for comparisons.
LES of seven different flow configurations; four twin round jets and three twin
plane jets were carried out. The jets emerge from round or slot nozzles normal to a wall
into the ambient. The new parameter is S/d, which is the initial separation between
the two jets 𝑆 scaled with d, the initial diameter (round jets), or width (plane jets). In
addition, LES of single round and plane jets were also performed for comparison. LES
was employed to compute the unsteady flow-field evolution of the interactions between
the jets.
For both configurations, stationary turbulent flows were obtained, from which
statistics were collected. Profiles of velocity and fluctuations, pressure, spectra and
correlations were examined. Close quantitative agreement with all experiments were
examined. In both flows, after merging into a single jet, profiles collapse on local
scales (configuration centerline velocity and half widths). A new finding is that both
configurations exhibit intrinsic scaling. For twin round jets, the mean streamwise
velocity along the configuration centerline rises from zero at the inflow plane (wall) to
a maximum 𝑈𝑝𝑒𝑎𝑘 where the jets merge ¹𝑋𝑝𝑒𝑎𝑘 º and then decays. For all separations
S/d, configuration centerline velocity scaled with𝑈𝑝𝑒𝑎𝑘 , and streamwise distance scaled
with 𝑋𝑝𝑒𝑎𝑘 collapse to a single curve. Moreover, there are simple relations connecting
these two intrinsic scales with the input scales (jet velocity and separation at inflow).
While there is nearly a similar behavior for plane jets, the collapse is not as definite.
The difference arises from the development of the converging region. Twin round jets
develop downstream with little bending and gradually merge. Twin plane jets do bend
towards each other, increasingly with initial separation. For small separation the jets are
transitional at merger, but turbulent at large separation. This has other consequences
too: the spreading rate of the merged plane jet increases with separation, as observed in
the early experiments. All these similarities and differences, which were incompletely
understood before, have been obtained from the LES.
The first LES corresponds to a subsonic twin round jets configuration with a single
nozzle separation distance. The flow parameters are Reynolds number 𝑅𝑒 = 230 000,
nozzle diameter 𝑑 = 0 07m and 𝑆 𝑑 = 5, where 𝑆 𝑑 is the scaled distance between
the centers of the two nozzles (Okamoto et al. 1985). This would make it the first LES
simulation of twin round jet undertaken at very high Reynolds number. The mean flow
field quantities are accurately captured by the current LES approach. The predicted
spreading growth rate was in good agreement with the available experimental data. In
addition, the simulation has revealed the turbulence intensity profiles collapse when
scaled with the local maximum velocity indicating self-preservation beyond 𝑥 𝑑 = 20,
where 𝑥 𝑑 is the measure of downstream distance scaled with the jet diameter. A
cross-sectional velocity contour plots of the simulation also showed axis switching.
The next part of the current study is concerned with understanding the influence
of varying the separation distances between the nozzles using numerical approach and
glean further insight from the simulations. Three LES are carried out to investigate
the variation in nozzle separation distance. Here the configuration and flow parameters
are similar to the experimental investigation reported in Harima et al. (2005). The
Reynolds number is 𝑅𝑒 = 25 000 and the three nozzle separation distances considered
are 𝑆 𝑑 = 2 4 and 8. One important contribution of this study is also revealed
here. That is, the streamwise evolution of the mean streamwise velocity along the
configuration centerline, 𝑈𝑐, for all the nozzle separation distances 𝑆 𝑑 = 2 4 and 8
was discovered to collapse to a single curve. This behavior is also discovered to be valid
for the LES of the twin round jet with nozzle separation distance 𝑆 𝑑 = 5 discussed
above. Hence, this scaling seems to hold in the case of twin round jets irrespective of
the differences in the Reynolds number and nozzle separation distances. In addition,
scaling 𝑈𝑐 with its peak/maximum value 𝑈𝑝𝑒𝑎𝑘 and scaling the downstream distance
with the nozzle separation distance 𝑆 revealed a linear inverse relationship between
𝑈𝑝𝑒𝑎𝑘 𝑈𝑐 and 𝑥 𝑆 in all the nozzle separation distances considered in the current study
indicating the presence of a virtual jet with 𝑈𝑝𝑒𝑎𝑘 and 𝑆 as its jet exit velocity and
jet width respectively which is analogous to a single round jet inverse centerline mean
velocity relation with its downstream distance. To the best knowledge of the author this
scaling is the first to be reported for twin round jets with different nozzle separation
distance and Reynolds numbers.
