The mathematical models for penetration of a liquid jets into a pool

IVAN V. KAZACHKOV
Department of Energy Technology, Division of Heat and Power
Royal Institute of Technology
Brinellvägen, 68, Stockholm, 10044
SWEDEN
[email protected] http://www.kth.se/itm/inst?l=en_UK

Abstract: - The peculiarities of a jet penetrating the liquid pool of different density were examined by means of
the non-linear and linear mathematical models derived including bending instability. Based on experimental
observations reported in the literature for a number of situations, the penetration behaviour was assumed to
govern the buoyancy-dominated regime. A new analytical solution of the one-dimensional non-linear model
was obtained for the jet penetration in this condition, as function of Froude number, jet/ambient fluid density
ratio and other parameters. The solution was analysed for a number of limit cases. Analytical solution of the
non-linear second-order equation obtained can be of interest for other researchers as the mathematical result.

Key-Words: - Jet, Penetration, Pool of Liquid, Non-linear, Analytical Solution, Bifurcation, Bending

1 Introduction
The penetration dynamics of a liquid jet into the
other liquid (or solid) medium has been
investigated by a number of researchers [1-15].
Most of the earlier studies have been performed
in the metal and nuclear industries, e.g. [1, 4-7,
9-11]. But the problem still remains, especially
in the case of the thick jets when they are
penetrating a pool of other liquid without
disintegration and in case of dominated inertia,
drag and buoyancy forces.
For the thin jets it has been shown [16] that
the jet instability might be caused by the
bending perturbations of its axis. The objective
of present paper is determining the penetration
Fig. 1. Experimental illustration of a jet penetrating
behaviours of a thick jet into a fluid pool and the pool of other liquid
deriving a penetration depth as a function of the
conditions and properties of a jet and a pool.
The jet penetration non-linear model is General scheme of the penetration process is
developed and some analysis is made for a illustrated by experimental data borrowed from
number of limit cases, which maybe of interest [17] shown in Fig. 1. It is clearly observed that
for some practical applications. the penetrating jet is going first with
General scheme of the penetration process is approximately stable radius and then changing
illustrated by experimental data borrowed from its radius abruptly to a bigger one. This is some
[17] shown in Fig. 1. It is clearly observed that interesting bifurcation point, which has got
the penetrating jet is going first with explanation from the analytical solution
approximately stable radius and then changing obtained in this paper.
its radius abruptly to another bigger one. This Some amount of air may also be entrained
bifurcation point is explained from the into a pool together with a jet. As shown in a
analytical solution obtained below. number of papers [18-20], when liquid jet
impacts a liquid pool, air is entrained in a pool consideration are of an order of a jet cross-
if jet’s velocity exceeds the threshold value. section and a surface tension is of an order of a
Phenomenologically based correlations for jet circular.
an air entrainment have been proposed in a few
papers, for example in [20]. Then it has been
considered [19] that instability responsible for
the air entrainment was caused by the gas
viscosity.
The analysis presented in [18] is based on
inviscid flow theory assuming that the air
entrainment was a result of a Helmholtz-Taylor
instability. This is an interesting complex
problem for a separate study, therefore an
influence of the air entrainment on a jet
penetration features is not considered here.
G. K. Batchelor [21] has also given the
equation to compute the momentum looses by a
shock of the jet on a liquid pool surface at the
initial moment of a jet penetration when moving
jet touches a pool having liquid with a zero
velocity. Using those equations one can
compute the abrupt change of a jet velocity at
the entrance to a pool. This phenomenon is not
taken into account here because it is easy to do
and it does not influence the solution considered
in this paper.
Fig. 2. Scheme of a jet penetration into the pool of
other liquid: phases by penetration
2 Problem Formulation
2.1 Physical model of a jet penetration 2.2 Non-linear mathematical model of a jet
Consider a jet penetrating the pool of other penetration
liquid as a body of a variable mass assuming Based on the above physical description of the
that the jet is moving under an inertia force problem, the equation of a jet momentum
acting against the drag and buoyancy forces conservation (considering a jet as a body of a
(see Fig. 2). The surface forces are supposed to variable mass) is the following:
be negligible comparing to those ones.
d ( hv1 ) 1
Then a jet radius is assumed approximately ρ1 = h( ρ1 − ρ 2 ) g − ρ 2 v12 , (1)
constant during the jet penetration or at least dt 2
during some part of the depth of penetration,
where h is a depth of a jet penetration into the pool,
which allows considering the jet being partly of
ρ1 , ρ2 are densities of the jet and fluid in the
a nearly constant radius. It allows calculating
the jet penetration step by step in general case pool, respectively, v1 is the jet velocity.
approximately taking the first constant jet Obviously here is v1 = dh/dt.
radius, then next constant jet radius, and so on. For the thick jets one can neglect surface forces
Strictly saying, such assumptions are always retaining the only drag force together with the
satisfied in case of a solid rod penetration into buoyancy and inertia forces. To estimate this
the liquid pool. But mainly it is also attainable simplification, consider when the ratio of the surface
assumption in case of a thick jet penetration force µ (∂v1 / ∂z ) taken by the entire jet surface to
into the pool because all the forces taken in a the drag force acting on a jet’s head is negligibly
small. Here µ is the dynamic viscosity coefficient, moment of time (t=0) the jet has velocity u0, and the
z is the coordinate perpendicular to a jet axis. Thus, fluid pool at the jet/pool contact area (h=0) changes
it yields to the following condition: abruptly its velocity from 0 to u0 (actually less than
u0, if energy dissipation is taken into account).
µ1 (∂v1 / ∂z ) s 2π r0 h << ρ2 v12π r02 / 2 , To avoid this singularity, let consider further the
following initial conditions instead of the above-
mentioned conditions:
where from estimating the velocity gradient as
(∂v1 / ∂z ) s ≈ v1 / r0 , one can finally get t =0, h = h0 , dh / dt = u p , (5)
Re >> 4(h / r0 ) ρ1/ 2 .
where h0 and up are the initial depth and velocity of a
Here ρ1 / 2 = ρ1 / ρ2 , Re = v1r0 ρ1 / µ1 is the
jet penetration (after a first contact of a jet with a
Reynolds number. For example, from the condition pool), which should be calculated later on. For some
obtained follows that by h / r0 = 10, ρ1/ 2 = 0.1 limit cases they could be taken from the studies of a
surface force is negligible comparing to the drag high-speed jet penetration [1, 5, 6, 13], e.g.
force by Re >> 4 .
λ
up = u0 ,
1+λ
2.3 Singularity of the initial conditions (6)
The initial conditions for the jet’s momentum
equation (1) should be stated as follows: where λ = ρ1 / 2 . Taking into account (5), (6), one
can obtain for ρ1 / 2 =1 :
t = 0, h =0, dh / dt = u0 ,
(2)  5t 
2/5
u p = 0.5 , h = h0  + 1 ,
where u0 is the initial jet velocity (before  4 h0 
penetration into the pool). (7)
−3 / 5
In case of ρ1 = ρ2 one can obtain from equation dh 1  5t 
=  +1 .
(1) the following simple equation: dt 2  4 h0 
2
d 2 h 3  dh  Thus, in case of the same densities of a jet and a
h +   =0,
dt 2 2  dt  pool, the jet velocity tends to zero asymptotically.
(3) Then the jet velocity decreases twice at the depth
h = h0 .
which is integrated through the next transformation: Here and further the penetration depth is
dimensionless value, and the scale is the jet’s radius
−1
 dh  d  dh  3 dh r0 .
   + = 0,
It is also interesting to calculate the characteristic
 dt  dt  dt  2h dt
distance where the jet loses its velocity of a given
where from yields value. This is easily determined from the equation
(4):
dh 3 / 2
h = c1 , v1 = u p ( h0 / h ) 3 / 2 , (8)
dt

so that where u p = 0.5 . As one can see from the equation

(8), the velocity of a jet penetration into a pool is
2 5/ 2 decreasing by the jet penetration depth as 1/ h 3 / 2 .
h = c1t + c 2 .
5 So far in a case of the same densities, the jet looses
(4) a half of its velocity at the depth h0 , then a half of
that velocity at the depth h1 = 2 h0 ≈ 1,6h0 , and
2/3
As one can see we have here some singularity with
the initial conditions (2) because at the initial
then ten times velocity decrease happens at the
depth h10 = 10 h0 ≈ 4,5h0 .
2/3 H 1
h0 = , up = . (12)
In a general case of the different densities of a jet 1 + 1 + ρ2 / 1 1 + ρ2 / 1
and a pool ( ρ1 / 2 ≠1 ) one needs to solve the non-
linear equation (1), which has the following The last formula above corresponds to (6),
dimensionless form: which was taken from the literature for the high-
speed jet/solid rod penetration into the liquid pools
2
d 2h  ρ  dh  ρ −1 and solid plates [1, 5, 6, 13]. Now (6) is rewritten as
h 2 + 1 + 2 / 1   + 2 / 1 h =0,
dt  2  dt  Fr
(9) u0
up = . (13)
1 + ρ 2 /1
where u0 is the velocity scale and r0 /u0 is the time
scale, Fr= u02/(gr0) is the Froude number, which The correspondence of (12) and (13) by u p is
characterizes the inertia and buoyancy forces' ratio. clearly observed from the Table 1 below.
As one can see from the above, the Froude
number and the density ratio totally predetermine
the process of a thick jet penetration into a pool of Table 1. The initial velocity and the depth of a jet
other liquid. penetration

