THEORETICAL SOLUTIONS FOR TURBULENCE GENERATED BY VIBRATING

GRIDS AND BY WALL FLOWS USING THE k- MODEL IN LINEAR FORM

Harry Edmar Schulz
Johnnes Grson Janzen
Karine Cristiane de Oliveira Souza
[email protected]
Departamento de Hidrulica e Saneamento, Escola de Engenharia de So Carlos,
Universidade de So Paulo
Av. do Trabalhador Sancarlense 400, 13566-590, So Carlos S.P. Brasil
Abstract. Theoretical solutions for turbulence are presented considering the basic equations
of the k- model. The equations are used with the terms of diffusion, generation and
dissipation, together with the "eddy viscosity" or "turbulent viscosity" defined by the k-
model. In the present formulation only stationary (steady) situations are considered and
solutions are obtained for different flow conditions. The deduction procedures followed to
obtain linear governing equations for different flow conditions are presented in detail, and
the proposed solutions may be obtained through traditional mathematical tables or adequate
softwares. Experimental results are compared with the presented theoretical predictions. The
obtained results permit to suggest, for example, that "wall flows" and "grid flows" are well
represented through the proposals of the present study. Based on the results it is concluded
that: a) theoretical solutions are important for the better understanding of turbulent flows
(possibility to verify basic assumptions, which is not always the case in numerical
simulations); b) the k- assumption represents an adequate ad-hoc approximation for flows
such as those considered in this study.
Keywords:Turbulent flows, Turbulence models, k- model, Grid turbulence, Wall Turbulence

1.

INTRODUCTION AND OBJECTIVES

The k- model is perhaps the most largely model used to predict mean characteristics
of turbulent flows. Its use is mostly coupled with numerical procedures, applied directly to the
original equations for k and . The original equations are non-linear, and relatively less effort
has been directed in the literature to find linear forms which permit theoretical predictions of
flow characteristics. Theoretical solutions are important because they may furnish details of
the behavior of turbulence parameters in the flow. Moreover, theoretical solutions may be
used to check mathematical procedures used in numerical codes, through the comparison
between numerical and theoretical predictions. In this sense, the main objective of this study
is to present theoretical solutions for flows described by the k- model. The terms of
diffusion, generation and dissipation of turbulent kinetic energy are considered, together with
the "eddy viscosity" or "turbulent viscosity". Through the use of mathematical tools, linear
governing differential equations are first obtained. Further, the program Mathematica was
used to obtain the solutions presented here for different situations. For wall flows, a finitedifference scheme is used to show the form of the solution of the governing linear equation
for k and . The theoretical predictions are compared with experimental results of several
sources, showing very good agreement between data and theory.
2.

k- MODEL FOR TURBULENCE WITH DIFFUSION-DISSIPATION TERMS

The general equations for turbulent flows, considering the k- approximation, are giving

by:
The turbulent kinetic energy (k) equation:
k
k

+ ui
=
t
xi xi

t
k

k
xi

ui
u ' i u ' j

xj

(1)

The equation for the dissipation rate () of turbulent kinetic energy:

t

+
xi
uj
u
+ C1 t i +
k x j x i

+ ui
=
t
xi xi

ui
2

C2
xj
k

(2)

The equation for the turbulent viscosity (t):

t = C

k2

(3)

In Eq. (1), (2) and (3) u i represents the i component of the mean velocity, u ' i represents
the i component of the velocity fluctuation, C1, C2, C, C, and k are constants of the
model. For the cases studied here, stationary conditions and no mean motion are considered,
so that the left side of both equations vanish. When considering only diffusion and dissipation

phenomena, also the middle term of the right side in both equations vanish (which
corresponds to production). Eq. (1) and (2) simplify to:

xi

C k 2 k

k xi

(4)

xi

C k 2

xi

2
= C 2
k

(5)

Basic situations of turbulent flows may be studied as one-dimensional cases. Considering

one-dimensional problems, the first integration of Eq. (4), furnishes (Schulz and Chaudhry,
1998):
=

d 2 C 3
k + w1
dx 3 k

(6)

w1 is an integration constant. Although Eq. (6) is still nonlinear, and k appear isolated (only
one variable at each side of the equation), and suggest that
= (k )

k = k (x)

and

(7)

