You are on page 1of 16

International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer


journal homepage: www.elsevier.com/locate/ijhmt

Upstream penetration behavior of the developed counter flow jet


resulting from multiple jet impingement in the crossflow of cylindrical
duct
E.V. Kartaev ⇑, V.A. Emelkin, M.G. Ktalkherman, S.M. Aulchenko, S.P. Vashenko
Khristianovich Institute of Theoretical and Applied Mechanics, SB RAS, Institutskaya str., 4/1, Novosibirsk 630090, Russia

a r t i c l e i n f o a b s t r a c t

Article history: Experimental and numerical investigations of ultimate upstream penetration of developed counter flow
Received 27 April 2017 jet formed as result of impingement of multiple round jets radially injected into a high-temperature con-
Received in revised form 17 August 2017 fined crossflow have been performed. These investigations aimed to reveal an influence of strength of the
Accepted 27 September 2017
developed counter flow jet on mixing intensity of jets in crossflow of cylindrical duct. Based on analysis of
Available online 17 October 2017
experimental data, the analytical dependence between dimensionless parameter of upstream penetration
depth of counter flow jet and the square root of jets-to-mainstream momentum-flux ratio J has been
Keywords:
established. The dependence proved to consist of linear region as well as nonlinear and asymptotic ones.
Impinging jets
Developed counter flow jet
For a given geometry an ultimate upstream penetration depth of the counter flow jet has been estimated
Confined crossflow to be approximately 2.1–2.3 diameter of the cylindrical duct. The corresponding value obtained on the
Upstream penetration depth base of numerical simulation turned out to be 1.8–2.0 diameter of the cylindrical duct. The dimensionless
Mixing parameter h/D of radial jet penetration depth proved to be an appropriate to describe adequately an
empirical character of upstream penetration of the counter flow jet within the linear region. Based on
results of numerical simulation, axial velocity, temperature and pressure profiles as well as centerline
turbulent kinetic energy contours have been obtained. It has been has also shown that an increase of
the momentum-flux ratio promotes mixing in upstream recirculation flow zone and improves overall
mixing performance as well.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction C as an indicator of the mixing uniformity which is proportional to


the square root of the momentum-flux ratio J related to the num-
Mixing of transverse round multiple jets-in-crossflow (JIC) of ber of jets n. Extensive calculation and experimental works were
the cylindrical duct is of special importance in various technologi- also addressed to the same challenge in [22–24].
cal applications; mixing in the Rich-Burn/Qiuck-Mix/Lean-Burn The complexity of multi-parameter JIC problem governed the
(RQL) staged combustor is among most important. Optimization further theoretical investigations; the target was to find a universal
of mixing implies an almost uniform profile of the preset temper- dimensionless parameter describing the physical character of the
ature and components composition at the mixing zone exit, and dependence of the trajectory of the transverse jets issuing into
interaction between radially injected quenching jets and crossflow the crossflow [25]. Ktalkherman [26–28] proposed the h/D param-
plays the key role here. Comprehensive reviews of the works eter of the radial jet penetration as a value to describe reagent mix-
devoted to JIC, both about a single jet in the crossflow and a ing in fast-mixing chemical reactors and final product quenching;
multi-jet injection are presented in [1–4]. A number of investiga- this parameter was also used in [29].
tions of JIC, both experimental and numerical, have been per- Numerical analyses of JIC conducted by Clayton and Jones [30]
formed within recent decades; they dealt with the effect of the and Ivanova et al. [31] demonstrated a superiority of Large Eddy
geometry of the RQL combustor mixer and jet parameters on their Simulation (LES) model over Reynolds Averaged Navier-Stokes
interaction in reacting and non-reacting flows of the cylindrical (RANS) approach under conditions of multiple jet injection and sin-
duct [5–21]. In this work cycle, Holdeman proposed the parameter gle jet injection, respectively. Nevertheless, Urson et al. [32]
applied RANS-based standard j-e model of turbulence to study
multiple jet injection into crossflow of cylindrical duct and com-
⇑ Corresponding author.
pared numerical results with experimental data of [5]. Recently,
E-mail addresses: kartayev@mail.ru, kartayev@itam.nsc.ru (E.V. Kartaev).

https://doi.org/10.1016/j.ijheatmasstransfer.2017.09.111
0017-9310/Ó 2017 Elsevier Ltd. All rights reserved.
1164 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

Davoudzadeh and Forliti [33] used low Reynolds number j-e (1) carry out numerical and experimental investigations of
model to perform numerical study of flow and mixing properties behavior of the developed counter flow jet formed by
of single transverse jet in a confined system. The results show that impinging round cooling jets transversely injected into
the numerical simulations predict the experimental data with a high-temperature crossflow for given geometry of the cylin-
good degree of accuracy. The reported results of the experimental drical duct;
and numerical simulation of the processes in the mixing region in (2) conduct the analysis of the mixing intensity in the recircula-
RQL combustor geometry with water tests [34–37] are of special tion flow zone upstream of the jet injection plane (JIP) as
interest. Numerous works dealing with the JIC processes of function of J or derived parameters under conditions of
quenching in mixers and fast mixing of components in chemical developed counter flow jet.
reactors prove that this problem is topical [38–45].
Understanding of the JIC processes occurring during the forma- 2. Analogy of CJIM and JIC
tion of the counter flow directed toward a crossflow of the cylindri-
cal duct is very important from the viewpoint of design and Fig. 1 presents two configurations of the counter flow jet in the
optimization of the units of fast-mixing of reagents, or the zones crossflow confined by cylindrical duct walls: (1) as a result of jet
of rapid quenching in chemical reactors. In this case, the conditions emanating from the tube on the channel axis – the counterflowing
of counter flow mixing and fast jet dilution of crossflow could be jet-in-mainstream CJIM (Fig. 1a) [56]; (2) as a result of impinge-
fulfilled. The studies by Kartaev et al. [46,47] have been devoted ment of the jets transversely injected into the channel – the mul-
to the analysis of the gas dynamics of the interacting impinging tiple jets-in-crossflow JIC (Fig. 1b) [35,36]. The same figure shows
jets and crossflow and their mixing under the conditions of counter the flow velocity profiles ux on the channel axis normalized by
jet formation and development. By now, however, a relation the mass-average velocity Um of the mainstream. The positive
between the axial penetration depth of the developed counter jet direction of the x axis coincides with the velocity vector Um direc-
and large values of momentum-flux ratio J is still unclear. At the tion, thus the flow velocity ux on the channel axis is negative in the
same time a lot of experimental data has been accumulated, a recirculation flow zone, RFZ [30], which is shaped as a toroidal
number of numerical calculations have been performed concerning eddy.
counterflowing jet-in-mainstream (CJIM)) interaction in the cylin- It is evident that the areas of recirculation mixing form in both
drical or non-cylindrical duct [48–56]. Here, the jet emanates from cases, though they are of different shapes. In the first case, the exit
a tube on the channel axis towards mainstream flow. Morgan et al. tube section means the jet injection plane with the axial velocity
[49] found the nonlinear relation between counter flow jet pene- Vj; in the second case, it is the plane formed by intersection of ori-
tration depth and jet-to-mainstream velocity ratios (equal to J for fice axes, the jets are injected with the radial velocity Vj (jet injec-
equi-density flows); the ultimate counter jet penetration depth tion plane (JIP)). The directions are presented in Fig. 1. In both
was evaluated in [56]. Li et al. [57] controlled the crystal phase for- cases, the penetration depth lp of the counter jets is deemed to
mation and size fraction of synthesized titania particles by means be the axial distance between the flow stagnation point and JIP.
of the centerline counter flow jet issuing from the tube in the In the JIC case, two points of flow stagnation form on the channel
plasma-chemical reactor. Settumba and Garrick [58] applied cold axis (see Fig. 1b), the 1st stagnation point is located downstream
argon counter jet to control the rate of aluminum vapor condensa- of the JIP at low jets’ flow rate, whereas at higher flow rates, the
tion in the heated flow of argon plasma. Recently, the counter flow stagnation point is almost coincident with the JIP. In the case of
model was used for the axial separation of the counter jets (axial mixing of non-isothermic flows the axial depth hv of cold counter
jet separator) in the accelerator mass spectrometry [59]. jet penetration into the high-temperature crossflow can be evalu-
The objective of this work: ated on the base of the temperature decrease by a certain preset