2.4 The initial depth and corrected initial u p by u p by h0 by (12)

velocity of a jet penetration into a pool (12) (13)
The equation (9) is solved with the initial conditions
(5), where h0 and u p are determined using the ρ2 / 1 = 0 1 1 0
equation of a jet momentum and the Bernoulli
equation in the form:
ρ2 / 1 = 1 1/ 2 1/2 ( 2 −1 H )
ρ1 Hu 0 = ρ1 Hu p + ρ2 h0 u p ,
(10) ρ2 / 1 >> 11 / ρ2 / 1 1/ ρ2 / 1 ρ2 / 1 H (1 − ρ2 / 1 )
1 1 1
ρ1u 02 = ρ1u 2p + ( ρ1 − ρ 2 ) gh 0 / u 02 − ρ 2 u 2p ,
2 2 2

where H is the initial length for the finite length 2.5 The analytical solution of the second-
jet falling into the pool. In case of a jet spreading order non-linear differential equation
out from a nozzle (not of a finite length), this value The equation (9) can be solved using the following
is determined by the pressure at the outlet. special coupled transformations for the both
Now an analytical solution to the equation array dependent and independent variables, which were
(10) is presented in the following dimensionless found by the method described in [22]:
form:
2
2
 2 A +1 2 A +1 ,
H  1 + ρ2 /1  h =  X 2 A +1

h0 =  − 1 ,  2 
ρ2 /1  1 + 2(1 − ρ )h / Fr  (14)
 2 /1 0 
1
(11)  1  2 A +1
1
dt =   X 2 A +1
dτ ,
1 + 2h0 (1 − ρ 2 / 1 ) / Fr  2 A +1 
up = .
1 + ρ2 /1
where are: ρ2 / 1 = ρ2 / ρ1 , A = 1 + ρ2 / 1 / 2 .
Then, by a small density difference or by a small
initial depth of a jet penetration (comparing to the Implementation of (14) and a few further simple
Froude number) when h0 (1 − ρ2 / 1 ) << Fr , the transformations lead to the following linear second-
order equation in the new variables:
simpler approximations follow from (11):
 1
c1' +c 2
' kτ

 1  3+ρ2 / 1 3 + ρ2 / 1 3+ ρ2 / 1
d y ρ −1
2 2
t =   e + c3 .
+ =0. 2 /1
 3 + ρ2 / 1  c2' k
dτ 2 2 Fr 2
(15) From these equations requiring t = 0 , which leads
to τ = 0 , the constants c3 are got.
Here y is the new variable given by X = e y . The
Consequently, the real dimensionless time t is
solution of (15) is y =c1 e kτ + c2 e −kτ , where c1 , expressed through the artificial variable τ :
c2 are the constants computed using the initial
conditions (5). The eigen value k is ρ2 / 1 <1,
1
  3+ρ2 /1 3 + ρ2 / 1
k = (1 − ρ 2 / 1 )[1 + 0.5(1 + ρ 2 / 1 )] / Fr .
1
t =
3 + ρ 
 ⋅
 2 /1  k (c1 − c 2 )
(16)
(20)
In case of ρ2 / 1 >1 (a pool is denser than a jet), 1 c1 + c2
the eigen values are imaginary, and the solution is [ ( c1 + c2 ) + k (c1 − c2 )τ ]
3+ ρ 2 / 1 3+ ρ 2 /1
⋅ {e −e }
;
c cos kτ + c sin kτ .
´
1
´
2 (17)

2.5.1 Dimensionless time

The dimensionless time t is determined through the ρ2 / 1 >1,
variable τ by (14), which gives
1
 1  3+ρ2 /1 3 + ρ2 / 1
t =
3 + ρ 
 ⋅
ρ2 / 1 <1,  2 /1  c 2' k

(18) (21)
1 c1' + c2' kτ c1'
c1e kτ +c2 e − kτ
 1  3+ ρ2 / 1 3+ ρ 2 /1 3 + ρ2 /1
t =   ∫e
ρ2 / 1 +3
dτ + c3 ; ⋅(e −e )
 3 + ρ2 / 1 
Strictly saying, these equations are satisfied in a
ρ2 / 1 >1, small ε -surrounding of τ = 0 . In general case one
(19) needs to compute integrals in (18), (19) numerically.
1
c1' cos kτ +c2' sin kτ But for ρ2 / 1 ~1 and Fr>>1, the multiplayer of τ
 1  3+ ρ2 / 1

∫e
ρ2 / 1 +3
t =   dτ + c3 , has to be small value, which is possible using
 3 + ρ2 /1  approximations (20), (21) in a wider region of τ ,
and even if τ is not small but the condition kτ
where the constants c3 are calculated later on. For <<1 is satisfied.
τ << 1 , the following linear approximations by And further the expression (20) is presented in
the form:
kτ are satisfied: e ±kτ ≈ 1 ± kτ , c o ksτ ≈ 1, 1
c1e kτ +c2 e − kτ
sin kτ ≈ kτ .  1  3+ ρ 2 / 1 3 + ρ2 /1 3+ ρ2 / 1
Thus, the equations (18), (19) yield: t =   e − t0 ,
 3 + ρ 2 / 1  k ( c1 − c2 )
ρ2 / 1 <1, (22)

1
1 c1 +c2
 1 3+ρ2 / 1 3 + ρ2 / 1
1
[ ( c1 +c2 ) +k ( c1 −c2 )τ ]  1  3+ ρ2 / 1 3 + ρ2 / 1 3+ρ2 / 1 .
t =
3 + ρ   e 3+ρ2 / 1 + c3 t 0 = 
 3 + ρ  e
 2 /1  k ( c1 − c2 )  2 /1  k ( c1 − c2 )
. (23)

ρ2 / 1 >1,
For ρ2 / 1 >1 the corresponding expressions are 1
−
1
1
up Fr  2 3+ρ2 / 1 2 −
3+ρ2 / 1
obtained from (21) similarly. c = '
  (3 + ρ2 / 1 )
1 − ρ2 / 1  3 + ρ2 / 1 

2
h0 
2.5.2 Caclulation of the constants ,
Now using the initial condition (5) and correlations
so that c1 = ( c1 + c2) / 2, c2 = ( c1 − c2) / 2.
' ' ' '
(11), one can substitute (14) into (5) and calculate
constants. Thus, for ρ2 / 1 <1 (a jet is denser than a Then from the equations (23), (24) follows
pool) the equations for constants are: t 0 ≈1.02 (1 + ρ2 / 1 )h0 , t 0 = 2h0 / u p . By
ρ2 / 1 <<1, there is t0 ≈ h0 , and by ρ2 / 1 >>1
  2  3+ ρ 2 / 1  there is t0 ≈ ρ2 / 1 h0 .
c1 + c2 = ln   h0 2  ,
  3 + ρ 2 / 1  
1 1 Explicit form of the solution obtained
−
up Fr  2  3+ ρ 2 / 1 2 The solution (14) can be transformed to an explicit
c1 − c 2 =   ⋅ form as the function of t (exclude the artificial time
1 − ρ 2 / 1  3 + ρ 2 / 1
h0  τ ). For this purpose, from (23), (24) yields
1
−
3+ ρ2 /1
⋅ (3 + ρ 2 / 1 ) 1
 1  3+ρ2 /1 3 + ρ2 / 1
t + t0 = 
3 + ρ 
 ⋅
 2 /1  k (c1 − c 2 )
where from yields:
1
[ ( c1 +c2 ) +k ( c1 −c2 )τ ] ,
3+ρ2 / 1
⋅e
1  2 3+ ρ 2 / 1
 up Fr
c1 =  ln  h0 2
+ ⋅
2   3 + ρ 2 /1  h0 1 − ρ2 /1 and further it goes to

1
−
1
1  e kτ = [ (t + t0 )k (c1 − c2 ) ⋅
 2  3+ ρ 2 / 1 2 −
3+ ρ 2 / 1 
⋅  (3 + ρ 2 /1 ) 3+ ρ 2 / 1
,
1
−1
 3 + ρ 2 /1   1  3+ ρ2 /1
c +c
− 1 2  c1 −c2 , (26)
 ⋅ e 3+ ρ 2 /1


