Atomization and Sprays, 19(12):xxxxxx, 2009

NEW IMAGINARY-VALUED SIMILARITY

SOLUTIONS FOR LAMINAR BOUNDARY
LAYERS WITH A SPRAY
R. Z. Orland and D. Katoshevski
Department of Biotechnology and Environmental Engineering, Ben-Gurion University of the
Negev, Beer-Sheva 84105, Israel

D. M. Broday
Faculty of Civil and Environmental Engineering, Technion Israel Institute of Technology, Haifa
32000, Israel
Original Manuscript Submitted: 20/10/2009; Final Draft Received: 22/10/2009
A new similarity solution for spray distribution in a boundary layer flow, involving an imaginary-valued variable,
is presented. The flow field similarity solution has three branches, the two known solutions for accelerating flow
and moderately decelerating flow and a new solution, which is the focus here, for high-rate deceleration. This last
family of solutions involves the use of imaginary expressions in the course of the similarity procedure. Although
the procedure employs imaginary expressions, the solution in the physical plane admits a fine behavior. The mass
distribution of spray droplets in the boundary layer for the latter case shows that the highest concentration is
either at the wall or at the edge of the boundary layer, depending on the nature of downstream variation of
the aerosol concentration at the outer flow and on the particular branch of the flow solution. The new similarity
solution presented here opens the possibility to describe a new range of boundary/shear layer problems. We
demonstrate that the option for describing droplet/particle distributions in such configurations has ramifications
for transport and deposition of droplets/particles in boundary layers developed in industrial facilities and sizing
instrumentation, and we suggest its possible application for cases of spray-surface interactions.

INTRODUCTION
Analytical and numerical solutions of laminar
boundary layer and shear layer flows often involve
similarity solutions, which have been extensively
discussed in the literature [14]. Blasius [5] presented for the first time such a solution for a
boundary layer with constant free-stream velocity.
Falkner and Skan [6] extended the Blasius solution to a family of flows for which the free-stream
velocity relates to the longitudinal distance along
the boundary by a power law. The dependence of
solutions on the power of the distance x, x ,
by which the free-stream velocity changes downstream, has also been studied extensively [711].

The parameter = 2/( + 1) has often supplanted in the resulting equations. For < 1,
but close to 1, admits large positive values, and
these are the cases we address here. Solutions for
large values of and for negative have been
studied [12]. However, these solutions involve an
artificial reversal of the flow direction and relate to
a different class of flows than those discussed here
(as elaborated in the model formulation section).
The use of similarity solutions to describe spray
dynamics has been demonstrated in the works of
Katoshevski and coinvestigators [2, 3, 13], who
studied the shear layer formed between two parallel streams that move at different velocities and
carry spray droplets. The velocities of the streams

Correspondence concerning this article should be addressed to D. Katoshevski, Department of Biotechnology

and Environmental Engineering, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel; e-mail:
davidk@bgu.ac.il
10445110/09/35.00
c 2009 by Begell House, Inc.

1

R. Z. ORLAND, D. KATOSHEVSKI, AND D. M. BRODAY

were considered constant or to vary as power

of the distance downstream, x . The power was
taken to be zero or close to zero, producing no or
only moderate downstream velocity variation. The
current study, as mentioned, extends the range of
values to < 1. Solutions of the pertinent
boundary layer equation in these cases require the
use of imaginary-valued variables. Stewartson [14]
conjectured that since the variable and related differential equations are imaginary-valued, there is
no point in considering this branch of solutions.
However, the analysis presented here shows that
these solutions are actually feasible in the physical sense because the imaginary values of the
variable and functions in the mathematical solution do not, in fact, entail imaginary values of the
physical properties. These new solutions, elaborated on here, open the door to an additional range
of applications that can be described by similarity solutions. The new solution for the flow field
serves as a platform for describing the spray mass
distribution in a boundary layer flow, assuming a
convective mass transport mechanism inside the
boundary layer. Broday [1] has presented an application that has some similarities to the current one:
the deposition of ultrafine particles in the upper
airways of the respiratory system, assuming a diffusional mechanism as governing mass transport.
This application, however, utilizes wedge flows,
which do not pertain to the problems described
here as wedges are limited by the condition < 2.
Wall suction is also accounted for in the current
work, which brings the particles/droplets closer to
the wall. This modification has ramifications when
modeling spray surface interactions such as nonhomogeneous surface coating.
In the course of the article, we first describe
the similarity model, and then we address the new
issue of the use of an imaginary-valued variable
and the resulting new solution for the velocity
field. This sets the platform for including a spray
through an additional, coupled similarity solution.