Fig. 1. Configurations of the counter flow jet in mainstream confined by cylindrical duct walls and respective profiles of centerline dimensionless velocity: (a) – tube
counterflowing jet-in-mainflow (CJIM); (b) – multiple jets-in-crossflow (JIC). Zero axial velocity surfaces (stagnation surfaces) are drawn by dotted lines based on streamline
curvature.
E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178 1165

value which is taken to be 100 K in [46,47]. The calculations gave [26–28] in order to describe to a certain degree the process of JIC
hv  1.1  lp. mixing, though at h/D > 0.5 jets impinge on the axis, hence this
As seen from Fig.1a, on the initial section, the absolute velocity parameter becomes formal. This could be interpreted as following.
equals to Vj, it determines the jet core length at the tube exit; then If the jets had not impinged, each of them would have radially pen-
the velocity module drops to zero in the flow stagnation point. In etrated up to value of h/D in duct diameters for a given value of J.
JIC case (Fig. 1b) the counter jet first accelerates and then is decel- The jet-to-mainstream momentum-flux ratio is defined as
erated by the crossflow. following:
It is seen that the difference in RFZ shapes results above all from
the fact that in the second configuration (JIC), the radially injected
qj V 2j V 2j
J¼ ¼ DR  2 ð6Þ
jets form a basis of RFZ. The maximal area of the ‘‘effective” RFZ qm U m 2
U;
cross section lies in the plane near the JIP. Thus it can be expected
Here qj is the density of jets, qm is the density of the crossflow, DR is
that mixing in the second configuration will be more intensive,
the density ratio. Therefore, the dependence of parameter hv/D of
since, aside from the RFZ ‘‘surface” of mixing (stagnation surface
counter flow jet penetration depth can be written as:
or zero axial velocity surface), a part of the crossflow pushed
rffiffiffiffiffiffiffi 
toward the walls will be ejected by the jets inside of RFZ. As seen, hm K 0 qi d V j d pffiffi
   K0 J; ð7Þ
inside of the region limited by the stagnation ‘‘surface” local axial Cd Cd qm D U m DC d
velocity is co-directional to counter flow jet.
As noted above, there are many experimental, numerical and where K0  0.67 [47].
analytical studies [48–56] devoted to the counter jet penetration Comparing (2) and (7), one can notice their similarity up to a
in CJIM configuration. Morgan et al. [49] (water tests) and Rao constant, taking into account that (7) is formulated for variable-
[48] (incompressible air tests) experimentally analyzed the pene- density flows. Furthermore, in (7) d is the orifice diameter, whereas
tration depth lp of the counter jet in the crossflow with respect in (2) it is the inner diameter of the tube; in (7), Vj is the velocity of
to the parameter: jets’ injection into the channel, in (2) it is the jet velocity at tube
exit on the channel axis.
 2 Hence it can be presumed that parameter h/D of orifice jet radial
d Vj
Z¼  ; ð1Þ penetration depth (based on linear character of relation with hv/D
D Um
according to (5) and (7)) can be used as an appropriate parameter
where d is the inner tube diameter, D is the diameter of the cylin- for description of counter flow jet penetration depth.
drical duct. In both cases, the following dependence was obtained:
  3. JIC mixing parameters
lp d Vj
¼ 2:5  ¼ 2:5  Z 1=2 ; ð2Þ
D D Um
The following parameter is used, in [2,7,26] and other works, for
which has the linear character with respect to Z1/2. Moreover, it was the evaluation of local mixing nonuniformity:
found in [49], that, with further increase of the counter jet hydrody- Tm  T
namic head, this dependence transforms and becomes nonlinear fT ¼ ; ð8Þ
Tm  Tj
one – the one-third power law parameter Z1/2 (0.5 < Z1/2 < 1.5):
where Tm is the crossflow temperature, Tj is jet temperature, T is the
 1=3
lp d Vj temperature at a local point. The case when fT = 0 corresponds to
¼ 1:8  ¼ 1:8  Z 1=6 : ð3Þ
D D Um local temperature equal to Tm, while fT = 1  local temperature
equal to jet temperature Tj. For non-reacting flows, f is the con-
As noted in [49], this dependence was obtained within the served scalar and corresponds to component concentration to be
range D/d from 8.3 to 82 for equi-density flows, it is considered mixed [60,61], and (8) is valid when Tm and Tj are not too different.
to be in generalized form. The numerical study (standard k-e Taking into consideration temperature dependence of fluid specific
model) of the penetration depth of the turbulent counter jet into capacity the more general expression is used at Tm  Tj:
the crossflow of the cylindrical duct is presented in [56], the results
are compared with experimental data [49,52]. It was found that im  i
fi ¼ ; ð9Þ
influence of initial jet turbulence intensity is negligible concerning im  ij
to ultimate counter jet penetration depth into mainstream. The
where i is the gas enthalpy. As shown in [47], the respective values
asymptotic ultimate value of the counter jet penetration depth is
obtained by the formulas (8) and (9) for the temperature below
evaluated based on numerical results and experimental data
1500 K, differ within the range of 4%, thus in this work, the mixing
analysis:
nonuniformity was evaluated on the base of the measured
 
lp temperatures.
¼ 3:57: ð4Þ When crossflow and jets are completely mixed, f is determined
D max
by the ratio of jets’ flow rate Gj to the crossflow flow rate Gm:
No such investigations have been carried out for the JIC config-
urations. In JIC case [46,47], for given mixer geometry, a region of Gj =Gm
f eq ¼ : ð10Þ
the linear relation of the parameter hv/D of axial penetration depth 1 þ Gj =Gm
of the counter flow jet and parameter h/D of the orifice jet radial The mixing nonuniformity in the cylindrical duct cross section
penetration depth was revealed, here h/D is estimated from for- of area A can be found from the expression:
mula below: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

P 2
h d pffiffi f
0 1
A i ai ðf i  f eq Þ
¼K  J; ð5Þ ¼ ; ð11Þ
D DC d f eq f eq

where K = 1.7, D is the cylindrical duct diameter, d is the orifice where fi is the local mixing nonuniformity in the point i, ai is the
P
diameter, Cd is the orifice discharge coefficient. The parameter of area of the surface element near the i-point, here A ¼ i ai . In the
the jet radial penetration depth h/D is proposed by Ktalkherman cylindrical system of coordinates, (11) is reduced to the form [26]:
1166 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R R1 2 turbine burners, where the ratio of dilution jets mass flow rate to
0
1
R2 0
ðf  f eq Þ 2rdr
f 1 the mass flow rate of crossflow is fixed (for example, in [7,19] it
¼ ; ð12Þ was 2.5, and in [6] – 2.2), and variation of J is possible only by vary-
f eq f eq
ing the orifice diameter if their number is given [63]. It was found
where integration is performed within the range from 0 to R1 = 0.9  that the minimal number of the orifices needed to reduce crossflow
R, where R – cylindrical duct radius. This integration limit results passage between the jets without mixing is 6 [8]. That is why the
from the necessity to exclude the effect of the thermal boundary circumferential distance S between the orifices normalized to their
layer as the temperature is considered as the conservative scalar diameter d is an important parameter for the qualitative and fast
[7,26]. mixing:
The spatial unmixedness parameter Us determined as [62]:
P
pD
2 S=d ¼ : ð17Þ
i ai ðf i
 f eq Þ
1
nd
Us ¼ A
: ð13Þ
f eq ð1  f eq Þ The inverse value is referred to as the parameter of blockage [7].
is also used to evaluate the mixing efficiency.
The unmixedness parameter Um have been applied in [25] and 4. JIC experimental setup
other studies as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Fig. 2 shows the schematic of the part of cylindrical duct conju-
u P
u1 ai ðf  f Þ2 pffiffiffiffiffiffi gated with the mixer and the annular air manifold. The facility and
U m ¼ tA i
i eq
¼ Us ð14Þ its operating modes used are the same as in [46,47]. The main flow
f eq ð1  f eq Þ
(nitrogen) passed into the duct from the water-cooled direct-
All these parameters are linked by the formula: current linear-scheme plasma torch with inter-electrode inserts
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (IEI plasma torch), with the rated power [64].
pffiffiffiffiffiffi f
0
f eq Either room-temperature air or nitrogen (Tj = 293 K) was used
Um ¼ Us ¼  : ð15Þ as a cooling gas. It was injected tangentially into annular manifold
f eq 1  f eq
and then supplied uniformly through the n = 8 inserts with drilled
Total mixing is reached when the parameters of (11), (13), and (14) orifices, the insert length t = 4 mm (or 5 mm), orifice diameter d =
are equal to zero. In this work, the mixing quality is estimated using 3 mm (cooling gas direction in the manifold is shown by arrows).
formula (11). The jets are injected perpendicularly to the main flow (crossflow)
As underlined in [2,63] and in a number of earlier works, the of the cylindrical duct of diameter D = 32 mm. JIP is formed by ori-
profiles of jet penetration in the radially-axial and radially- fice axes and is perpendicular to the duct axis (x = 0, the axis x is
tangential planes downstream of JIP are similar at the constant co-directed to the crossflow). The flow rate of the nitrogen cross-
value of the parameter C which is proportional to square root of flow was measured by Camozzi manometers and nozzle flow
momentum-flux ratio J divided by orifice number n (C does not meters, air volume flow rate was measured using SMC PFMB
depend explicitly on orifice diameter d): 7501 flow meters (below 500 L/min), and with manometer and
pffiffiffiffiffi nozzle flow meters – at high flowrates.
p 2J
C¼ : ð16Þ Upstream the JIP, 4 type K thermocouples are located to mea-
n sure the temperature on the duct axis. Two basic versions of ther-
According to [2], at C = 2.5, the uniform mixing of the jets and cross- mocouple positions were applied. In the first one, the junctions of
flow is reached at the minimal distance downstream the JIP. Holde- thermocouples ##1, 2, 3, and 4 have been located at x = 25 (or
man solved the JIC problem of optimization of the radial jets 2), 38, 51, and 64 mm respectively upstream of JIP. In the
penetration into crossflow based on specific RQL geometry of gas second version, the corresponding distances were 62 (or 56),

Fig. 2. Schematic of the part of cylindrical duct with the 1st configuration of type K centerline positioning. The mixing zone, the annular air manifold and the supplementary
section are shown. All sizes in mm.
E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178 1167

Table 1
The principal operating conditions of IEI plasma torch of the experimental tests.