(24)  3 + ρ 2/1  
1  2 3+ ρ2 /1
 u Fr
c2 =  ln  h0 2  − p ⋅ or
2   3 + ρ 2 /1  h0 1 − ρ2 /1
α
kτ  u 
e = (t + t0 ) p  . (27)
1
−
1
1   2h0 
 2  3+ ρ 2 / 1 2 −
3+ ρ 2 / 1 .
⋅  (3 + ρ 2 /1 )
 3 + ρ 2 /1  
 With account of (16) yields:

ρ 2 / 1 +1
For ρ2 / 1 >1 (a pool is denser than a jet), from (14), 1 − ρ 2 /1 h0  2  2(3+ ρ2 / 1 )
α=   ⋅
(5), accounting (17), yield the constants c1′, 2 : up Fr  3 + ρ 2 /1  (28)
1
+1
3+ ρ 2 / 1
⋅ (3 + ρ 2 /1 )
  2  3+ ρ 2 / 1 
c = ln 
'
1  h0 2  ,
 3 + ρ 2 /1   kτ
Accounting that e c1e +c2e = e c1e e c2 e , and
− kτ kτ − kτ

(25) using the equations (27), (28), (22)-(23), one can

come to a solution (14) for the penetration depth h
as a function of the real temporal variable t (for kτ
<<1):
 Thus, here we have two different cases
2
(1−chk τ ) correspondingly for the pool, which is denser than a
 3 + ρ2 / 1 3+ρ2 / 1
2
shk τ
h =  h0chk τ e α , (29) jet and for the inverse situation. Peculiarities of a jet
 2  penetration to the pool are different for these two
where ch, sh denote the hyperbolic cosine and sine, cases.
respectively, e kτ is expressed through t by (27). 3 Analysis of the solution obtained for
The velocity of a jet penetration into a pool is some limit cases
determined from (14) or (29) using the equations
Further analysis of the analytical solution obtained
v1 = dh / dt = ( dh / dτ )dτ / dt . In the cases of ρ2 / 1 is easier performed for the limit cases when the
<1 and ρ2 / 1 >1, it results in solution is substantially simplified.
If ρ2 / 1 <<1, (1- ρ2 / 1 )h0<<Fr, then (12), (23),
2
−1 (26), (28) result in the following solution:
dh  3 + ρ 2 /1  3+ ρ2 / 1 1
= ( 3 + ρ 2 /1 ) 3+ ρ ⋅
dt  2 
2 /1

(30) H H
c1e kτ + c2e −. kτ
h0 ≈ , t0 ≈ H , α ≈ 2.85 ,
2 Fr
3+ ρ 2 / 1
⋅k (c1e kτ − c2 e −.kτ )e (34)
α
 t 
2
−1
e kτ ≈  + 1  ,
dh  3 + ρ 2 /1  3+ ρ 2 / 1 1 H 
= ( 3 + ρ 2 /1 ) 3+ ρ ⋅
dt  2 
2 /1

(31) which can be easily analysed. Here should be noted

c1' cos kτ + c2' sin kτ
that this approximation satisfies a wide range of
3+ ρ 2 / 1
⋅k (c2' cos kτ − c1' sin kτ )e parameters because many practical situations
correspond to the large Froude numbers.
correspondingly. Accounting (16), (24), from (29), (30) yields the
following approximate solution for the depth of a jet
penetration, as well as for its velocity and
Parameters of a jet penetrating a pool acceleration:
The equations obtained, e.g. (29)-(31), allow
2 / 3(1−chk τ ) chk τ 1 Fr
computing the parameters of the jet penetrating the 3  H  shk τ

pool. For example, the penetration depth h* is h =    e 1, 43 H

,
2  2 
determined by condition dh/dt=0, therefore h* and
the correspondent penetration time t*, for ρ2 / 1 <1 −1
dh h H  t 
and ρ2 / 1 >1 are computed as v1 = = 2.85  + 1 ⋅
dt H Fr  H 
2
4 ,
ρ2 / 1 <1,  3 + ρ2 / 1  3+ρ2 / 1 3+ρ2 / 1 c1c2
,  H  2  2 / 3
h* =   e 1 Fr
 2  ⋅{ln     shkτ + chkτ }
(32)  2  3   1.43 H
1 (35)
( 3 + ρ 2 /1 ) 1− 2 c1c 2
3+ ρ 2 / 1 2 −1
t* = e 3+ ρ 2 / 1 - t 0 ; d h 1 H  t 
a1 = = 2.85  + 1 ⋅
k ( c1 − c2 ) dt 2
H Fr  H 

2
c1' + ( c2' ) 2 / c1'
 3 + ρ 2 / 1  3+ ρ 2 / 1 2 H 2
2/ 3
 1 Fr
ρ2 / 1 >1, h* =   e 3+ ρ 2 / 1 , ⋅({ln     shkτ + chkτ} ⋅
 2   2 3  1.43 H
(33)
1
( 3 + ρ 2 /1 ) 1−
3+ ρ 2 / 1
c1' + ( c 2' ) 2 / c1'
3+ ρ 2 / 1
t* = '
e − t0 .
kc 2
  dh h  t −1
  h H t
1

−
H t 
2

⋅ −  1 +  +
2.85  1 + ⋅ h≈  + 1 ,
2 H 
 dt H  H   H Fr H 
,  H  2 2 / 3   t
dh H   t 
H  2  2/3
 ≈ {4,06 ln     ln  +1 +1} +1
1 Fr dt Fr 
⋅{ln     τ
chk + shk τ })  2 3   H  H 
 2  3   1.43 H ,
(37)
or, with explicit expression for dh/dt, (35) is d 2h  t
1  dh 1 
−1

2 / 3 chkτ ≈ 2{
  + 1 − +
 3   H 2  
2/3 1 Fr − 1
2.85 H t  2
dt  H  dt 2 
H
shkτ
v1 =      
1,43 H
e 1 +  ⋅
 2   2 3   H Fr H  
1 H  2  t  
2/3 2

+ ln     + 1  .
Fr  2  3   H  
 H  2 2 / 3  1 Fr
⋅ {ln    shkτ + chk τ }
 2  3   1.43 H
With an order of the term

where are:
[
ln H / 2( 2 / 3) 2/3
]
ln( t / H + 1) restricted by 1, a
further simplification is as follows

dh t
1  t 
2.85 H / Fr
 t 
−2.85 H / Fr  ≈ +1 ,
chk τ ≈  +1 + +1  dt H
2  H   H  
 
, d 2h 1 2  H  2 2 / 3  t   1
2

(36) ≈ + ln     +1  ≈
dt 2 H Fr  2  3   H   H
.
1  t 
2.85 H / Fr −2.85 H / Fr
  t 
shk τ ≈  +1 − +1 
2  H   H   Here H~1 or H>>1 were considered because by
 
. H<<1 there is actually no jet (a length of a jet
supposed to be at least larger than its diameter). But
this case might be also considered using the solution
3.1 Influence of the Froude number and the obtained.
initial length of a jet
Analysis of the expressions (35), (36) shows the
solution dependence on parameters H / Fr , 3.2 The case of a long finite jet or a jet
t / H . A key feature of a jet penetration is coming from a nozzle
determined by the Froude number and initial jet The case of H>>1 is considered separately due to its
length, e.g. for H / Fr <<1: most practicality. It corresponds to a long jet or to a
jet coming from the nozzle. For this case, the
H
equation (11) yields
2.85
 t  H  t 
+1 ,
Fr
 +1 ≈1 + 2.85 ln 
H  Fr H  ( h0 ρ 2 / 1 / H + 1) 2 = 1 / u 2p ,
(38)
H  t  (1 + ρ2 / 1 )u 2p =1 + 2h0 (1 − ρ2 / 1 ) / Fr ,
shk τ ≈ 2.85 ln  +1 , chk τ ≈ 1 ,
Fr  H 