MODEL FORMULATION
The momentum and continuity equations for a
laminar boundary layer are, respectively,

u
u
U
2u
+v
=U
+ 2
x
y
x
y

(1)

u v
+
=0
x y

(2)

where x and y are the longitudinal and lateral

distances, respectively (with x = 0 being the leading edge of a flat plate or the tip of a wedge
in wedge flow); u and v are the longitudinal and
lateral velocities, respectively; U is the free-stream
velocity (assumed to be longitudinal); and is the
kinematic viscosity.
If we assume the relation U = Cx , where C
is a constant, the well-known Falkner-Skan transformation may be applied. We define a stream
function and a similarity variable, respectively:
2 Cx+1
=
+1


 12

f ()

(3)

 12

(4)

( + 1) C x1
=
2


Thus u and v are defined as

= U f 0 = Cx f 0
(5)
y
r



+1 Cx+1
1 0
v=
=
f+
f (6)
x
x
2

+1
u=

where prime denotes differentiation with respect to

. Substituting Eqs. (3)(6) into Eq. (1), we arrive
at the well-known Falkner-Skan equation
h

2 i
f 000 + f f 00 + 1 f 0
=0
(7)
where

2
(8)
+1
As already mentioned, we will focus on cases
with < 1( > 2), corresponding to cases in
which the similarity variable takes on imaginary
values. Previously, such cases have been regarded
as nonphysical because of the imaginary nature
of resulting functions (f , f 00 ). However, since the
values of all physical properties derived from the
solution are real, the solution is applicable. Brodie
and Banks [12] have addressed the range > 2,
but their method was based on considering reverse
=

SIMILARITY SOLUTIONS FOR LAMINAR BOUNDARY LAYERS

flows (C < 0), that is, corresponding to U < 0. Although their solution addresses the same equation
and accounts for the same range of values of ,
the nonlinear characteristics of the boundary layer
equation (7) reveal that the solutions presented here
(for U > 0) and those presented by Brodie and
Banks (for U < 0) refer to different classes of
flows.

cases in which overshoot exists are those where

< 0.1984, unless penetration into the boundary
is allowed [5]. We find that this applies also for
cases characterized by > 2 addressed here.

The physical boundary conditions normally used

for solving Eq. (7) are no slip, u(y = 0) = 0; no
penetration, v(y = 0) = 0; and a finite thickness
of the boundary layer. These conditions translate,
respectively, into

As mentioned, solutions for < 1 ( > 2)

have relied on reversal of flow direction (since
was restricted from receiving imaginary values), so
highly decelerating flows have not been considered. However, the imaginary nature of does not
necessarily entail a loss of physical applicability
as such values may still correspond to physical
values in real x y coordinate space. Since f 0
must be a positive real number that increases as
we move away from the boundary for any value
of , both f and f 00 must take on negative imaginary values at the boundary. Such values still result
in all physical properties, that is, longitudinal and
lateral velocities, shear at the boundary and the
stream function, maintaining real values. Hence we
show that the Falkner-Skan-type solution can be
extended beyond the range of parameters previously considered. Such a solution is presented in
Fig. 1 in physical coordinates, along with the wellknown Blasius similarity solution, corresponding to

= 0 : f0 = 0

(9)

=0:f =0

(10)

: f 0 = 1

(11)

The second condition constrains the possibility of

fluid penetration into the boundary, a condition
that may be formulated as f ( = 0) = f0 , where
f0 > 0 enables wall penetration, according to the
expression for the lateral velocity v (Eq. (6)) at
= 0.
A viable solution should also require that there
be no overshoot, that is, that u does not exceed U
anywhere and hence that f 0 1. The well-known