Test# Plasma torch electric Plasma forming nitrogen flow rate Nitrogen enthalpy at plasma torch outlet Mass-average temperature of nitrogen at
power, kW Gm, g/s (l/min) cross-section, kJ/kg plasma torch outlet, K
1 8.8 0.93 (44.6) 7.4  106 5100
2 9.9 1.13 (54.2) 7.0  106 4950
3 10.3 1.60 (76.8) 5.2  106 4100

75, 88, and 101 mm upstream of JIP. The distances were mod- In the RQL burners, the typical ratio of the wall thickness to the
ified by means of the relocatable supplementary insert (see Fig. 2), orifice diameter t/d = 0.1, i.e. it is below 0.25 (thin-wall orifices
in order to extend the range of the measured counter jet penetra- [34]), and normally the values of orifice discharge coefficient Cd
tion depth. The diameters of thermocouple junctions were 0.5–0.8 is within the range 0.55–0.70 [65]. In this work, the orifices were
mm, the diameter of corundum thermocouple case was 3 mm. assumed to be thick-wall (t/d = 1.33 or 1.67), Cd = 0.8 [66,67].
Each successive thermocouple was rotated about the previous As ratio d/D = 3/32 = 0.09, hence, at high jet flow rates, the path
one by 90° to reduce the disturbance introduced by the thermo- passed by the jet in orifice diameters before the impingement
couple into the flow. The readings were registered (repeatedly within the duct is D/2d = 5.3, in [46,47] this value was 3.2. Because
from 2 to 3 times) by the multi-channel temperature meter MIT- of this, in this mixer configuration, more accurate ‘‘focusing” of tur-
12TP. bulent jets is necessary. It has been provided by the thick wall of
In addition, in some experiments the junction of W-Re thermo- orifices; moreover, the gas flow in the manifold in the vicinity of
couple (type A-1) embedded into the section with watch window each orifice was almost co-directed to orifice axis, it enabled to
was placed on the duct axis at the distance of 17 mm upstream ensure radial injection of jets to the channel.
of 4th type K thermocouple. The readings of the tungsten-
rhenium thermocouple were indicated by the millivoltmeter to 5. JIC numerical simulation
control mainstream temperature upstream of the type K thermo-
couples. Thermal radiation of the junction of type A-1 thermocou- 5.1. Computational domain and boundary conditions
ple can be visually inspected using the watch window.
Electric power of the plasma torch, the flow rate of the cross- The interaction between jets and high-temperature crossflow in
flow (plasma forming nitrogen), as well as water flow rate for cool- the mode of formation of the counter flow jet, was simulated using
ing of duct walls were chosen in such a way to provide the ANSYS Fluent 15.0. The geometry of the cylindrical duct and radial-
maximal temperature of the main flow at arbitrary measurement axial symmetry plane which divided the channel in half and passed
point being less than the upper measurable value of the type K through the centers of opposed orifices, the grid density of the
thermocouples (1600 K). The principal operating conditions of IEI computational domain with refinement in the vicinity of orifices
plasma torch as heater of nitrogen main flow are presented in were approximately the same as in [46,47]. Total number of the
Table 1. The temperature distribution along duct centerline indi- unstructured hexagonal cells in the computational domain was
cated by the thermocouples was steady for each experimental test 252865, the minimal cell size near the orifice of 3 mm in diameter
when high-temperature mainstream was not disturbed by jets. and wall thickness of 5 mm is about 0.27 mm, the maximal size is
The systematic inaccuracy of temperature measurement caused 2.5 mm, the cell growth rate is 1.10, minimum orthogonal quality
by radiation of thermocouple junction has been evaluated. Taking is about 0.256.
into account the emission of the corundum body of the thermocou- Analysis of sensitivity of the temperature and velocity profiles
ple, it reached 30–70 K (2–5%) and was neglected in the further to the successive grid refinement reveals that the flow pattern is
analysis. almost unchanged. Incompressible ideal air is considered as the
The measurement procedure was as follows. First, temperature fluid of the main flow and injected jets. Preliminary numerical JIC
distribution on the channel axis upstream of JIP without injection tests in which air specific heat, thermal conductivity and molecular
of jets was measured, and then the temperature distributions were viscosity were set as temperature-dependent values demonstrated
obtained at various jet flow rates. The higher the jet flow rate, the almost identical turbulent flow fields compared with those
stronger is their impingement, and the deeper is the penetration of obtained when air thermal properties were treated as constant.
cold counter flow towards the mainstream high-temperature flow. To reduce computational cost, in further numerical calculations
It is implied that the resulting counter flow jet reaches a location of air thermal properties were kept constant. All computations have
thermocouple junction if the measured temperature decreases been performed in double precision. Mixing nonuniformity of jets
100 K below the value measured without jet injection. In this case (Tj = 293 K) with high-temperature mainstream was evaluated
the counter flow jet axial penetration hv is assumed to be equal to based on calculated temperature fields of the cylindrical duct.
the axial distance from JIP where the thermocouple junction is All orifice and channel walls are assumed to be smooth, and no-
located. The position of the thermocouple junction farthest from slip conditions are applied to them. Uniform temperature and
the JIP (4th type K thermocouple) on the duct axis was selected velocity profiles, as well as uniform turbulence intensities were
so that this thermocouple was not reached by cold counter flow specified at orifice inlets and crossflow entry. Zero-gradient bound-
jet at highest jets’ flow rates, in order to evaluate the ultimate ary conditions were set at the outlet plane. Wall temperature mea-
counter flow jet axial penetration depth hv. sured under conditions of the test#1 [47] was specified on channel
The orifice diameter d = 3 mm instead of 5 mm was chosen walls. The effect of the wall temperature on the core flow of the
since it was required to reach larger values of J at the same jet flow channel is estimated to be ignorable.
rates Gj as in [46,47]. At high jet flow rates, the excess pressure in
the manifold was measured by the manometer. The maximal jet 5.2. Computational approach
flowrate in the experiments was Gj  13.3 g/s (the mass-average
jet velocity through each orifice Vj  200 m/s), the excess pressure Our estimates show that Reynolds number for the crossflow
did not exceed 0.35 atm. The calculation involving gas compress- exceeds 3500, for jets – 35000. Obviously, impinging of jets in
ibility gave approximately the same jet velocity values, thus the crossflow is turbulent. The approach based on the steady-state
gas compressibility effects were omitted in this work. RANS equations [68] was utilized to calculate the turbulent flow
1168 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

Fig. 3. Multijet Annulus/Core-Flow case: (a) - test section geometry and measurement planes and (b) – general view of computational domain.

fields; to do this, the realizable steady-state j-e model with stan- 5.3.1. Comparison with LDV measurements of axial and radial
dard wall functions for an incompressible fluid was chosen [69], components of mean velocity
since the model demonstrates a good convergence for this prob- Spencer [34], Baker and McGuirk [35] carried out water flow
lem. At high values of momentum-flux ratio J, nevertheless, the simulation of jets-in-crossflow flowfield in dilution zone of com-
solution exhibits time-dependent behavior, and unsteady RANS bustor, test section geometry consisting on concentric inner and
approach (URANS) has been applied alongside with RANS. outer cylindrical tubes is given in Fig. 3a. In addition, in [35–37]
For the variable-density flows, the RANS (URANS) equations of CFD mutijet annulus/core-flow calculation incorporating standard
continuity and momentum may be interpreted as the Favre- j-e model has been conducted to compare with the acquired
averaged Navier-Stokes equations; note that the flow velocity pre- experimental data on mean velocity components. Here multiple
sents the mass-averaged value [70]. jets injected from annulus with radial velocity Vj impinge on each
Compared with the other j-e models, realizable j-e model other and mix with mainstream with axial velocity of Um = 0.168
shows the best agreement for the flows with high streamline cur- m/s. As result, upstream of JIP recirculation flow zone forms. Diam-
vature, rotational flows with complicated secondary vortex struc- eter of inner tube D = 94 mm, annulus flow height chosen for tests
tures; it predicts more accurately an expansion of the submerged H = 30 mm, water velocity at annulus inlet Uan = 0.16 m/s, bleed
circular jet [71]. It is stressed in [72] that j-e model poorly predicts velocity Ub was evaluated based on 28% bleed flow ratio. Jets are
the streamline pattern in the reversed flow zones, but gives satis- injected through single row of n = 6 equispaced holes with d = 19
factory results beyond this zone (flow recovery). Our previous JIC mm in diameter. Hence, momentum-flux ratio was constant: J1/2
calculations based on this model [46,47] showed just the qualita- = 4.6. Wall thickness of inner tube was 3 mm.
tive agreement with experimental findings, whereas the LES-based Test section tubes were made of clear acrylic to provide optical
simulation [30] gives very good quantitative agreement with the access for measurements of axial U (x direction), radial V (r direc-
experiment. Nevertheless, since the RANS requirements of the tion), and azimuthal W (h direction) components of mean velocity
model to the computational resources are much lower, and the cal- using single-component helium-neon LDV system with Dantex
culation time is shorter by orders as comparing to the LES, the sim- 55 transmitting optics. Statistical uncertainty in measuring the
ulation was carried out using realizable j-e model. Not claiming the mean velocity was kept below 4% by collecting 30,000 samples in
quantitative agreement, the authors of this paper expected to highly turbulent zones. The five measurement planes have been
reveal the dependence of the upstream axial penetration depth chosen for comparison: x = 40 mm (x/D = 0.43, upstream plane),
parameter hv/D of the developed counter flow jet on J, parameter x = 0 (JIP), and three downstream planes x = 6 mm (x/D = 0.06), x =
C (formula (16)) and radial penetration depth parameter h/D of 12 mm (x/D = 0.13) and x = 40 mm (x/D = 0.43).
the jets (formula (5)). General view of computational domain is shown in Fig. 3b. The
The SIMPLE algorithm of pressure-based solver was employed central x-y plane with half cross-section of the hole corresponds to
to correct the pressure term. For pressure, momentum, energy, tur- test section plane presented in Fig. 3a. As seen, in our calculations
bulent kinetic energy (TKE) and turbulent dissipation rate (TDR) we consider only flow fields in inner tube without coupling the
equations first-order upwind scheme have been applied since cho- annulus geometry. Instead, mean jet velocity Vj has been evaluated
sen discretization method with lower-order schemes generally based on velocity profile across hole diameter on castellated grid
offering the closest agreement to JIC experiment [37]. The model approximation proposed in [37]. In our computations grid density
default parameters have been specified, in particular, TKE Prandtl of the computational domain was following: the minimal cell size
number is equal to 1.0, TDR Prandtl number – 1.3. It was assumed was 0.2 mm, the maximal size – 2.0 mm, the cell growth rate –
that the solution convergence reached as the residuals reduced 1.15. To validate the model, x-component (U) and y-component
within all RANS equations by three orders, and by 4–5 orders for (V) in predicted velocity flowfields have been compared with axial
the energy equation. U (x direction), radial V (r direction) components of the mean
velocity measured in [35].
In Fig. 4 measured and predicted profiles of axial velocity nor-
malized by Um are drawn along the inner tube centerline. It can
5.3. Validation of the numerical model be seen that numerical solution indicates a noticeable under
prediction of counter flow jet axial velocity. Upstream
The realizable j-e model has been validated using JIC experi- penetration depth of counter jet flow marked by stagnation point
mental data by Baker and McGuirk [35] as well as recent data of U/Um = 0 – experimental value of lp/D  0.57 against numerical
experimental study by Thong et al. [41]. value  0.31. The latter is close enough to numerical results
E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178 1169