up to a limit t / H ~1 and even higher. For example, where u p ≈ 1, and the last equation (38) gives the
10 0,1 ≈1,23 , 1000 0,1 ≈ 2 , therefore the
approximate initial depth of a jet penetration:
approximations used here satisfy a wide range of the
varying parameters. By such assumptions, ρ2 /1
linearization of the solution (35) by the parameter h0 = Fr . (39)
H / Fr yields 2(1 − ρ 2 / 1 )
But the formula (39) according to (38) is justified
only for ρ2 / 1h0 << 1 , therefore H>> ρ2 / 1 Fr / 2 is
2
The equations (41) yield for t << ρ2 / 1 Fr the
required. For example, if ρ2 / 1 = 0.1 , and Fr=102, following approximations:
then H>>0.5 has to be, and h0 ≈ 5 , up ≈ 1. By
shk τ ≈ 2t /( ρ 23//12 / Fr 2 ) , chk τ ≈ 1 ,
Fr=104, there are H>>50, and h~500, respectively.
Thus, the assumption made is reasonable.
It should be noted that this case is absolutely therefore solution of the problem in a form (40)
different from the case considered in [1, 5, 6, 13]. goes to the following simplified expressions:
t 1
h ≈ h0 + , v1 ≈ . (42)
3.3 Parameters of the jet’s penetration into ρ2 / 1 Fr ρ2 / 1 Fr
the pool
The formula (39) expresses h0 through two Analysis of the simple partial limit solution (42)
parameters, the density ratio and the Froude shows that at the beginning of the jet penetration,
number, e.g. h0 does not depend on H. Substitution the depth of penetration is a linear function of time,
of (39) into (23), (26)-(29) results for ρ2 / 1 <<1 in and the velocity of penetration is nearly constant
being inversely proportional to the density ratio and
the following:
to the Froude number.
chk τ 1
3
2/3
 2 2 / 3 ρ  ρ2 / 1
shk τ
h ≈    2 /1
Fr  e ,
2  3 
 2 
 3.4 The approximate solution for the
1
−    2
ρττ extended time interval

ρ
2h  
ρ ρ
dh t  
2 chk  3
≈ 
 +
1ln
  2 /1
 Fr
shk + Similar approximation for the extended time
dt Fr  Fr  

3 
2  t >> ρ2 / 1 Fr is the following:
  2 /1 
2 /1 2 / 1

1 t 2 ρ2 /1
−2  
d 2h 4h  t  3
2/3
 2  2 / 3 ρ 2 /1  2  ρ2 /1Fr 
≈  + 1 ⋅ h≈    Fr  ⋅
dt 2
ρ 2 /1Fr 2  ρ 2 /1Fr  2  3  2 
(43)
(40)
2
  2  2/ 3 ρ  chkτ  1  t 
 
⋅ ln   2 /1
Fr  ⋅ shkτ +  + ⋅e
ρ 2 / 1  ρ2 /1 Fr  2 ρ2 / 1

  3  2  ρ 2 /1 

with the depth of a jet penetration growing in time.

 2
3 ρ

  2  shkτ Analysis of the solution (40) reveals an
+ ln   2 /1
Fr  chkτ + + interesting feature with a jet velocity, which can be
 3  2  ρ 2/1
  decreased if and only if

  2
   2  2 / 3 ρ 2 / 1 
1   2  3 ρ 2 /1  chkτ  ln   Fr  < 0 , ⇒
− ln   Fr shkτ +  .  3  2 
2 ρ 2 /1   3  2  ρ 2/1 
    2 3
2/3
2,62
Fr <   ≈ .
ρ2 /1  2  ρ2 /1
where are:
(44)
ρ 2 /1
t0 = 2h0 = ρ2 / 1 Fr , h0 ≈ Fr , The condition (44) is necessary but not
2 satisfactory. Actually one needs to know when the
(41) jet acceleration is negative. A full penetration is
 t 
2 ρ2 / 1
determined by the condition of v1 = 0 , where from
e kτ ≈ 
 ρ Fr + 1
 .
 2 /1 
  2  2 / 3 ρ 2 / 1  chk τ * ρ 1/ 2
shk τ * ln  
 3  2
Fr  = −
 ρ2 /1
,
h* ≈ 0.5 Frρ 2 / 1 e
−
[
ln ( 2 / 3 ) 2 / 3 0.5 ρ 2 / 1 Fr ].
(47)
with a time and a depth of penetration, τ* , h* ,
respectively.
Solving this equation with (41) yields 4 Peculiarities of the jet penetration by
different parameters
chkτ *
−
chkτ *
 3
2/3
 2  2/3
ρ2 / 1 
 ρ 
ρ 2 / 1 ln  ( 2 / 3 ) 2 / 3 2 / 1 Fr  4.1 Accelerating jet ( a1 > 0 )
2
h* =     Fr  e  
By t >> ρ2 / 1Fr , a simple condition for a1 > 0
 2  3  2 
(positive acceleration of a jet, velocity is growing)
, follows from (40). Due to the correlations
 2 2 / 3 ρ  chk τ ≈ shk τ ≈ 0.5e kτ , the condition a1 < 0
−1
γ = − ρ2 / 1 ln   2 /1
Fr  , results as:
 3  2 

(45) (
ln 2 β + 1 + 1.5 / ρ 2 /1 ln β + )
2 ρ2 /1 − 2 2 /1ρ
 t*   t*  , (48)
 +1  −  1+  =
 ρ 2 /1Fr  2 /1ρFr  (
+1/ ρ 2 /1 1+ 0.5 / ρ 2 /1 < 0 )
,
 t 2 ρ2 /1
 t
− 2 2 /1ρ
  where β = ( 2 / 3) 2 / 3 0.5 ρ 2 / 1 Fr . The solution of the

=γ  *
1+  + * 1 +  
  ρ 2 /1Fr equation (48) is the following:
  2 /1ρFr  

− 1 − 0 .5 / ρ2 / 1 < ln β < −1 / ρ2 / 1 , ⇒
and further goes for the penetration time: (49)
2( 3 / 2 ) < Fr < 2( 3 / 2)
2/3 −1 / ρ 2 /1 2/3 −1−1 / ρ 2 /1
/ ρ 2 / 1e / ρ 2 / 1e
1 + γ 
0.25 / ρ2 / 1 
t * =   −1ρ2 / 1 Fr , .
1 −γ  
  
(46) For the density ratio ρ2 / 1 = 0.1 , from the
2 ρ2 / 1 0.5 inequalities (49) yields approximately:
 t*  1+γ 
 + 1 =   ,
 ρ 2 / 1 Fr  1−γ 
F r∈ (1.07; 1.95).
  1−
0.5
ρ2/1 ln (2/ 3 ) 0.5 Fr
2/3
ρ 2/1
    + By t >> ρ2 / 1 Fr there is a narrow range of the
  1+
ρ2/1 ln (2/ 3 ) 0.5 Fr
2/3
ρ 2/1 Froude numbers where a jet velocity may decrease.
1     
chk τ* =   −0.5
Normally velocity is growing in time if the density
2   1−
ρ2/1 ln (2/ 3 ) 0.5 Fr ratio is small because the gravity force exceeds the
2/3
ρ 2/1
    drag force.
 ρ2/1 ln (2/ 3 ) 0.5
2/3
ρ 2/1 Fr 1+

     
If 4.1.1 Condition for the jet’s velocity decrease
In general, the condition of velocity decrease
follows from (40):
( 2 / 3) 2 / 3 0.5 ρ2 / 1Fr <<1 ⇒
[
− ln ( 2 / 3)
2/3
0.5 ρ 2 / 1 Fr >>1, ] c h kτ * ≈ 1, then (
A2 (τ ) − 2 − 1/ ρ 2/1 e kτ A(τ )+ )
, (50)
from (45) yields: (
−4 1/ ρ 2/1 − ln β e kτ < 0 )
 (
A(τ ) = ln β + 1/ ρ 2 /1 e 2 kτ+ ) Both conditions (55) must be satisfied
simultaneously (not separately!). The first one in
. (51)
( )
+ 2 − 1/ ρ2/1 ekτ + 1/ ρ 2 /1− lnβ
case of

ln β + 1 / ρ2 / 1 > 0 , (57)
Solving the quadratic inequality (50) for the
function A(τ) results in
which corresponds to the left side of (49), gives the
A1 (τ ) < A(τ ) < A2 (τ ) , following two solutions:
(52)
e kτ ≤ B1 , e k τ ≥ B2 ,
where the limits of the interval are: (58)

( ) − ( γ 1 + γ 2 ) m ( γ 1 + γ 2 ) + 4( ln 2 β − ρ 1 / 2 )
2
A1,2 = 1 − 0.5 / ρ 2 /1 ekτ +
B1, 2 =
(53)
(
2 ln β + ρ 1 / 2 ) .