IMAGINARY-VALUED FUNCTION AND

VARIABLE

Fig. 1 Variation of the normalized longitudinal velocity f 0 = u/U with distance from boundary (y), at constant
longitudinal distance (x). For 1 < , no wall penetration was simulated, and x = 103 m. For < 1, wall
penetration was set at v = 0.3161 m/s and x = 101 m

R. Z. ORLAND, D. KATOSHEVSKI, AND D. M. BRODAY

= 0. Solutions were achieved numerically via a

Runge-Kutta-4 method.

tion (by substituting Eqs. (4)(6), and (12) into

Eq. (13)), it may be shown that
2

q = C1 f +1

EXTENSION TO SPRAY FLOW AND

DEMONSTRATION
The idea behind the solution of the flow field may
be extended to a second phase, for example, of
spray droplets. It is reasonable to assume [3, 13]
that the concentration of the liquid phase, Q, may
also be described by a similarity solution, using
a power law for the downstream concentration
behavior outside the boundary layer:
Q = q () x

(12)

where is a constant and q is the concentration in

the boundary layer normalized by maximal concentration in it. Assuming convective mass transfer as
the dominant mass transfer mechanism, a conservation equation for the liquid phase is given:
(uQ) (vQ)
+
=0
x
y

(13)

We assume that the spray droplets are small

enough to closely follow the host flow streamlines
and that the spray phase is dilute, so that dynamic
effects of the spray on the continuous fluid phase
may be neglected. Employing the flow-field solu-

(14)

where C1 is a constant. This solution for q exhibits

an interesting double branching at both = 0 and
= 1. To avoid complex values of q for <
1, C1 may, in these cases, be expressed as C1 =
C2 i[2/(+1)] , where C2 is a positive real constant.
Note that this in no way limits the applicability
of the analysis to cases with < 1, but merely
shows that such cases may be addressed. Figure 2
shows q as a function of lateral distance from the
boundary for a common negative value of and
different negative values of . Results for positive
values of would simply be the reciprocals of the
results shown. For consistency, the three solutions
presented are for the same degree of boundary
penetration (lateral velocity at the boundary, which,
as explained later, is crucial for the solution), and
thus boundary conditions for f , f 00 differ.
Unlike the case of = 0, the branch switch of
the similarity solution at = 1 does not manifest
itself in the physical plane. This is to be expected
as the branch splitting at = 0 appears at a point
of physical significance (the point between the two
branches of increasing and decreasing downstream
free-stream particle concentrations), whereas the

Fig. 2 Variation of the normalized spray concentration similarity function, q, in the physical plane for constant
values of x = 0.1 m and = 1.6

SIMILARITY SOLUTIONS FOR LAMINAR BOUNDARY LAYERS

= 1 singularity point owes more to the employed similarity variable. Figure 3 displays the
spray distribution in the physical boundary layer
by means of contours of equal normalized concentration, normalized through the maximum value.
In the case presented, the maximum value is attained at the minimum value of x. As x may not
reach zero (it is a point of singularity for ), the
minimum value of x chosen was 0.1 m. It is important to note that the new similarity solutions must
involve nonzero values of f ( = 0), denoted by
f0 , to avoid instances in which f 0 exceeds unity.
Therefore, for a solution to be applicable, we allow
the flow to penetrate the wall (through suction).
While this may be regarded as a restriction on the
applicability of the proposed solutions, such cases
may be of value for various applications.
Nonhomogeneous Surface Coating as an
Application for the New Solution
One possible application would be nonhomogeneous coating of surfaces, where droplets are
forced to approach or even deposit on a surface by
transverse flow. This application is addressed here

in a simple manner for the purpose of demonstration only. Assume a monodisperse droplet population with a diameter of 50 m (as previously used
for similar studies [13]) suspended in a laminar
boundary layer proximate to a plane surface. Deposited spray concentration on the surface would,
assuming stationary capture, be equal to the concentration of droplets at a distance from the boundary equal to (or less than) the droplet radius (usually termed the capture or interception distance).
Such a simplistic demonstrative case is described
in Fig. 4 for = 1.56 105 m2 /s, C = 1 m1 /s
(where C x is the free-stream longitudinal velocity). Free-stream spray concentrations are assumed
to qualitatively change in a way similar to the
case presented in Fig. 3. However, here has been
reduced further (values of 1, 0.5 were used, instead of the value of 0.1 in the previous figure)
for clarity.
CONCLUDING REMARKS
Our analysis indicates that use of imaginary-valued
similarity solutions can extend the applicability
of similarity solutions to a much wider range