streaming mode, in the rest cases centerline values of axial velocity


are significantly underpredicted whilst on the periphery there is a
some overprediction of U/Um. This seems to be result of under pre-
diction of strength of jets impingement demonstrated by this
numerical model.

6. Results and discussion

6.1. Experimental results

Fig. 7a and b presents the typical axial temperature profiles for


each of two versions of type K thermocouple positions according to
Section 4, at various J and parameter C (test #1, see Table 1). Each
Fig. 4. Dimensionless measured and predicted axial velocity profiles along the
center-line.
axial temperature profiles corresponds to the case when developed
counter flow jet (which moves from the JIP toward negative x/D)
reaches a successive thermocouple. It is seen from the profiles in
obtained in [35–37]. Calculated centerline axial velocity profiles Fig. 7b that at J = 4956.2 temperature measured by the 2nd ther-
near JIP and downstream demonstrated a good agreement with mocouple decreases 100 K below the value measured without jet
experiment. injection. Therefore the ultimate axial penetration depth of the
In Fig. 5 radial profiles of measured and calculated axial U and counter flow jet hv  74–76 mm, consequently (hv/D)max  2.34.
radial V components of mean velocity normalized by Um in the cen- The similar centerline temperature profiles have been obtained
tral x-y plane are depicted. A satisfactory quantitative agreement in tests #2 and #3.
with experiments for both components is observed. Main features Under conditions of test #1 the radial temperature profiles were
of the components of mean velocity are well reproduced. A sub- also read in the plane of the axial cross section of the thermocouple
stantial discrepancy appear in Fig. 5a and f corresponding to #2 (2nd version of type K thermocouple positioning, x = 38 mm)
upstream plane x = 40 mm. This could be explained by numerical at different jets’ flow rates. Thermocouple motion pitch was 2 mm.
under prediction of counter flow jet penetration depth seen in The radial temperature profiles are shown in Fig. 8. The measure-
Fig. 4. ments were carried out: (1) without jets – denoted with the num-
ber 1 (J = 0), (2) with the jets’ flow rate Gj  3.1 g/s – denoted with
the number 2 (J = 441.0); (3) with the jets’ flow rate Gj  4.1 g/s –
5.3.2. Comparison with PIV measurements of axial velocity at various denoted with the number 3 (J = 745.3). Moreover, in the 1st type K
values of J thermocouple version of positioning (1st thermocouple position x
Thong et al. [41] experimentally studied mixing field and flow = 2 mm), the threshold value Jthr  4.0 was found, at which the
structures of turbulent radially injected jets in cylindrical cross- counter flow jet started to form (not shown in Fig. 2).
flow. They have utilized particle image velocimetry to measure As seen from Fig. 8, at J = 441.0, the counter jet just only reaches
planar velocities flowflields. The experimental nozzle was 1-m thermocouple #2, thus the farther from the centerline, the less the
long Perspex central cylinder with internal diameter D = 56 mm. temperature on the periphery differs from the undisturbed cross-
Four equi-distant jets with initial diameter of d = 6 mm were flow temperature. At J = 745.3, the temperature is almost constant
placed at one central diameter upstream of the nozzle exit. Water in every point, which means that the plane x/D = 1.19 completely
of jets and primary flow was seeded with 50-lm Dantec Dynamic lies in the recirculation flow zone, RFZ. At the point nearest to the
PSP-50 polyamide seeding particles. The nozzle and laser light wall, the temperature is almost constant, which seems to be result
sheet (532 nm Quantel Brilliant B Nd:TAG laser) were aligned such of influence of wall boundary layer.
that the sheet was coincident with nozzle centerline plane. 2-D PIV To generalize the available experimental data on the multi-jet
image ensemble is processed using single-pass standard fast Four- JIC in the cylindrical duct suitable for the evaluation of the axial
ier transform to obtain raw velocity vector data which then are penetration depth of the counter flow jet directed toward the
analyzed using OMA-X freeware. PIV image sample size of 10 crossflow, available literature sources have been analyzed. The
images provides uncertainty level < 5%. results are presented in the summary Table 2 below. In this table
Three modes of flow structure have been selected for compar- the leftmost column gives the experimental data source, successive
ison: (1) streaming mode (primary flow Reynolds number Re = 4 columns present the parameters of the mixer geometry, i.e. num-
6467, J = 9.3) when jets do not impinge; (2) impinging mode (Re ber of orifices n, ratio of orifice diameter to the channel diameter d/
= 4629, J = 18.1) when jets impinge without back-flow formation; D, ratio of the circumferential distance between the orifices to the
(3) back-flow mode (Re = 1386, J = 203.0) when impinging jets orifice diameter S/d (formula (17)), ratio of the channel wall thick-
form counter flow jet. These correspond to momentum ratio of ness to the orifice diameter t/d. The next five columns correspond
0.1, 0.2 and 2.3, respectively in [41]. to the jet and crossflow parameters, i.e. the fluid type, ratio DR of
Computational domain spanned 180° sector of the cylindrical jet density to the main flow density (formula (6)), the square root
nozzle, grid parameters were close to the previous ones: the min- of momentum-flux ratio J1/2 (formula (6)), Re number of the jets
imal cell size 0.17 mm, the maximal size – 1.75 mm, the cell and orifice discharge coefficient Cd. The rightmost column presents
growth rate – 1.10. the method of evaluation of the counter flow jet axial penetration
In Fig. 6 predicted and measured radial profiles of axial velocity depth into the crossflow. It is seen that geometrical characteristics,
U normalized by primary flow velocity Um at nozzle outlet (one turbulent flow parameters of mixers and their ranges presented in
duct diameter downstream of JIP) for above mentioned modes of Table 2 are close enough or overlapped and appear to be pertinent
flow structure are presented. to further generalization.
As seen, measured profiles in Fig. 6b and c happened to be Fig. 9 presents the experimental and calculation dependencies
asymmetrical. It can be also noticed that predicted and measured of hv/D on J1/2 in order to compare the present results (Case #6)
profiles of axial velocity agree quantitatively only in case of and previous ones (Case #5). As is seen from Fig. 9, at the same
1170 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

Fig. 5. Dimensionless axial velocity profiles as function of radial distance from centerline: (a) – x/D = 0.43, (b) – x/D = 0 (JIP), (c) – x/D = 0.06, (d) – x/D = 0.13 and (e) – x/D =
0.43; dimensionless radial velocity profiles as function if radial distance from centerline: (f) – x/D = 0.43, (g) – x/D = 0 (JIP), (h) – x/D = 0.06, (i) – x/D = 0.13 and (j) – x/D = 0.43.
E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178 1171

Fig. 6. Predicted and measured radial profiles of dimensionless axial velocity U at nozzle outlet (one duct diameter downstream of JIP): (a) – streaming mode (J = 9.3); (b) –
impinging mode (J = 18.1); (c) – back-flow mode (J = 203.0).