( ) ( )
2
m 1 − 0.5 / ρ 2 /1 + 4 1/ ρ2 /1 − ln β e kτ As far as in (58) B1 < 0 is, only the second
solution supposed to be real. Similarly, the other
Required A1, 2 (τ) be the real functions, with
inequality in (55) has the following solution:
account of (53) and (49), after some simple
transformations, one can get the following condition D1 ≤ e kτ ≤ D2 ,
for the Froude number: (59)

Fr ≤ 2( 3 / 2 )
2/3
e
(
0.25 1+3 / ρ 2 / 1 + 0.25 / ρ 2 / 1 ) /ρ . ( γ 2 − γ 1 ) m ( γ 2 − γ 1 ) 2 + 4( ln 2 β − ρ 1/ 2 )
D1,2 =
( ) .
2 /1
(54) 2 ln β + ρ 1/ 2

Thus, for ρ2 / 1 = 0.1 , the condition (54) gives

Fr ≤ 785 . Then, putting (51), (53) into (52) 4.1.2 Conditions for the Froude number
yields: When ln β > ρ1 / 2 , both B1, 2 and D1, 2 are
the real values. And this is the sufficient but not the
necessary condition. It is satisfied by small, as well
( ln β +1/ )
ρ 2 /1 e 2 kτ + (γ 1+ γ 2 ) ekτ + as by large values of the Froude number:
,
Fr > 2( 3 / 2 )
2 / 3 1 / ρ2 / 1
e / ρ 2 /1 , or
+ 1/ ρ 2/1 − ln β ≥ 0
(60)
(55)
Fr < 2( 3 / 2)
2/3 −1 / ρ 2 / 1
e / ρ 2 /1 .

( ln β +1/ )
ρ 2 /1 e 2 kτ + (γ 1− γ 2 ) ekτ +
For ρ2 / 1 << 1 considered here, γ1 < 0 ,
,
therefore it goes to ( γ 1 + γ 2 ) < ( γ 2 − γ 1 ) . When
2 2

+ 1/ ρ2 /1 − ln β ≤ 0 D2 is real value, B2 is always real. That is why

more simple condition than (60) is considered when
where are: D2 is real value: ( γ 1 + γ 2 ) 2 + 4(ln 2 β − ρ 1 / 2 ) ≥ 0 .
Then it goes to the simpler condition than (60):
γ 1 = 1 − 0.5 / ρ2 / 1 ,
(56)
γ2 = (1 − 0.5 / ρ2 / 1 ) 2
+4 1/ ( ρ 2 / 1 − ln β . )
 Due to D1 < 0 , the solution (60) changes to the
2(ln β − ln β + 0.5
2
1/ 2ρ −0.8751/ 2 ρ +
0.5) ≥ following one: 0 < e kτ ≤ D2 , where the left side is
always satisfied. Therefore the common solution
.
(56) yields: B2 ≤ e kτ ≤ D2 , where from with
( ) ( ) ( 4+ )
2
≥ 0.5 ρ1/ 2 −1 −
1 0.5 ρ
1/ 2 ρ ln − β
1/ 2 account of (57)-(60) and the last correlation of (41),
as well as the expressions for β from above, one
can compute the temporal interval t1 ≤ t ≤ t 2
For ρ1 / 2 ≥ 4 the right side of the inequality is
corresponding to the case of a jet velocity decrease (
positive. The left side is positive if
a1 < 0 , decelerating jet flow):
ln 2 β − ln β + 0.5 ρ1 / 2 − 0.875 ρ1 / 2 + 0.5 ≥ 0
,

where from following ln β < ( ln β ) 1 or { +1

t1 = ρ2/1 Fr −
ln β > ( ln β ) 2 , 0.5 ρ1/ 2 
 γ −γ + γ −γ +2 4 β
 ( 2 1) (2 1 ) ( ln −2 ρ 

1/ 2 ) 
,
+ 
(ln β)1, 2 = 0.5 m 0.875 ρ1 / 2 −0.5 ρ1 / 2 −0.25 , 

(
2 ln β + ρ 1/ 2 ) 




which is real by ρ1 / 2 ≥4. Therefore, taking into (62)
account the previous condition ln β > − ρ1 / 2 , one t2 = ρ2 /1 Fr { −1 +
can come to the requirements: 0.5 ρ1 / 2 
 ( 2 1) ( 2 1) (
 − γ + γ + γ + γ 2 + 4 ln 2 β − ρ
1/ 2 )   .
ln β <0.5 − 0.875 ρ1 / 2 −0.5 ρ1 / 2 −0.25 , 

 (
2 ln β + ρ1/ 2 ) 
 

or
ln β >0.5 + 0.875 ρ1 / 2 −0.5 ρ1 / 2 −0.25 ,

where from:
4.2 The bifurcation points of the jet
The non-linear solution thus obtained is an exact
( 7 / 36) 2 ≈ 0,04 ≤ ρ 2 / 1 ≤ 0,25 , analytical solution for a solid rod penetration into
the pool and for some initial part of a jet penetration
(61) before remarkable growing of its radius. It might be
Fr < 2 ρ1 / 2 (1.5)
0.5 − 0.875 ρ1/ 2 −0.5 ρ1/ 2 −0.25
2/3
e , used as approximate step-by-step solution for a jet
penetration into a pool for small temporal intervals
correcting the jet radius from one to another one.
or Therefore it is crucial to estimate an evolution of the
jet’s radius to get an idea how to correct solution
Fr > 2 ρ1 / 2 (1.5) 2 / 3 e
0.5 + 0.875 ρ1/ 2 −0.5 ρ1/ 2 −0.25
aiming at good correspondence with the
. experimental data. With this purpose, the Bernoulli
equation and the mass conservation equation are
For ρ2 / 1 = 0.1 , from (61) yields solution considered for the jet in the following form:
Fr <~ 3,12 , or Fr >~ 600 . Comparing the last
condition with the request of real values A1, 2 , one [ ]
S1 ( ρ 1 − ρ 2 ) hg + 0,5 ρ 1v 12 = 0,5 ρ 1u02 S 0 ,
can get: 600 < Fr < 785 . It is very narrow gap by
the Froude numbers (except the low Froude ρ1v1 S1 = ρ1u0 S 0 ,
numbers) when the velocity decreases with time.
When (58) is not satisfied, the case is not
interesting because it requires too small Froude where S is the area of the jet’s cross section. Index 0
numbers determined by the last condition (61), e.g., denotes the initial state while the index 1 denotes
for ρ2 / 1 = 0.1 there is Fr <~ 1 . some current state afterwards.

4.2.1 dimensionless conservation equations

In a dimensionless form, retaining the same Fr 1
symbols: h1 = = ,
8(1 − ρ 2/1) 8 Ri

[ ]
S 1 2h(1 − ρ 2 / 1 ) / Fr + v 12 = 1 , when further existence of the two possible jet’s
(63) radiuses is impossible. Here Ri is the Richardson
S1v 1 = 1 . number (the ratio between the momentum and
buoyancy forces of a jet).
Substituting S1 = 2 into the last expression (63)
The equation array (63) has the following
solution: gives v1 = 0,5 . The jet is going from h = h0 to
1
h1 = and during this time its radius is growing
S1 =
Fr
[
4h(1 − ρ 2 / 1 )
]
1 ± 1 − 8h(1 − ρ 2 / 1 ) / Fr ,
8Ri
from 1 to r1 = 2 , when the jet velocity becomes
(64) v1 = 0,5 , e.g. for the density ratio 0.1 the total
v1 =1 / S1 . depth of a jet penetration into a pool up to this point
is computed as

4.2.2 Bifurcation point h0 + h1 ≈ 5,5 + 13,9 ≈ 19 ,4 .

There are two possible solutions for the jet radius
with the point of bifurcation: From the equation (64) a jet cross-section at the
depth of penetration of h = h0 is as follows:
Fr
h= .
8(1 − ρ2 / 1 ) S1 = 0,5ρ1 / 2 (1 ± 1 − 4 ρ2 / 1 , ),
(65)
After this point the solution (64) does not exist
anymore in real numbers, therefore the jet can where from for the density ratio 0.1 follows
change its solution abruptly between these two
available solutions. S1 ≈ 1,15 , r1 ≈ 1,07 , v1 ≈ 0,87 ,
The jet starts penetration into the pool with initial
or
cross-sections, thus, S1 = 1 . Analysing the (66)
equation (64) one can note that for a small S1 ≈ 8,87 , r1 ≈ 2,98 , v1 ≈ 0,11 ,
penetration depth or, more common,
8h (1 − ρ2 / 1 ) << Fr , it goes to:
so that the first set of parameters (66) is close to the
Fr assumptions made above, while the other set of
S1 ≈ 1 or S1 ≈ >>1. parameters is a possible solution, which may occur
2h (1 − ρ2 / 1 )
abruptly at the point of bifurcation h = h1 due to an
instability of the jet when any regular solution, as it
There is no reason for a jet to become abruptly
is shown by (64), does not exist.
from the section area 1 to the bigger one because the
jet momentum directs mainly along its axis. But
4.2.4 Basic features of a jet penetration into a
further on, due to instability causing by the free
pool
surface perturbations and due to a loss of
The phenomenon of a jet penetration in a pool
momentum, the jet area may change at any moment.
accounting the results obtained and the experimental
Strictly saying, it requires complete instability
data presented in Fig. 1 seems to be as follows. First
and bifurcation analysis, therefore it is a subject of a
separate paper. Here only some estimation has been a jet penetrates into a pool at the distance h0
done for the moment. determined by the initial length of a jet, the Froude
number and the density ratio. In case of a long jet
4.2.3 Parameters of a jet with bifurcation (as well as the jet permanently spreading out of the
From S1 = 1 the jet should become to S1 = 2 at nozzle) the initial penetration length is determined
by the Froude number and the density ratio.
the point
 Then jet is going with a slight increase of its
radius till h1 , which represents the bifurcation
point.. After this bifurcation point, the jet is sharply
2
enlarged and goes further with a nearly constant 5x10
Finite Jet
radius. Applying the solution obtained to those parts ρp/ρj=9.4 ρp/ρj=1.9
with their own initial data, the whole jet might be 2 H=5 ;
10
computed based on the analytical solution got here. H=25;
H=50;