Overall the twin round jets at 𝑅𝑒 = 230 000 and 𝑅𝑒 = 25 000 have an order
of magnitude difference in Reynolds number however the corresponding LES showed
the flow development in both cases to be similar. In both cases, self-similar velocity
profiles are obtained in the combined regions. Also, the variation of the velocity along
the centerline between the jets have similar curves. The behavior of the intensity profiles
in both cases is observed to be same. Hence, features of turbulent twin jets are already
established at 𝑅𝑒 = 25 000.
The last part of the current research work deals with twin parallel plane jets at
relatively larger nozzle separation distances, which are rarely studied. Plane twin jets
have features that are not found in round twin jets. First, they enclose a recirculating
region. Secondly, the spreading rate of the combined jet increases with nozzle separation
and then saturates. Three LES of twin plane jets were conducted. The Reynolds
number was 𝑅𝑒 = 8750, and the nozzle separation distances were 𝑆 𝑑 = 4 8 5 and 12.
The numerical study was carried out to investigate the flow evolution of the interaction
in particular the influence of the nozzle separation distance on the growth rate of the
combined flow regions given that the physics of slot jet interactions differs from that of
twin circular jets due the presence of pocket of confined fluid between the slots.
In all the cases, the inner shear layers of the two planar jets merge albeit at different
downstream distances. The simulation showed the merge points of the two inner shear
layers and the location of the combined point are influenced by the variation of nozzle
spacing. At far downstream stations in the combined flow regions, self-similarity
of the velocity profiles is observed similar to the one obtained in single plane jet.
The numerical study also showed the value of the maximum mean velocity and its
location along the centerline to be influenced by the nozzle separation distance, that is
with increasing nozzle separation distance the actual value for the maximum velocity
decreases and its location moves further downstream.
Another contribution of this simulation is giving a new insight on the variation
in the relationship between the growth rate of the combined flow region, expressed in
terms of jet half-width, and the nozzle separation distances. In all cases the jet halfwidth
grows linearly with downstream distance 𝑥 𝑑 similar to a single plane jet however
its value and spread angle varies for larger nozzle separation distance (𝑆 𝑑 = 8 5 12)
but not much for the small nozzle separation distances. The numerical study showed
different flow structure of the jets before merging. That is, for 𝑆 𝑑 = 4 the simulation
indicated that the jet break up is barely completed before merging. At 𝑆 𝑑 = 8 5,
a more turbulent jet reaches the merge point and at 𝑆 𝑑 = 12 the jet breakdown is
completed further upstream of the merge point and fully turbulent jets merge. Hence,
the increased lateral fluctuations at merge are probably responsible for forcing a lateral
movement of the merged jet resulting in greater spreading. This phenomenon at merger
exists for twin plane jets but not twin round jets.
In general, the study showed detailed flow development and interaction of subsonic
turbulent twin jets. The mean and turbulent intensity characteristics of the interactions
were quantified. Distinct differences in the flow features of twin round and plane jets
close to the nozzle exit were observed. In the case of twin round jets, a new scaling of
the mean centerline velocity was showed to prevail for different Reynolds numbers. In
twin plane jets the spreading of the combined region was shown to increase with nozzle
separation distances. This study also further reinforces the robustness and capability
of the current numerical algorithm by investigating the flow interaction between twin
turbulent round and plane jets and providing excellent solutions as well. | en_US |