Fig. 7. Cylindrical duct centerline temperature profiles as a function of non-dimensional distance x/D upstream of JIP (x/D = 0) at various J and corresponding parameter C: (a)
– 1st version of type K thermocouples positioning; (b) – 2nd version.

5 mm, grey circles), where only the linear dependence is observed.


Experimental data of Case #6 show that axial penetration depth hv
of counter flow jet does not exceed 2.4 times the duct diameter D.
Solid black and grey lines depict the numerical results in the
conditions of Case #6 and #5, respectively. It is obvious, that if
the calculation for Case #6 gives the satisfactory qualitative agree-
ment with the experiment, the calculation for Case #5 does not at
J1/2 above 15. Moreover, the ultimate penetration depth for Case #5
is even lower than one for Case #6 (1.8–2.0 of the channel diameter
D), which contradicts to the experimental data. Evidently, for Case
#5, the experimental curve must lay above the one for Case #6.
The discrepancy between the experimental data of the Case #5
and #6 show that at J1/2 > 15 both parameters J1/2 and C (constant
number of orifices n = 8, formula (16)) are not universal from the
viewpoint of description of counter flow jet - crossflow interaction.
Fig. 10 presents the experimental data for the dependence hv/D
on the radial penetration depth parameter h/D for each Case pre-
sented in Table 2. Opposite to the dependence presented in
Fig. 8. Local temperature radial profiles at the cross-section x/D = 1.19 (upstream
Fig. 9, here an interval can be observed (Region I), wherein black
of JIP) under test #1 conditions.
circles (Case #6) and grey circles (Case #5) are located along one
dashed line: ¾ < h/D < 6, which corresponds to 4.0 < J1/2 < 30.0 in
number of orifices n = 8, the orifice diameter d = 3 mm (Case #6), Fig. 9. Note that in this interval, Fig. 9, both groups of experimental
the experimental dependence (black circles) has a linear, non- points lie on its straight line each, and for Case #5, the tangent of
linear, and asymptotic regions, in contrast with the Case #5 (d = the straight line inclination angle is about 5/3 = 1.67 times higher
1172 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

Table 2
JIC experimental cases with the displayed upstream counter flow jet penetration.

Case Geometry of mixer JIC flow parameters Method for evaluation of axial upstream jet penetration
1/2 depth
n d/D S/d t/d Fluid DR J Orifice jet Cd
Re
number
#1 6 0.22 2.36 0.25 Water 1 2.7 >2.4  104 0.64 Distance between the 2nd stagnation point and JIP
Spencer 5.0
[34]
#2 6 0.17 3.14 0.16 Water 1 4.6 >2.4  104 0.64 Distance between the 2nd stagnation point and JIP
Baker and McGuirk 0.20 2.59
[35]
#3 4 0.25 3.14 0.50 Water 1 4.0 2.7  103–3.5  104 0.80* Distance between JIP and a point at which local
Luo et al. 5.6 dimensionless concentration of the tracer ft = 0.8 (ft = 1.0
[38,39] 8.0 for crossflow, ft = 0.0 for jets) or a point at which
temporal unmixedness Itemp = 0.1 (Itemp = 1.0 for
segregated crossflow and jets, Itemp = 0.0 for complete
mixing)
#4 4 0.11 7.33 25.0 Water 1 4.2 2.1  103 0.55* Distance between the 2nd stagnation point and JIP
Thong et al. 7.2
[41] 14.2
#5 8 0.16 2.51 0.60 Air 4.2–4.6 3–25 1.4  103–1.3  104 0.80 Distance between JIP and a point at which at least 100 K
Kartaev et al. N2 temperature drop indicated in comparison with the
[46,47] temperature without disturbing counter flow
#6 8 0.09 4.19 1.33 Air 4.1–4.8 3–75 2.8  103–5.1  104 0.80 Distance between JIP and a point at which at least 100 K
Kartaev et al. 1.67 N2 temperature drop indicated in comparison with the
(present work) temperature without disturbing counter flow
*
The discharge coefficient Cd has been estimated on the base of the Idel’chik handbook [66].

Fig. 9. The parameter of counter flow jet penetration depth hv/D as function of Fig. 10. The parameter of counter flow jet penetration depth hv/D as a function of
square root of J: black circles – experimental data of our tests (Case #6); grey circles parameter of orifice jet radial penetration depth h/D: black circles – experimental
– experimental data obtained in [46,47] (Case #5). Black solid line – simulation data of our tests (Case #6); grey circles – experimental data obtained in [46,47]
results for Case #6; grey solid line – simulation results for Case #5. Black dashed (Case #5). Rest of symbols: inversed triangles – experiments of Case #1, ordinary
lines correspond to Regions I, II and III; grey dashed line fitted to experimental triangles – Case #2, stars – Case #3, squares – Case #4. Black dashed lines
points of Case #5. correspond to Regions I, II and III of Case #6. Extended dashed line adapted to
experimental points of Case #5.

than it is for the straight line plotted for Case #6, which is equal to
the ratio of the orifice diameters. It correlates well with the for- experimental data of Cases #1–#4 from other studies
mula (7). [34,35,38,39,41]. The right shift of the threshold values of h/D
The non-linear region (Region II) for Case #6 lies within the (Cases ##1, 4 against Cases ## 5, 6) could be a result of the lower
range 6.0 < h/D < 12.0 (30.0 < J1/2 < 60.0), and, similarly to the coun- values of Cd.
ter flowing jet emanating from the tube (CJIM, dependence (3)), Note that the symbols of Cases ##1, 2 (n = 6) are located within
here it is possible to perform the approximation by the dependence the linear region (also owing to the close values of S/d), whereas for
hv/D on (h/D)1/3, J1/6 (or C1/3). The asymptotic region (Region III) for Case #3 and especially Case #4 (n = 4), the deviation from the lin-
Case #6 is found at h/D > 12.0 (J1/2 > 60.0). Fig. 10 also presents the ear region is clearly seen. Possibly, other things equal, the lower
E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178 1173

the jet number, the earlier the non-linear character of the depen- ious J calculated based on steady-state model (RANS approach).
dence hv/D = f(h/D) becomes apparent. The same conclusion seems They correspond to two pointed regions: Region I and Region II. It
to be valid for the orifice diameter d: its reduction causes the tran- is evident that as J grows, the counter flow jet penetration depth
sition to the non-linear character of this dependence. As clearly increases, and a certain flow pattern asymmetry is seen. The reason
seen the dependence hv/D = f(h/D) has a common initial linear is that at J1/2 > 15, the solution exhibited clearly unsteady behavior.
region for all Cases, but at higher values of h/D a divergence of Within this range of J continuity residuals were (1–3)  103
curves is expected. Hence, we can assume that the parameter of whereas the remaining residuals were at level of 104–105. As
jet radial penetration depth h/D also cannot be a universal indica- underlined by Muirhead and Lynch [73] although RANS models
tor of the counter flow jet penetration depth upstream of JIP. used to have low residuals and could be considered converged,
solutions were prone to be unstable owing to the difficulty of RANS
models in capturing the time-average behavior of highly unsteady
6.2. Results of numerical study
phenomena. Thus, for chosen values of momentum-flux ratio (J1/2
> 15) URANS model has been applied.
This subsection presents the results of the numerical calcula-
Time step size of 103 s was set, with total number of time steps
tions for Case #6, and, as seen from Fig. 9, they are in qualitative
being equal to 5000, in order to resolve low-frequency fluctuations.
agreement with obtained experimental findings.
Therefore the lower limit of frequency of fluctuations was 0.2 Hz
Fig. 11 shows the temperature fields and streamlines in the
whilst the maximal frequency – 500 Hz.
radial-axial plane (symmetry plane) of the cylindrical duct at var-

Fig. 11. Results of modeling of jets gas dynamics in radial–axial plane (RANS approach) - contours of temperature with streamlines: (a) – J1/2 = 21.0 (Region I); (b) – J1/2 = 38.1
(Region II).