Lp/Dj
5 Correspondence of the model to 10
1

experimental data
To validate the model developed and the analytical
solution obtained, the computed penetration depth 0
10 1 2 3
of a jet had been compared to experimental data 10 10 10
from the literature. Fr
The maximum penetration depth h from the non-
linear analytical model for a continuous jet and for a
finite jet of the length H compared to the Fig. 4. Maximum penetration depth h by the
experimental data [8, 17] are given in Fig. 3 and non- linear analytical model vs experimental
Fig. 4, correspondingly. data [8, 17] for finite jet
The dark bands (trust regions) in the Figs 3, 4
include the region between the upper line
corresponding to the experimental data [8] and the Although the idealistic assumptions were
bottom line corresponding to the experimental data employed in the analytical model, for the continuous
of the work [17]. jet, the solution showed reasonable match with the
experimental data until the Froude numbers up to
5.1 The results by the model without account 300, in the wide range of the density ratio (up to ten
of bifurcation point times).
The data obtained by model for the continuous jet However, the solution strongly overpredicted the
are presented in Fig. 3, while the data by the finite penetration depth after the Froude numbers over 300
jet are drawn in Fig. 4. First of all Fig. 3 illustrates (approximately), the higher density ratio was, the
that the penetration depth increases with a decrease more inconsistency with experimental data was
of the pool-to-jet density ratio. observed. For instance for the density ratio ρ 2 /1 =1.9
the results by the model obtained were out of the
trust region approximately at Fr=100 while for the
h/2 ρ 2 /1 =9.4 the results by the model leaved the trust
2
5x10 region approximately at Fr=300.
Continuous Jet
ρ p/ρ j=9.4 The correspondence of the presented results and
2
10 ρ p/ρ j=1.9 experimental data was good despite of the model
that was not accounted for the jet radius evolution
Saito [8] with a penetration depth, which would decrease the
L p/D j

penetration depth due to increase of the drag force.

1
10 Then increase of the velocity (correspondingly
increase of the Froude number) caused the air
Present data entrainment after the velocity threshold, which also
0 was not taken into account (and might be an
10 1
10
2
10
3
10 additional reason for the jet expansion and
Fr consequently for the growing drag force). This
caused an additional shortening of a penetration
depth. Therefore an account of the above-mentioned
Fig. 3. Maximum penetration depth h by the additional factors would improve the model.
non- linear analytical model vs experimental Presently analytical solution showed good results
data [8, 17] for continuous jet important for the model validation.
 For the finite jet, the Fig. 4 shows the jet
penetration depths estimated from the analytical
solution in the terms of a jet length H and of a
density ratio. As expected, the solutions
demonstrated the increase of a penetration depth
with a jet length H*.
For ρ 2/1= 9.4, the estimated penetration depth
remained nearly constant at a longer jet. However,
the Fr-term dominated at higher Fr. As expected, a
jet penetrated deeper into a liquid pool for the lower
pool/jet density ratio ρ 2/1 ( ρ p/ ρ j).
The asymptotic solution shown in Fig. 4 took
over the prediction by the Fr-dominant penetration
when h*>>H*. It was noted that for the low Fr, the
estimated penetration depths was longer than those
estimated by the continuous jets (Fig. 3). It was
resulted from the definition of the continuous jet, Fig. 5. Maximum penetration depth h against the
i.e., H*=Fr/2. Therefore for low Fr the actual jet Richardson number by analytical model
penetration depth for the continuous jet was smaller against experimental data [3, 23].
than those for H*=25 and 50.
A dimensionless mean jet penetration depth
scaled with the jet radius can be expressed in terms 5.2 The results by the model with account of
of the Richardson number Ri : h = C / Ri b , where bifurcation point
As it was discussed before and shown in Fig. 1, real
C , b are the constants. The correlation is normally
jet is penetrating a pool similar to the peculiarities
used in the form h = C / Ri , where C=4 got by the model with account of the bifurcation
corresponds to the closest fit with the Turner’s point.
results [3]. The results of computations by the model
Fig. 5 shows the comparison of the experimental described are presented in Fig. 6, where from may
results [3, 23] with analytical solution represented be clearly observed that after Fr=100 the calculation
as a curve 1 for the ρ 2 /1 =9.4 and 2 – for ρ 2 /1 =1.9, with account of bifurcation point give much shorter
length of a jet penetration into a pool, so that by
respectively. Evidently analytical solution 1 the best
Fr=300 the difference is nearly 50%.
fitted the Turner’s data [3] and two groups of the
data [23] for the range of the Richardson numbers
approximately Ri ≈ 0.03 − 1.0 .
For Ri < 0.03 the analytical solution went far
away from the data. Another solution 2 ( ρ 2 /1 =1.9)
fitted well both experimental results [3, 23] only in
a narrow region by Richardson number around
Ri ≈ 0.004 .
For smaller Ri it overpredicted the experimental
data – the less Ri , the higher overprediction. In the
range Ri = 0.004 − 0.01 the analytical solution
differed from the experimental data [3, 23] mostly
less than 30% (only close to Ri = 0.01 it is about
70%) while the maximal measurement errors of the
experimental data exceeded 100% in some points.
Fig. 6. Comparison of the model calculations with
account of bifurcation

Another illustration of these peculiarities is given in

Fig. 7 and Fig. 8, Fig. 9 presenting the numerical
solution of the full Navier-Stokes equation array for
the jet penetrating pool and experimental data
borrowed from [17]:

Fig. 7. Jet penetration with initial velocity 4m/s:

experimental and numerical data

Numerical simulation was performed with the

computer code Casper [17]. Both data, experimental
and numerical, are presented in Figs 7-9 for the
corresponding moments of time (in ms).

Fig. 8. Experimental data by the jet penetration into

the pool: with initial velocity 4m/s, 6m/s, 9m/s,
respectively, from the top to the bottom picture

Presented data have evidently shown that the jet

penetration is really going according to the model
developed here, with one critical bifurcation point.
 when inertia, drag and buoyancy are the dominating
fluid forces [25].
Several analytical and experimental studies have
been conducted in the past to obtain the breakup
length of buoyant jet penetrating into another fluid.
Most of these studies were related to the injection of
gas jet into another fluid as in fluidized bed, e.g.
Yang and Kearins [26] and Blake et al [27].
The mechanism associated with the interaction of
jet and ambient fluid is predominant for high-
velocity laminar jets, which breakup as a result of
growing bending disturbances on the jet axis.
Theoretical studies on the dynamics of bending
disturbances of liquid jets were initiated by Weber
[28] and continued further by other investigators
(see for example [14], where is also bibliography on
the subject).
Quasi-one-dimensional equations were obtained
by Entov and Yarin [29] for an arbitrary
parameterisation of a jet and successfully applied to
predict breakup length of a buckling jet.
The objectives of the works [16, 24] were to
determine the penetration behaviours and breakup
length of a thin jet penetrating into a fluid pool.
There are evidences that the buoyant jet bending
mechanism seem to occur and eventually dominate
Fig. 9. Numerical simulation by computer code the breakup behaviour. The jet breakup behaviour
Casper [17]. and the jet breakup length was obtained and
compared to the experimental data available in the
literature.
Further investigations are needed for a
clarification of the diverse factors influencing the 6.1 Mathematical model’s formulation
features of a jet penetration into a pool in the Study of the bending jet decay during its penetration
different flow regimes in a wide range of into a pool was performed according to the scheme
parameters. shown in Fig. 10:
A general conclusion is that all experimental data
of different authors and analytical solution suffer on
restricted ranges of parameters where they are valid,
therefore presently the results cannot be generalized.