Fig. 3 Contours of equal spray concentrations in the boundary layer for the case of = 1.2 and = 0.1

R. Z. ORLAND, D. KATOSHEVSKI, AND D. M. BRODAY

Fig. 4 Normalized liquid- phase concentration deposited on the wall surface, as a function of downstream location
x; for each plot, the concentration is normalized by its maximum value, which is at minimum x

of flow configurations than previously addressed.

Such flow configurations may be associated, for
example, with drastically changing diameters of
pipe systems or with systems that require control of spray distribution at a boundary. For such
systems, a similarity solution has been introduced
that addresses the spray phase, with further study
needed to ascertain effects on heat transfer. The
obtained solution for the liquid phase distribution
admits a double branching, where one branching
point is physical and corresponds to switching
from decrease to increase of the liquid concentration downstream ( > 0, , 0) and the other
branching point is of a mathematical nature and
switches from real values of the similarity variable and functions to imaginary ones, enabling us
to describe a new range of cases by a similarity
solution. As deposition and evaporation would appear to be the obvious candidates for downstream
free-stream spray concentration variation, the value
of would likely be negative in most systems, and
deposition efficiency would decrease downstream.

REFERENCES
1. D. M. Broday, Deposition of Ultrafine Particles at
Carinal Ridges of the Upper Bronchial Airways,
Aerosol Sci. Technol., vol. 38, pp. 9911000,
2004.

2. D. Katoshevski, A Mathematical Study of a Fire

in the Interaction Zone between Flows with Different Velocities, AAQR, vol. 6(2), pp. 193212,
2006.
3. D. Katoshevski and Y. Tambour, A Theoretical
Study of Polydisperse Liquid Sprays in a Shear
Layer Flow, Phys. Fluids A, vol. 5, pp. 3085
3098, 1993.
4. K. R. Rajapogal, A. S. Gupta, and T. Y. Na,
A Note on the Falkner-Skan Flows of a NonNewtonian Fluid, JCR., vol. 18, pp. 313319,
1983.
5. H. Blasius, Grenzschichten in flussigkeiten mit
kleiner reibung (The Boundary Layer in Fluids
with Very Little Friction), Z. Math. Phys., vol.
56, pp. 137, 1908.
6. V. M. Falkner and S. W. Skan, Some Approximations of the Boundary Layer Equations, Philos.
Mag., vol. 12, pp. 865896, 1931.
7. D. R. Hartree, On an Equation Occurring in
Falkner and Skans Approximate Treatment of the
Equations of the Boundary Layer, Proc. Camb.
Philos. Soc., vol. 33, pp. 223239, 1937.
8. L. Rosenhead, Laminar Boundary Layers, Oxford
University Press, London, 1963.
9. G. K. Batchelor, An Introduction to Fluid Dynamics, Cambridge University Press, Cambridge,
1967.
10. T. Cebeci and P. Bradshaw, Momentum Transfer
in Boundary Layers, McGraw-Hill, New York,
1977.
11. A. E. P. Veldman and A. I. Van De Vooren, On
a Generalized Falkner-Skan Equation, J. Math.

SIMILARITY SOLUTIONS FOR LAMINAR BOUNDARY LAYERS

Anal. Appl., vol. 75, pp. 102111, 1980.

12. P. Brodie and W. H. H. Banks, Further Properties
of the Falkner-Skan Equation, Acta Mech., vol.
65, pp. 205211, 1986.
13. D. Katoshevski and J. B. Greenberg, A New
Solution for an Evaporating Spray in a Shear

Layer, Including Downstream Flow Deceleration/Acceleration, Atomization and Sprays, vol.

17, pp. 641658, 2007.
14. K. Stewartson, Further Solutions of the FalknerSkan Equation, Proc. Camb. Philos. Soc., vol. 50,
pp. 454465, 1954.