Fig. 12. Results of modeling of jets gas dynamics in radial–axial plane (URANS approach) – instantaneous contours of temperature with streamlines at J1/2 = 21.0 (Region I): (a)
– t = 0.767 s; (b) – t = 1.208 s; at J1/2 = 38.1 (Region II): (c) – t = 1.036 s; (d) – t = 1.421 s.
1174 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

In Fig. 12 instantaneous contours of temperature with flow in the center of the channel 300 mm  300 mm, i.e. the hydraulic
streamlines are presented: first J1/2 = 21.0 (Region I) at 0.767 and channel diameter was D = 300 mm (d/D = 0.017 and 0.033).
1.208 s (Fig. 12(a) and (b), respectively), and second J1/2 = 38.1 It should be emphasized that it is not exactly known whether
(Region II): at 1.036 and 1.421 s (Fig. 12(c) and (d), correspond- counter flow jet ‘flapping’ in calculations is a phenomenon
ingly). It is seen an arbitrary ‘flapping’ of counter flow jet and sub- observed in above mentioned experiments or not. Further consid-
stantial distortion of toroidal vortex, note that instability rises eration is based on RANS approach. It is assumed that time-average
along with J. For each couple of contours (Fig. 12(a) and (b) and RANS approach allows us to reveal basic features of developed
Fig. 12(c) and (d)) there is respective intermediate temperature counter flow jet formed as result of impingement of multiple round
flowfield being close enough to that depicted in Fig. 11(a) and (b). jets radially injected into a confined crossflow. Unsteady behavior
Based on elapsed time of calculations, frequency of ‘flapping’ of counter flow jet in cylindrical duct, is of special interest and,
has been evaluated to be within 1–2 Hz. Averaging of 15–20 possibly, will be studied in our future work.
instant axial temperature contours taken with 55–60 ms time slice Fig. 13 shows the dimensionless velocity and turbulent inten-
has been done, and mean parameter of axial penetration depth hv/ sity profiles on the duct axis normalized by mean crossflow veloc-
D has been evaluated. It turned out to be slightly lower (3–4%) ity Um (Fig. 13(a) and (b)), plus the axial dimensionless
compared to that obtained in RANS calculations for selected values temperature profiles referred to the crossflow temperature Tm
of J1/2 = 21.0, 38.1 and 70.4 shown in Fig. 9. (Fig. 13(c)). These profiles are in line with temperature flowfields
Similar counter flow jet ‘flapping’ around the duct centerline shown in Fig. 11 for J1/2 = 21.0 and 38.1. The point x/D = 0 lies in
was observed experimentally in [41] (Case #4) with parameter d/ the JIP plane, the region x/D < 0 corresponds to the channel axis
D = 0.11, interestingly, that calculations under conditions of Case part along which the counter flow jet moves toward the crossflow.
#6 have been done at d/D = 0.09. Perhaps, the less d/D, the more It is seen that, along with an increase of J, the axial velocity of
unstable counter flow jet. The same phenomenon was observed the counter jet and fluctuation intensity rise, too. Counter jet accel-
in the water tests in the CJIM configuration [51], when the inner eration begins from the jet impingement point (x/D = 0, the 1st
tube diameters were equal to 5 mm and 10 mm, they were located stagnation point) up to x/D  0.2, and then counter jet starts to

Fig. 13. Duct centerline dimensionless profiles (at various J): (a) – axial velocity; (b) – turbulent intensity; (c) – temperature. Rhombs correspond to experimental data at J1/2
= 21.0, circles – J1/2 = 70.4.
E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178 1175

somehow downstream of 2nd stagnation point, it means that the


maximal velocity fluctuations on the axis are observed in the
RFZ, though the zone of turbulent fluctuations on the axis is located
both downstream and upstream of the 2nd stagnation point (see
Fig. 13(a) and (b)).
As J rises the profiles of the dimensionless temperature profiles
shift farther upstream (Fig. 13(c)). At the normalized jet tempera-
ture Tj/Tm  0.2, the jets are not mixed with the crossflow nearby
the 1st stagnation point at high J (Region III), the mixing zone dis-
places progressively upstream of JIP. Rhombs in Fig. 13(c) desig-
nate the experimental data at J1/2 = 21.0 (Region I), circles shows
the same at J1/2 = 70.4 (Region III). Evident that the higher J, the
more the difference between the calculations and experiment.
One more fact can be pointed out too, namely that the area of
maximal temperature gradients on the channel axis feasibly coin-
cides with the turbulent fluctuation zone, which indicates the high
intensity of the turbulent mixing which is likely due to Kelvin-
Helmholtz instability developed on the RFZ stagnation surface or
zero axial velocity surface (see Fig. 1(b)). Kelvin-Helmholtz insta-
bility occurs in flows with different densities moving at various
speeds. This could be seen from Fig. 14 in which local dimension-
less radial profiles of axial velocity and temperature (cross-section
x/D = 0.5) upstream of JIP are shown for two values of J1/2: 13.2
and 21.0. In Fig. 14(a) cross-section of counter flow jet zone is lim-
ited within the interval where axial velocity is negative. It is clear
that there is a strong velocity difference across the interface (RFZ
stagnation surface) between two fluids with variable density (dif-
ferent temperature). This difference proved to become larger with
increasing value of J.
In Fig. 15(a) profiles of centerline gauge pressure at various
momentum-flux ratios are drawn. Initial gauge pressure is zero-
referenced against atmospheric pressure of the cylindrical duct.
Fig. 14. Local dimensionless radial profiles at the cross-section x/D = 0.5
As J grows a substantial increase of pressure peak at stagnation
(upstream of JIP) for values of J1/2 13.2 and 21.0: (a) – axial velocity; (b) –
temperature. point x/D = 0 is observed where jets impinge. In essence, there is
a significant correlation between plots in Figs. 13(a) and 15(a). As
could be seen the pressure minima are found at the points corre-
decelerate. As J rises, the axial point of the counter jet stagnation sponding to maxima of axial hydrodynamic head of upstream
(the 2nd stagnation point) shifts upstream, but no shift of this and downstream jets (Fig. 13(a)), with total pressure being con-
point is observed in the Region III. Downstream (x/D > 0), the stant. A remarkable growth of absolute value of pressure gradient
almost symmetrical acceleration of the reflected jet followed by DP located within RFZ is seen. In addition, according to calculation
its deceleration due to the channel expansion is observed, too. results there is a noticeable increase of gauge pressure in the zone
Turbulence intensity profiles at every J on the channel axis have undisturbed by counter flow jet up to 370 Pa at J1/2 = 70.4. In Fig. 15
two peaks: in the jet impingement point in the JIP, and in the point (b) the absolute value of pressure gradient DP as a function of J1/2 is

Fig. 15. Profiles of gauge pressure on duct centerline at various J – (a); absolute value of pressure gradient on duct centerline upstream of JIP as function of square root of J –
(b). Dashed lines denote mean inclination angle for each Region.
1176 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

Fig. 16. Temperature distributions and streamlines at JIP, TKE j profiles along jet Line 1 and bisectoral Line 2 for two values of J1/2: (a) – 21.0, (b) – 70.4.

plotted. It is seen an increasing contribution of drag force (propor- profiles along these lines. Similarity of the profiles along each line
tional to DP) which decelerates the counter flow jet when increas- for both J is noticeable; as J rises, the TKE rises more than one order.
ing J. Moreover, a rate of growth of this contribution depends on At constant J, the difference in TKE distribution profiles along the
chosen Region, this has been marked by inclination angle of dashed Line 1 and Line 2 is evident. The possible reason of such a difference
lines for each Region. At low J pressure gradient grows slowly and is a turbulent interaction of neighboring jets prior to the impinge-
counter flow jet penetrates upstream of JIP linearly with J1/2 (Region ment and hence generating the additional turbulent energy. Note
I). In Region II drag force starts to increase faster and counter flow that it means that at the 22.5° turn, the character of the TKE j
jets seems to penetrate according to the one-third power law of dependence along the assigned direction changes. In turn, it per-
J1/2, i.e. J1/6. At high values of J (Region III) drag force grows rapidly mits concluding the presence of the substantial anisotropy of the
and counter flow jet cannot penetrate deeper. Hence, this might be turbulent viscosity lt, related with TKE j.
the reason of finite penetration depth of counter flow jet. In con- Fig. 17 presents the profiles of the mixing nonuniformity degree
0
trary to Case #6, in Case #5 calculated profiles of gauge pressure f =f eq , calculated at several cross sections of the duct channel, both
upstream of JIP (left branches of profiles) were all above or a bit upstream and downstream of JIP, for three values J: J1/2 = 21.0 and
lower at minima the axis of abscissae. Nevertheless, we still have 27.3 (Region I), J1/2 = 70.4 (Region III); the respective values of the
no evidence-based arguments to explain such a difference between
Cases #6 and Case #5, possibly, our future work will be focused on
this issue.
Fig. 16 demonstrates the temperature fields and streamlines at
JIP for two values of square root of momentum-flux ratio – J1/2 =
21.0 (Fig. 16(a)) and J1/2 = 70.4 (Fig. 16(b)). The flow pattern turned
out to be very similar to the one obtained at lower J1/2 < 10 by the
planar digital imaging method [8], where the mixing indicator was
the Mie scattering of radiation from aerosol particles injected
together with the jets. The jets flowing out from the orifices, form
two counter rotating vortices of equal strength, which remain in
the near-wall region. It can be noted from Fig. 16, that, as the jet
strength rises, these vortices from adjacent jets start to interact
and promote more intensive mixing of that part of crossflow which
penetrates between jets. It is seen that at J1/2 = 21.0, the tempera-
ture between jets is about 700–800 K, whereas at J1/2 = 70.4, it is
within 400–600 K. In addition, Fig. 9(a) also illustrates a displace-
ment of the jet impingement point from the channel axis due to
the turbulent interaction between primarily opposite jets.
In the same figure two lines are drawn: Line 1 connects the cen- 0
Fig. 17. Calculated centerline profiles of f =f eq (dashed lines and solid one) for three
tral point formed by the crossing channel axis and JIP, and the values of J and those of corresponding parameter C: J1/2 = 21.0 and 27.3 (Region I); J1/
point on the orifice jet injection point; Line 2 lies in the bisectoral 2
= 70.4 (Region III). Location of filled rhomb is based on experimental data at plane
plane between the orifices. On the right, there are the plots of TKE j x/D = 1.19 for J1/2 = 21.0, filled square – experimental data for J1/2 = 27.3.
E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178 1177