6 Bending instability and breakup

length of the thin jets
Special case of a jet penetration into a pool, namely
bending instability and breakup phenomenon of a
thin liquid jet penetrating into another fluid has been
studied in [16, 24].
Although remarkable progress has been made in
explaining the effects of surface tension (Weber
number) and viscosity (Reynolds number) on jet
breakup behaviour, little attention has been given to
the thin jets (i.e. the ratio of the characteristic
transverse size to the longitudinal is small)
penetrating into pool of fluid. Especially, in the case
Fig. 10. Scheme of a bending jet in a pool
 The flow of fluid jet into another fluid can be liquid particle) in the jet is determined by three
characterized by a set of dimensionless numbers. parameters: y, z, and s, which serve as the
The density ratio ( ρ21 = ρ2 / ρ1 ) is important as it coordinates in a moving frame of curvilinear (non-
determines the penetration rate of the jet head and orthogonal in the case κ ≠ 0 ) coordinate system.
plays an important role in the jet instability (here
ρ2 , ρ1 is the density of the fluid pool and the jet, 6.2 Modelling of the bending jet’s
respectively). perturbations
In the case of large Reynolds and Weber The distributed forces qi and gi are calculated and
numbers, the inertia force is dominating compared applied to the axis in a similar way as given in
to viscosity and capillary forces. For a jet references [28, 29]. The jet’s axes are parametrized
discharging vertically under the influence of gravity and the equations along the axes are written in the
into a fluid pool, the density ratio and Froude following form
number may be dominant parameters, which
determine the jet instability. In such condition the η = H ( s, t ) , ζ = Z ( s, t ) ,
moment and moment of momentum equations for (69)
the jet can be written on the following form [28,
29]: where H and Z are the displacements of the axis in
the directions O1η and O1ζ , respectively at
ρ1 f 0∂Vn / ∂t = ∂Qn / ∂s −κQb + qn + ( ρ1 − ρ2 ) g n , ξ =s.
Only projection of the momentum and the
ρ1 f 0∂Vb / ∂t = ∂Qb / ∂s + κQn + qb + ( ρ1 − ρ2 ) gb moment-of-momentum equations to the normal and
, binormal axis are retained from equation (67), i.e.
only the bending perturbations are considered.
ρ1I ∂ / ∂t (∂Vb / ∂s + κVn ) = −∂M n / ∂s + Those equations describe small bending
(67) perturbations of the liquid jet penetrating into the
+κ M b + Qb + k ( ρ 2 − ρ1 ) gτ I , fluid pool, neglecting the change in jet radius. These
disturbances have a growth rate γ .
ρ1I ∂ / ∂t (∂Vn / ∂s − κVb ) = ∂M b / ∂s + A projection of equation (67) to the tangent of
the axis describes the growth of small axially
+κ M n + Qn + k ( ρ 2 − ρ1 ) g b I , symmetric disturbances of the jet [28, 29]. The
perturbations on the jet axis are considered to have
where the hydrodynamic (qi) and buoyancy (gi) following form [28]:
forces are:
H = A exp( γt ) cos( χs / a0 ) ,
qn = − ρ 2U f χ / a exp(γ t ) ⋅
2
0 0
2 2
0 (70)
1 Z = B exp( γt ) sin( χs / a0 ) ,
⋅[ A2 cos 2 ( χ s / a0 ) + B 2 sin 2 ( χ s / a0 )] ,
2

where the curvature and torsion are respectively:

g n = f 0 g χ / a0 exp(γ t ) ⋅
1 (68) k = χ 2 / a02 exp(γ t )[ A2 cos 2 ( χ s / a0 ) +
⋅[ A2 sin 2 ( χ s / a0 ) + B 2 cos 2 ( χ s / a0 )] ,
2
1 , (71)
+ B 2 sin 2 ( χ s / a0 )] 2
g b = qb = 0, χ = 2πa0 / l .
1

Here, ao and Uo are the initial radius and velocity κ = χ / a0 AB[ A2 cos 2 ( χs / a0 ) + B 2 sin 2 ( χs / a0 )] 2
of the jet, respectively, f 0 = πa 0 is the area and
2

The bending jet perturbation and the moment Mb can

I = 1/ 4πa04 is the moment of inertia of the jet.
be written as:
The variables Q , M and χ represent the
shearing force, the moment of stresses in the jet H2 +Z2 ,
cross-section and the jet perturbation length,
respectively. The variable k is the curvature and κ is
M b = I [ ρ2U 02k + ( ρ2 − ρ1 ) g H s2 + Z s2 ] ,
the torsion of the jet. The position of a point (or a
 γ 2 = ( ρ21 − 1) g ( χ / a0 )( B / A) .
where are: (75)
H s = −Aχ / a0 exp( γt ) sin( χs / a0 ) ,
Equation (75) shows that bending jet instability is
Z s = Bχ / a0 exp( γt ) cos( χs / a0 ) . observed if and only if the density of the pool is
larger than density of the jet ( ρ21 >1).
The velocity normal and binormal to the jet axis Assuming that the bending jet perturbations are
can be obtained by differentiating (70) with respect of the order of few diameters of the jet:
to a time: Z 2 + H 2 = ∆a0 , for long wave perturbations
cos( χs / a0 ) ≈ 1 we obtain γ *t* = ln( ∆a0 / A) ,
Vn = H t2 + Z t2 = −γ exp(γ t ) ⋅
t* = 1/ γ * ln( ∆a0 / A) . The jet breakup length is
1 , (72)
then,
⋅[ A2 cos 2 ( χ s / a0 ) + B 2 sin 2 ( χ s / a0 )] 2
∆a0 U0
Vb = 0 . L* / a0 = U 0t* = ln( )
A a0 B / A( ρ21 − 1) gχ* / a0
Substituting (68)-(72) into the equation array ,
(67), leads to the following set of equations for
small perturbations on the thin jet: which can be written as

Qn = ρ1I∂2Vn / ∂s∂t − ∂M b / ∂s , L* / a0 = δ1 Fr / δ2 ( ρ21 −1) . (76)

Qb = ρ1I∂(κVn ) / ∂t −κVb , Here Fr = U 0 /( ga 0 ) is the Froude number,

2

δ1 = ln( ∆a0 / A) is a constant, which depends on

∂Qb / ∂s = −κQn , (73)
the initial perturbation level, δ 2 = χ* B / A is a
constant with regard to the wavelength and initial
ρ1 f 0∂Vn / ∂t = ∂Qn / ∂s −κQb + qn + g n ( ρ1 − ρ2 )
perturbation level of the two coordinates in the
. plane perpendicular to the jet curvilinear axis.
Now, taking into account the terms of second
Here in this case the momentum normal to the order in χ in equation (75), the growth rate γ of
axis, Mn, is assumed to be negligible. Inserting Qn the bending jet perturbations is got as follows:
and Qb from the first two equations of (73) into the
third and fourth, makes third equation an identity
and the forth equation for the moment of momentum 4U 0 χ B ρ 21 − 1
γ =± ρ 21 + , (77)
in the normal direction becomes a0 4 + χ 2 A χFr

ρ1I ∂ 2 (κVn ) / ∂s∂t − ∂ (κ M b ) / ∂s + where positive growth rate γ corresponds to a jet

(74) instability mode and negative to a stable
perturbation mode.
+ ρ1κ I ∂ Vn / ∂s∂t − κ∂M b / ∂s = 0.
2
From the equation (77) the optimal bending
perturbation length can be found, which determines
the jet’s decay on the fragments. Differentiating the
6.3 Bending instability analysis expression (77) as a function of the jet’s
Stability of the fluid jet due to linear perturbations perturbation length χ yields to the extreme point
on the jet’s axis was examined and a jet breakup at
length was derived that relates the jet coherent
length with dimensionless parameters.
Fr B Fr A 1
Substituting equation (71), (72) and Mb into (74) χ* = 4[ + ( )2 ( )2 + ] ,
and retaining only terms of first order in χ , the 1 − ρ12 A 1 − ρ12 B 4
following equation for the perturbation growth rate (78)
is obtained:
where ρ12 = 1/ ρ21 ( χ* is the most unstable L*
and it follows for ρ21 >> 1 : = δ 2 ρ12 Fr
perturbation length). Equation (78) was obtained by a0
making the following assumptions:
δ1 A
ρ 21 − 1 B where δ 2 = . The other limiting case, (
U 0 ≠ 0, χ* ≠ 0, ρ 21 + ≠ 0. 2 B
χ* Fr A
ρ21 −1 )~1 gives L* / a0 = δ2 Fr /( ρ21 −1) . The
length of bending jet decay is predominantly
Investigation of the equations (77) and (78)
determined by the square root of the Froude
showed that bending jet’s decay is possible in all
number. This is in good agreement with Blake et al.
cases for ρ21 > 1 . When ρ21 < 1 , the bending jet [27] and Yang and Keairns [26] and some other
perturbations are growing in time if and only if studies analyzed in [16, 24] who obtained the
ρ12 − 1 B following experimental correlation for the breakup
χ> . The other ones do not have to
Fr A length of jet penetrating into a fluidized bed:
grow with time (oscillations but not jet decay). Here
it should be marked that χ is considered small. L* Fr 9.16
≈ 9.16 = , (82)
Taking into account the above, the jet breakup a0 1 − ρ21 Ri
length can be obtained on the following form:
1 − ρ 21
L* δ1 Fr where Ri = is the Richardson number.
= . Fr
a0 2 χ ρ21 χFr + ( ρ21 −1) B / A Comparing the equations (81) and (82) one can
(79) obtain the constant in the theoretical solution (81) as
shown in Figs 11, 12. In case (82) the jet decay
In case of ρ21 = 1 (equal densities of the jet and depends only on the Richardson number.
pool) a perturbation length does not exist by this
model. In this case solution of the higher order
terms in χ should be considered.
Using equations (77) and (78) for the most
unstable bending jet perturbations the jet breakup
length was obtained as