Matsunaga et al.(1999) suggest the transformation (8), which simplify the representation
of the original equations:
d F k
=
d x t

(8)

Equations (4), (5) and (8) appear as:

d 2k
=k2
2
dF

and

d 2
= j k
d F2

(9)

where the following definitions apply:

k=

C
k

and

j=

C2
k

(10)

The first integration of Eq. (4), after the suggestion of Schulz and Chaudhry (1998, 1999)
and Schulz (2001), furnishes:
dk
2
= w2 + k 3
dF
3

(11)

w2 is an integration constant. Equation (6) is still valid, so that Eq. (7) may be expressed as:

 = (k )

and

k = k (F )

(12)

From Eq. (12) it follows immediately that:

d 2
d 2
=
d F2 d k2

dk
d d 2 k

+
dk d F2
d F

(13)

Using Eq. (8), (9), (12) into Eq. (13) leads to:
d
2 3 d 2

+k2
jk = 0
w2 + k
2
3 dk
dk

(14)

This result is important, because the original problem, composed by two nonlinear
coupled equations for = (k , F ) and k = k (F ) , is transformed into only one linear equation for
= (k ) . To generalize the equation and its results, a nondimensional form (Eq. 15) is
presented, considering the values of the turbulent kinetic energy (k0) and its dissipation rate
(0) at the origin.
d *
2 3 d 2 *

+ k *2
jk * * = 0
w* + k *
2
dk*
3

dk*

k
* =
and
k* =
0
k0

(15)

w* is a nondimensional constant. A solution of this linear equation is presented using

Hypergeometric functions (Hyp2F1 in Eq. 16), useful to represent turbulence generated in the
region between two oscillating grids.

0.667
0.667k *3

Hyp 2 F1 0.567,0.734,0.667,
+
1 w3 Hyp 2 F1 0.234,1.067,1.333,
w *
w *

3
k * w Hyp 2 F1 0.567,0.734,0.667, 0.667 Hyp 2 F1 0.234,1.067,1.333, 0.667k *

w
w
*
*

* (k *) =
0.667

Hyp 2 F1 0.567,0.734,0.667,
w *

(16)
w3 and w4 are integration constants. Figures 1 and 2 show the agreement between the
proposed solution and experimental data obtained by Janzen (2003) for turbulence generated
between two oscillating grids, in nondimensional form. Such analysis was also performed, in
dimensional form, by Souza (2004). The constants of the k- model used here are C,=1.44,
C2=1.92, C=0,09, =1.3 and k=1.0. The boundary condition adopted is: for k* = 1, * = 1.
The remaining integration constants were evaluated through adjustment with the experimental
data. Table 1 presents some of the studied experimental conditions.

1000
500
100
50
10
5
2

10

20

50

Figure 1 * versus k* (S = 2.0 cm, f = 2.0 Hz of table 1). The line is the prediction of Eq.
(16). Dots are measured values.

1000
500
100
50
10
5
2

10

20

50

Figure 2 * versus k* (S = 5.0 cm, f = 2.0 Hz of table 1). The line is the prediction of Eq.
(16). Dots are measured values.
Table 1: Some of the experimental conditions studied for turbulence generated by a pair
of oscillating grids
Frequency f
(Hz)

2.0
3.0
2.0
3.0

Stroke S (cm)

w3=w4

w*

2.0
2.0
5.0
5.0

4.5
6.0
2.0
2.5

1.0
1.0
1.0
1.0

Hypergeometric functions as presented by Eq. (16) are tabulated in the literature and may
also be evaluated through adequate softwares. The obtained solution also permits to calculate
other variables (although numerically). As an example, from definition (8) and Eq. (3), (11)
and (16) it is possible to obtain the evolution of k* with the nondimensional distance x*, in the
form of Eq. (17):

dx C
=
dk k

0.5

k 2
2
(k *) w * + k 3
3

x* =

, where

x 0
k 03 / 2

(17)

*(k*) is given by Eq. (16). A graph of k* for the same condition of fig. 1 is given by fig. 3.
Experimental points were obtained by Janzen (2003). Equation. (17) was solved using a
fourth order Runge-Kutta scheme, with boundary condition k* = 1 for x* = 0. Experimental
energy dissipation rates were obtained using =CDk3/2/L, where CD is an empirical constant
and L is the integral turbulence length scale. The value of CD used here is 0.5.