parameter C calculated by the formula (16), are given, too. Two [3] J.D. Holdeman, Mixing of multiple jets with a confined subsonic crossflow,
Prog. Energy Combust. Sci. 19 (1993) 31–70.
solid symbols (rhomb and square) correlate with two sets of exper-
[4] A.R. Karagozian, Transverse jets and their control, Prog. Energy Combust. Sci.
imental data obtained in the plane x/D = 1.19 at J1/2 = 21.0 and 36 (2010) 531–553.
27.3 (curves 2 and 3 in Fig. 8). Good agreement of the numerical [5] M.S. Hatch, W.A. Sowa, G.S. Samuelsen, J.D. Holdeman, Geometry and flow
results and experimental data is seen in the linear area (Region I). influences on jet mixing in a cylindrical duct, J. Propul. Power 11 (3) (1995)
393–402.
It is clearly seen that even in the linear Region I, the mixing [6] M.S. Hatch, W.A. Sowa, G.S. Samuelsen, J.D. Holdeman, Jet Mixing into a Heated
nonuniformity is below 10% at JIP, i.e. the mixing process is more Cross Flow in a Cylindrical Duct: Influence of Geometry and Flow Variations,
localized in the RFZ upstream of JIP when J increases. The mixing NASA TM 105390, 1992, pp. 1–26 (also AIAA-92-0773).
[7] J.T. Kroll, W.A. Sowa, G.S. Samuelsen, J.D. Holdeman, Optimization of orifice
character changes gradually: from the jet mixing at low J (<100) geometry for crossflow mixing in a cylindrical duct, J. Propul. Power 16 (6)
to the more intensive mixing in the RFZ, i.e. in the zone of crossflow (2000) 929–938 (also NASA CR 198482, 1996).
and counter flow jet interaction at J > 400. [8] A. Vranos, D.S. Liscinsky, B. True, J.D. Holdeman, Experimental Study of
Crossstream Mixing in a Cylindrical Duct, NASA TM 105180, 1991, pp. 1–14
(also AIAA-91-2459).
[9] D.B. Bain, C.E. Smith, D.S. Liscinsky, J.D. Holdeman, Flow coupling effects in jet-
7. Conclusions
in-crossflow flowfields, J. Propul. Power 15 (1) (1999) 10–16.
[10] D.S. Liscinsky, A. Vranos, R.P. Lohmann, Experimental Study of Cross Flow
The upstream penetration depth of the counter flow jet, formed Mixing in Cylindrical and Rectangular Ducts, NASA CR 187141, 1993, pp. 1–
54.
by the impinging radially injected jets in the high-temperature
[11] M.S. Hatch, W.A. Sowa, G.S. Samuelsen, J.D. Holdeman, Influence of Geometry
crossflow of the cylindrical duct has been analyzed numerically and Flow Variation on NO Formation in the Quick Mixer of a Staged
and experimentally as a function of momentum-flux ratio J for Combustor, NASA TM 105639, 1995, pp. 1–14.
given mixer geometry. It was obtained that, both in the experiment [12] M.Y. Leong, G.S. Samuelsen, J.D. Holdeman, Jet Mixing in a Reacting Cylindrical
Crossflow, NASA TM 106975, 1995, pp. 1–25.
and calculation, the dependence of the upstream penetration depth [13] M.Y. Leong, G.S. Samuelsen, Quick-Mixing Studies Under Reacting Conditions,
hv/D of the counter flow jet on J1/2 has the linear, non-linear, and NASA CR 195375, 1996, pp. 1–132.
asymptotic regions; their boundaries have been estimated. Accord- [14] M.V. Talpallikar, C.E. Smith, M.-C. Lai, J.D. Holdeman, CFD analysis of jet mixing
in low NOx flametube combustors, J. Eng. Gas Turbines Power 114 (1992) 416–
ing to the experimental data, the counter flow jet on the axis can- 424 (also NASA TM 104466, 1991).
not penetrate into the crossflow deeper than 2.3 of the channel [15] V.L. Oechsle, J.D. Holdeman, Numerical Mixing Calculations of Confined
diameter D, the calculation gives the approximate value of 1.8– Reacting Jet Flows in a Cylindrical Duct, NASA TM 106736, 1995, pp. 1–71
(also AIAA -95-0733).
2.0 of the channel diameter. [16] V.L. Oechsle, Mixing and NOx Emission Calculations of Confined Reacting Jets
Investigation results show that none of three parameters, Flows in a Cylindrical Duct, NASA CR-2003-212321, 2003, pp. 1–223.
namely momentum-flux ratio J, optimal mixing parameter C, [17] V.L. Oechsle, H.C. Mongia, J.D. Holdeman, An Analytical Study of Jet Mixing in a
Cylindrical Duct, NASA TM 106181, 1993, pp. 1–60 (also AIAA-93-2043).
parameter h/D of the orifice jet radial penetration depth, is univer-
[18] J.D. Holdeman, C.T. Chang, Mixing of Multiple Jets with a Confined Subsonic
sal from the viewpoint of the description of the dependence of the Crossflow. Part III The Effects of Air Preheat and Number of Orifices on Flow
parameter hv/D of axial depth of the counter flow jet penetration and Emissions in an RQL Mixing Section, NASA TM 2008-215151, 2008, pp. 1–
27.
into crossflow of cylindrical duct. Nevertheless, within the linear
[19] G.S. Samuelsen, J. Brouwer, M.A. Vardakas, J.D. Holdeman, Experimental and
region of the dependence hv/D = f(J1/2), the parameter h/D permits modeling investigation of the effect of air preheat on the formation of NOx in
evaluating the hv/D value. an RQL combustor, Heat Mass Transf. 49 (2013) 219–231.
The mixing intensity in the recirculation zone upstream of jet [20] A.J. Bishop, Confined mixing of multiple transverse jets, MS Thesis, California
Polytechnic State University, San Luis Obispo, 2012.
injection plane has been studied under the conditions of developed [21] G. Zhu, M.C. Lai, T. Lee, Penetration and mixing of radial jets in neck-down
counter flow jet. It is demonstrated that the mixing is progressively cylindrical crossflow, J. Propul. Power 11 (2) (1995) 252–260.
localized in the recirculation zone when J increases. The mixing [22] Y. Tao, W. Adler, E. Specht, Numerical analysis of multiple jets discharging into
a confined cylindrical crossflow, Proc. Inst. Mech. Engin., Part E: J. Process
character changes too: from the jet mixing at low J to the more Mech. Eng. 216 (2002) 173–180.
intensive mixing in the toroidal vortex of the recirculation zone [23] A. Nirmolo, H. Woche, E. Specht, Temperature homogenization of reactive and
at large values of J. non-reactive flows after radial jets injections in confined cross-flow, Eng. Appl.
Comput. Fluid Mech. 2 (1) (2008) 85–94.
[24] A. Nirmolo, H. Woche, E. Specht, Mixing of jets in cross flow after double rows
of radial injections, Chem. Eng. Technol. 31 (2008) 294–300.
Conflict of interest
[25] D.J. Forliti, D.V. Salazar, A.J. Bishop, Physics-based scaling laws for confined
and unconfined transverse jets, Exp. Fluids 56 (A36) (2015) 1–16.
This manuscript has not been published and is not under con- [26] M.G. Ktalkherman, V.A. Emelkin, B.A. Pozdnyakov, Influence of the geometrical
sideration for publication elsewhere. We have no competing finan- and gas-dynamic parameters of a mixer on the mixing of radial jets colliding
with a crossflow, J. Eng. Phys. Thermophys. 83 (3) (2010) 539–548.
cial interests. This manuscript is in compliance with Ethics in [27] M.G. Ktalkherman, I.G. Namyatov, V.A. Emelkin, High-temperature pyrolysis of
Publishing Policy as described in the Guide for Authors. liquefied petroleum gases in the fast-mixing reactor, Int. J. Chem. Reactor Eng.
All authors (E.V. Kartaev, V.A. Emelkin, M.G. Ktalkherman, S.M. 9 (2011) 1–23.
[28] M.G. Ktalkherman, I.G. Namyatov, V.A. Emelkin, K.A. Lomanovich, High-
Aulchenko and S.P. Vashenko) express consent for publication the selective pyrolysis of naphtha in the fast-mixing reactor, Fuel Process. Technol.
proposed article in the International Journal of Heat and Mass 106 (2013) 48–54.
Transfer. [29] E.V. Kartaev, V.P. Lukashov, S.P. Vashenko, S.M. Aulchenko, O.B. Kovalev, D.V.
Sergachev, An experimental study of the synthesis of the ultrafine titania
powder in plasmachemical flow-type reactor, Int. J. Chem. Reactor Eng. 12
Acknowledgements (2014) 1–20.
[30] D.J. Clayton, W.P. Jones, Large eddy simulation of impinging jets in a confined
flow, Flow Turbul. Combust. 77 (2006) 127–146.
The authors are grateful to Dr. J.D. Holdeman (retired, NASA [31] E.M. Ivanova, B.E. Noll, M. Aigner, A numerical study on the turbulent schmidt
Glenn Research Center, Cleveland OH, USA) for useful discussions. numbers in a jet in crossflow, J. Eng. Gas Turbines Power 135 (2013) 011505-
1–011505-10.
[32] H. Urson, M.F. Lightstone, M.J. Thomson, Numerical study of jets in a reacting
References crossflow, Numer. Heat Transfer, Part A: Appl.: Int. J. Comput. Methodol. 40
(2001) 689–714.
[33] F. Davoudzadeh, D. Forliti, Numerical and experimental investigation of
[1] R.J. Margason, Fifty Years of Jet in Counter flow Research, Computational and
confined turbulent multiple transverse jets, in: Proceedings of 50th AIAA/
Experimental Assessment of Jets in Counter flow Meeting, AGARD CP-534,
ASME/SAE/ASEE Joint Propulsion Conference, Cleveland, OH, 2014, pp. 1–38
Winchester, UK, 1993, pp. 1.1–1.41.
(AIAA 2014-3789).
[2] J.D. Holdeman, D.S. Liscinsky, V.L. Oechsle, G.S. Samuelsen, C.E. Smith, Mixing
[34] A. Spencer, Gas Turbine Combustor Port Flows PhD Thesis, Loughborough
of multiple jets with a confined subsonic crossflow: part I – cylindrical duct, J.
University, Leicestershire, 1998.
Eng. Gas Turbines Power 119 (1997) 852–862, also NASA TM 107185, 1996.
1178 E.V. Kartaev et al. / International Journal of Heat and Mass Transfer 116 (2018) 1163–1178