L* δ1 4 + χ*2
= .
a0 4 [ ρ21 χ* + B / A( ρ21 − 1) / Fr ]χ*
(80)

6.4 Discussion of the results by bending jet’s

instability analysis
A few limiting cases of the equations (79) and (80) Fig. 11. The results of computation by (81) against
were considered and analyzed for the governing experimental data [26]
parameters, density ratio and Froude number.
In a case of a small Froude number Fr <<1
(small jet velocity or large jet diameter) the
following estimation for an optimal perturbation
length was obtained from the equation (79), χ* ≈ 2
. In this case the jet breakup length is

L* δ1 Fr
= ,
a0 2 ( ρ21 −1) B / A + 2 ρ21 Fr
(81)
 [3] X3. J. S. Turner, Jets and plumes with negative
or reversing buoyancy, Journal of Fluid
Mechanics, Vol.26, No.4, 1966, pp. 779-792.
[4] X4. M.A. Lavrent'ev and B. V. Shabat, The
problems of hydrodynamics and their
mathematical models, Moscow, Nauka, 1973,
(In Russian).
[5] X5. J. Carleone, R. Jameson and P. C. Chou,
The tip origin of a shaped charge jet,
Propellants and Explosives, Vol.2, No.6, 1977,
pp. 126-130.
[6] X6. E. Hirsch, A formula for the shaped charge
jet breakup-time, Propellants and Explosives,
Vol.4, No.5, 1979, pp. 89-94.
Fig. 12. The results of computation by (81) against [7] X7. M.L. Corradini, B.J. Kim, M.D. Oh, Vapor
experimental data [27] explosions in light water reactors: A review of
It is important to note that well-known theory and modeling, Progress in Nuclear
experimental Saito correlation [8] is got here as one Energy, Vol.22, No.1, 1988, pp. 1-117.
limit case from the analytical solution obtained by [8] X8. M. Saito, et al. Experimental study on
mathematical model. penetration behaviors of water jet into Freon-11
and Liquid Nitrogen, ANS Proceedings, Natl.
Heat Transfer Conference, Houston, Texas,
7 Conclusion USA, July 24-27, 1988.
Analyses on the penetration phenomena of a jet into [9] X9. R.W. Cresswell, R.T. Szczepura,
another liquid with various densities at isothermal Experimental investigation into a turbulent jet
condition were performed and compared with the with negative buoyancy, Physics of Fluids, A5,
data from literature. No.11, 1993, pp. 86-107.
The non-linear analytical models for the [10] X10. M. Epstein, H.K. Fauske, Steam film
continuous and finite jets to predict the maximum instability and the mixing of core-melt jets and
penetration of the plunging jet were developed and water. ANS proceedings. National heat transfer
reasonably described the characteristics of the conference. Aug. 4-7. Denver, Colorado, 1985.
penetration behaviours. [11] X11. D. F. Fletcher, The particle distribution of
The general behaviours of a jet consisted of a solid melt debris from molten fuel-coolant
surface cavity of a pool liquid by the initial interaction experiments. Nuclear Engineering
mechanical impact of the jet, air pocket formation and Design, No.105, 1988.
during the penetration, radial bottom spreading of [12] X12. D.F. Fletcher, A. Thyagaraja, The
the jet and entrained air and interfacial instability
CHYMES mixing model, Progress in Nuclear
between the pool liquid and entrained air must be
Energy, Vol.26, No.1, 1991, pp. 3161.
taken into account to further improvement of the
[13] X13. S.A. Kinelovsky and K.K. Maevsky, On
model.
the cumulative jet penetration into hard plate,
Presently, an analytical solution obtained was
Journal of Applied Mathematics and Technical
accurate for the solid rod penetration into a liquid
Physics, No.2, 1989, pp. 97-105 (In Russian).
pool and is approximate for the jet penetration into a
[14] X14. A.L. Yarin, Free liquid jets and films:
pool of other liquid.
hydrodynamics and rheology, Longman
Scientific & Technical, Haifa, 1993.
References:
[1] X1. R. J. Eichelberger, Experimental test of the [15] X15. D. Goldman, Y. Jaluria, Effect of
opposing buoyancy on the flow in free and wall
theory of penetration by metallic jets, Journal
jets, Journal of Fluid Mechanics, Vol.166,
of Applied Physics, Vol.27, No.1, 1956, pp. 63-
1986, pp. 41-56.
68.
[2] X2. M. I. Gurevich, The theory of ideal liquid [16] X16. H.O. Haraldsson, I.V. Kazachkov, T.N.
Dinh and B.R. Sehgal, Analysis of thin jet
jets, Moscow: Nauka, 1961 (In Russian).
breakup depth in immiscible fluids, Abstracts
of the 3rd International Conference on
 Advances in Fluid Mechanics, 24-26 May Leschziner and Sommerfeld, eds.), KTH
Montreal, Canada, 2000. Stockholm, Sweden, June 27-29, 2, 2001.
[17] X17. H.S. Park, I.V. Kazachkov, B.R. Sehgal, [24] X24. I.V. Kazachkov, A.H. Moghaddam,
Y. Maruyama and J. Sugimoto, Analysis of Modeling of thermal hydraulic processes
Plunging Jet Penetration into Liquid Pool in during severe accidents at nuclear power
Isothermal Conditions, ICMF 2001: Fourth plants, National Technical University of
International Conference on Multiphase Flow, Ukraine “KPI”, Kyiv, 2008 (in Russian).
New Orleans, Louisiana, U.S.A., May 27 - June [25] X25. Lin, S.P. and Reitz, R.D., Drop and spray
1, 2001. formation from a liquid jet, Annual Review of
[18] X18. F. Bonetto, D. Drew and R.T. Lahey, Jr., Fluid Mechanics, Vol. 30, 1998, pp. 85-105.
The analysis of a plunging liquid jet-air [26] X26. W.C. Yang and D.L. Keairns, in
entrainment process, Chemical Engineering Fluidization, D.L. Keairns and J.F. Davidson
Communications, Vol.130, 1994, pp. 11-29. (eds.), Cambridge Univ. Press, Cambridge,
[19] X19. A.M. Lezzi, A. Prosperetti, The stability 1978.
of an air film in a liquid flow, Journal Fluid [27] X27. T.R. Blake, H. Webb and P.B.
Mechanics, Vol.226, 1991, pp. 319-347. Sunderland, The Nondimensionalization of
[20] X20. P. Lara, Onset air entrainment for a water Equations Describing Fluidization with
jet impinging vertically on a water surface, Application to the Correlation of Jet
Chemical Engineering Sciences, Vol.34, 1979, Penetration Height, Chemical Engineering
pp. 1164-1165. Society, Vol.45, No.2, 1990, pp. 365-371.
[21] X21. G.K. Batchelor, An Introduction to Fluid [28] X28. C.Z. Weber, Zum Zerfall eines
Dynamics, New York: Cambridge University Fluessigkeitsstrahles, Zeitschrift fuer
Press, 1967. Angewandte Mathematik und Mechanik,
[22] X22. P.L. Sachdev, Non-linear ordinary Vol.11, 1931, pp. 136-154.
differential equations and their applications, [29] X29. V.M. Entov and A.L. Yarin, The
Marcel Dekker, Inc., 1991. dynamics of thin liquid jets in air, Journal of
[23] X23. J. Dahlsveen, R. Kristoffersen and L. Fluid Mechanics, Vol.140, 1984, pp. 91-111.
Saetran, Jet mixing of cryogen and water. In
Turbulence and Shear Flow Phenomena, 2nd
International Symposium (Lindborg,
Johansson, Eaton, Humphrey, Kasagi,