1
0.8
*

0.6
0.4
0.2
100

200

300

400

500

Figure 3 k* as function of the normalized distance (S = 2.0 cm, f = 2.0 Hz, CD = 0.5). The
line is the prediction of Eq. (17). Dots are measured values.
3.

k- MODEL FOR TURBULENCE WITH PRODUCTION-DIFFUSIONDISSIPATION TERMS

Considering flows parallel to walls and using the turbulent viscosity defined with the
mean velocity gradient, general Eq. (1) and (2) are now simplified to:
Turbulent kinetic energy (k) equation:

xi

t
k

k
xi

+ t j
xi

= 0

(18)

Energy dissipation rate () equation:

xi

xi

u
+ C1 t j
k xi

C = 0
2

(19)

Usual definition of the turbulent viscosity:

= t

d uj
d xi

(20)

 is the shear stress and is the density of the fluid. Traditional studies of wall-flows consider
a constant , related to the shear velocity u* as u* = / . The production terms of Eq. (18)
and (19) may then be represented as:
duj
t
d xi

u *4
=
C k 2

(21)

Equations (20) and (21) assume the forms:

xi

C k 2 k

k xi

u *4
+
= 0
2
C k

(22)

xi

C k 2

xi

u *4 2
2
+ C1
C

=0
2
C k 3
k

(23)

Once more a set of two nonlinear coupled equations must be solved. As for the diffusiondissipation problem, algebraic simplifications were elaborated to transform this problem into
only one linear equation relating k and . The following nondimensional forms simplify the
presentation of the equations:
For the turbulent kinetic energy:
k# =

k k
u *2 C

(24)

For the energy dissipation rate:

h
# d y#
y
#
#
,
,
y
and
dF
#=
=
=
2
*3 444444h444244444444444
k #4
u4
144
3

(25)

First possibility, using a fluid depth h.

yu*

# d y#
#
#
,
y
and
dF
#=
=
=
2
44244444444444
*3 4444444
k #4
u4
144
3

(26)

Second possibility, using only the variables already defined.

Any form conducts to the following nondimensional equations:
d 2k #
= k# 2 A
d F#2

(27)

d 2 #
#
+ C1 AB # C 2 Bk # # = 0
#2
dF
k

(28)

Where A =

k2
and B = are known constants. Following the procedures of Schulz
3
k
C

and Chaudhry (1998, 1999) and Schulz (2001), the first integration of Eq. (27) leads to:
d k#
2 3
= w # + k # 2 Ak #
#
dF
3

(29)

w# is a nondimensional integration constant. Following Eq. (12), we have:

# = # (k # )

k # = k # (F # )

and

(30)

Equation(13) may be reproduced using the nondimensional k# and #, which together with
Eq. (27), (28) and (29), lead to the linear governing Eq. (31), for # as a function of k#:

2
#
3
d
2 #4
#2
# #d
+ k # Ak #
k 2 Ak + w k
2
#
dk
3
dk

#
#

(31)

+ B C1 A C 2 k # # = 0

As for the diffusion-dissipation case, this result is important, because the original
problem, composed by two nonlinear coupled equations is transformed into only one linear
equation. Particular solutions may be found for different combination of values of the
constants involved in this equation. A graph using C,=1.44, C2=1.92, C=0,09, =1.3 and
k=1.0 is quickly obtained using a numerical finite-difference scheme, for example with:
d 2
dk #

#
2

#
i 1

k #

#
i +1

d #
=
dk #

and

#
i

i 1
k #

(32)

Equations (31) and (32) lead to:

# # 2 #4
#2
w k i + k i 2 Ak i
3

#
i 1

2
k

#
i
#2

#
i +1

# i i 1 #
#
+ k i# 3 Ak i
k #

+ C1 AB C 2 Bk i# 2 # i = 0

(33)
A value of k=1.0 was used. The optimal w# for channel flows is evaluated to be 67850.
This value is obtained considering experimental data for k and extracted from figs. 6.4 and
6.5 of Nezu and Nakagava (1993), measured for channel flows. Figure 4 shows the
comparison between the theoretical model and the experimental results. It must be
emphasized that the squares (experimental data) correspond to lectures made on graphics, and
may include deviations related to this somewhat crude way to compile information.
Anyway, it is interesting to note that a strong change in the behavior of the experimental data
(for k# about 35) is well reproduced by the theoretical linear model.

120
100

80
60
40
20
0
0

10

20

30

40

50

60

k#

Figure 4- Solution of Eq. (31) (solid line) and experimental data (squares)
collected in the literature for energy dissipation rate as a function
of the turbulent kinetic energy in channel flows.

4. CONCLUSIONS
In this paper, linear forms for the k- model are presented and applied to flows generated
between oscillating grids and flows parallel to rigid walls. This linearity permits to find
particular solutions, with different boundary conditions, using adequate combination of values
of integration constants (and also k- model constants, not always "universal") involved in the
obtained equations. The comparisons between theoretical solutions and experimental data (for
grid-turbulence and wall-turbulence) show very good agreement and point to the validation of
the use of the ad hoc k- model for flows such as those studied here. It is important to stress
that the original k- problem is composed by two non-linear coupled differential equations
and that the present solutions consider mathematical procedures which lead to only one linear
differential equation relating k and . The results show that the theoretical procedures are well
conducted and that the propositions given by equations (7) and (30) are valid for the flows
considered here. In this sense, it is concluded that: a) theoretical solutions are important for
the better understanding of turbulent flows (possibility to verify basic assumptions, which is
not always the case in numerical simulations); b) the k- assumption represents an adequate
ad-hoc approximation for the flows considered in this study.

Acknowledgements
To FAPESP, CNPq and CAPES, Brazilian foundations for research support.

REFERENCES
Janzen, J.G., 2003. Details of turbulence properties in water agitated through a pair of
oscillating grids, Master degree thesis, School of Engineering at So Carlos, University
of So Paulo, Brazil (text in portuguese).

Janzen, J.G. & Schulz, H.E., 2003. Using PIV to determine turbulence characteristics in tanks
with two oscillating grids, Proceedings of the XVth Brazilian Simposium on Water
Resource,, ABRH, Published in CD-ROM, Curitiba PR, Brazil (text in portuguese), 23 to
27 november.
Matsunaga, N.; Sugihara, Y.; Komatsu, T. & Masuda, A., 1999. Quantitative properties of
oscillating-grid turbulence in a homogeneous fluid, Fluid Dynamics Research, v. 25,
p.147-165.
Nezu, I. & Nakagava, H., 1993. Turbulence in Open-Channel Flows, AA Balkema, IAHR,
Rotterdam.
Schulz, H.E. & Chaudhry, F.H., 1998. A Theoretical Solution for Turbulence Generated by
Oscillating Grids, Proceedings of the First Spring School of Transition and Turbulence,
ABCM, Brazilian Association of Mechanical Sciences, COPPE, Rio de Janeiro, Brazil,
21 to 25/09, pp. 181-194.
Schulz, H.E. & Chaudhry, F.H., 1999. Theoretical Solutions for Turbulence Generated by
Two Oscillating Grids. In: XVth Brazilian Congress of Mechanical Engineering, ABCM.
Publicado em CD-ROM, ISBN-85-85769-03-3. (Published in CD-ROM), guas de
Lindia, SP, Brazil.
Schulz, H.E., 2001. Alternatives in Turbulence (in Portuguese). Ed. EESC-EDUSP, ISBN 8585205-37-7, So Carlos, S.P., Brazil.
Souza, K.C.O., 2004. Analysis of Theoretical Solutions for Grid-Turbulence Flows. Master
degree thesis, School of Engineering at So Carlos, University of So Paulo, Brazil (text
in portuguese).