[35] S.J. Baker, J.J. McGuirk, Multijet annulus/core-flow mixing – experiments and [54] A.N. Sekundov, The propagation of a turbulent jet in an opposing stream, in: G.
calculations, J. Eng. Gas Turbines Power 115 (1993) 473–479. N. Abramovich (Ed.), Turbulent Jets of Air, Plasma and Real Gas, Consultants
[36] J.J. McGuirk, A. Spencer, Coupled and uncoupled CFD prediction of the Bureau, New York, 1969, pp. 99–109.
characteristics of jets from combustor air admission ports, J. Eng. Gas [55] Z.W. Li, W.X. Huai, Z.D. Qian, Large eddy simulation of a round jet into a
Turbines Power 123 (2001) 327–332. counterflow, Sci. China Technol. Sci. 56 (2) (2013) 484–491.
[37] J.J. McGuirk, A. Spencer, CFD Modeling of Annulus/Port Flows, ASME Paper 93- [56] M. Sivapragasam, M.D. Deshpande, S. Ramamurthy, P. White, Turbulent jet in
GT-185. confined counterflow, Sadhana 39 (2014) 713–729.
[38] P. Luo, H. Jia, C. Xin, G. Xiang, Z. Jiao, H. Wu, An experimental study of liquid [57] J.G. Li, M. Ikeda, R. Ye, Y. Moriyoshi, T. Ishigaki, Control of particle size and
mixing in a multi-orifice-impinging transverse jet mixer using PLIF, Chem. phase formation of TiO2 nanoparticles synthesized in RF induction plasma, J.
Eng. J. 228 (2013) 554–564. Phys. D Appl. Phys. 40 (2007) 2348–2353.
[39] P. Luo, Y. Tai, Y. Fang, H. Wu, Mixing times in single and multi-orifice [58] N. Settumba, S.C. Garrick, Modeling and simulation of nano-aluminum
impinging transverse (MOIT) jet mixers with crossflow, Chin. J. Chem. Eng. 24 synthesis in a plasma reactor, in: K.K. Kuo, J.D. Rivera (Eds.), Advancement
(2016) 825–831. in Energetic Materials & Chemical Propulsion, Begell House Inc., 2007, pp.
[40] P. Luo, Y. Fang, B. Wu, H. Wu, Turbulent characteristics and design of 643–655.
transverse jet mixers with multiple orifices, Ind. Eng. Chem. Res. 55 (2016) [59] G. Salazar, K. Agrios, R. Eichler, S. Szidat, Characterization of the axial jet
8858–8868. separator with CO2/helium mixture: toward GC-AMS hyphenation, Anal.
[41] C.X. Thong, P.A.M. Kalt, B.B. Dally, C.H. Birzer, Flow dynamics of multi-lateral Chem. 88 (2016) 1647–1653.
jets injection into a round pipe flow, Exp. Fluids 56 (A15) (2015) 1–15. [60] G.N. Abramovich, The Theory of Turbulent Jets, MIT Press, Cambridge,
[42] C.X. Thong, B.B. Dally, C.H. Birzer, P.A.M. Kalt, E.R. Hassan, An experimental Massachusetts, 1963.
study of the near flow field of a round jet affected by upstream multi-lateral [61] G.N. Abramovich, The Theory of a Free Jet of a Compressible Gas, NACA TM
side-jet, Exp. Thermal Fluid Sci. 82 (2017) 198–211. 1058, 1944, pp. 1–91
[43] X. Yu, T. Chen, Q. Zhang, T. Wang, Mixing behaviors of jets in cross-flow for [62] P.V. Danckwerts, The definition and measurement of some characteristics of
heat recovery of partial oxidation process, Int. J. Chem. Reactor Eng. 15 (1) mixtures, Appl. Sci. Res. 3 (1952) 279–296.
(2017) 21–33. [63] A.H. Lefebvre, D.R. Ballal, Gas Turbine Combustion: Alternative Fuels and
[44] Q. Zhang, J. Wang, T. Wang, Enhancing the acetylene yield from methane by Emissions, third ed., CRC Press, 2010 (pp. 113–151).
decoupling oxidation and pyrolysis reactions: a comparison with partial [64] V.I. Kuzmin, A.A. Mikhalchenko, O.B. Kovalev, E.V. Kartaev, N.A. Rudenskaya,
oxidation process, Ind. Eng. Chem. Res. 55 (2016) 8383–8394. Technique of formation of an axisymmetric heterogeneous flow during
[45] T. Wang, X. Yu, T. Chen, Q. Zhang, CFD simulations of quenching process for thermal spraying of powder materials, J. Therm. Spray Technol. 21 (2012)
partial oxidation of methane: comparison of jet-in-cross-flow and impinging 159–168.
flow configurations, Chin. J. Chem. Eng. (2016), https://doi.org/10.1016/j. [65] R.T. Dittrich, C.C. Graves, Discharge Coefficients for Combustor Liner Air Entry
cjche.2016.04.039. Orifices I, Circular Orifices, NACA TN 3663, 1956, pp. 1–39.
[46] E.V. Kartaev, V.A. Emelkin, M.G. Ktalkherman, V.I. Kuzmin, S.M. Aulchenko, S.P. [66] I.E. Idel’chik, Handbook of Hydraulic Resistance, third ed., Begell House, 1996
Vashenko, Analysis of mixing of impinging radial jets with crossflow in the (pp. 21–25).
regime of counter flow jet formation, Chem. Eng. Sci. 119 (2014) 1–9. [67] J.E. Rohde, H.T. Richards, G.W. Metger, Discharge Coefficients for Thick Plate
[47] E.V. Kartaev, V.A. Emelkin, M.G. Ktalkherman, S.M. Aulchenko, S.P. Vashenko, Orifices with Approach Flow Perpendicular and Inclined to the Orifice Axis,
V.I. Kuzmin, Formation of counter flow jet resulting from impingement of NASA TN D-5467, 1969, pp. 1–28.
multiple jets radially injected in a crossflow, Exp. Thermal Fluid Sci. 68 (2015) [68] B.E. Launder, D.B. Spalding, The numerical computation of turbulent flows,
310–321. Comput. Methods Appl. Mech. Eng. 3 (1974) 269–289.
[48] T.R.K. Rao, Investigation of the Penetration of a Jet into a Counterflow MS [69] T.-H. Shih, W.W. Liou, A. Shabbir, Z. Yang, J. Zhu, A new j-e eddy-viscosity
Thesis, Iowa State University, Iowa City, 1958. model for high Reynolds number turbulent flows, Comput. Fluids 24 (3) (1995)
[49] W.D. Morgan, B.J. Brinkworth, G.V. Evans, Upstream penetration of an enclosed 227–238.
counterflowing jet, Ind. Eng. Chem. Fundam. 15 (2) (1976) 125–127. [70] J.O. Hinze, Turbulence, McGraw-Hill Publishing Co., New York, 1975.
[50] K.M. Lam, H.C. Chan, Investigation of turbulent jets issuing into a counter- [71] S.B. Pope, Turbulent Flows, Cambridge University Press, Cambridge, England,
flowing stream using digital image processing, Exp. Fluids 18 (1995) 210–212. 2000.
[51] M. Yoda, H.E. Fiedler, The round jet in a uniform counterflow: flow [72] D.G. Sloan, P.J. Smith, L.D. Smoot, Modeling of swirl in turbulent flows, Prog.
visualization and mean concentration measurements, Exp. Fluids 21 (1996) Energy Combust. Sci. 12 (1986) 163–250.
427–436. [73] K. Muirhead, S.P. Lynch, A computational study of combustor dilution flow
[52] S. Bernero, A Turbulent Jet in Counterflow PhD Thesis, Technical University of interaction with turbine vanes, in: Proceedings of 55th AIAA Aerospace
Berlin, Berlin, 2000. Sciences Meeting, Grapevine, TX, 2017, pp. 1–16 (AIAA 2017-0109).
[53] S.F. Saghravani, A.S. Ramamurthy, Penetration length of confined counter
flowing free jets, J. Hydraul. Eng. 136 (3) (2010) 179–182.

You might also like