Escolar Documentos
Profissional Documentos
Cultura Documentos
Reynolds Equation
u i u i p 2ui
+uj = + (1)
t x j x i x j x j
u i
= 0. (2)
x i
During a turbulent motion u and p are random functions of space and time.
Hence, these may be decomposed into mean and fluctuating parts, i.e.
u = U + u , U i = u i , u i = 0 , (3)
p = P + p , P = p , p = 0 , (4)
where U and P are the mean quantities and u ' and p' are the fluctuating parts. Here, a
bar on the top of the letter stands for the (time) averaged quantity. That is
1 t+T
T T
u i = lim u i dt (5)
t0
+
< u i >= u f (u)du ,
i (6)
Note that the ergodicity hypothesis has not been proven for turbulence; however, it is
commonly used to relate the theoretical results to the experimental data.
ME637 1 G. Ahmadi
u i u j 0 , p' u i 0 , u i u j u k 0 . (8)
About a century ago, Reynolds suggested to use the decomposition given by (2)
and (3) into the Navier-Stokes equation and average the resulting equation. Noting that
u i
U i u j = U i u j = 0 , = 0, (9)
x j
it follows that
U U i uu
= P + U i i j ,
2
i + U j (10)
t x j x i x j x j x j
U i
=0 (11)
x i
Equation (10) is referred to as the Reynolds equation. The term u i u j is the stress
induced by the turbulent fluctuation. i.e.,
U U i U j
i + U j = P ij + ( U i + ) u i u j (13)
x j
t x j x j x i
U i U j
where ij = P ij + ( + ) is the mean viscous stress tensor and u i u j is the
x j x i
turbulent stress tensor.
The Reynolds stress (turbulent stress) is a symmetric tensor and its components
are given by
u 2 u v u w
tT
= u v v 2
v w .
(14)
u w v w w 2
Note that the turbulent stresses introduce six additional unknowns into the
averaged Navier-Stokes equation.
ME637 2 G. Ahmadi
Phenomenological Theories of Turbulence
Boussineq suggested
u k u k U U j
Tij = ij + T i + ,
(15)
3
j x x i
where
T = T
U
12
T
= T = T . (16)
y
It is now well recognized that T is not a constant and is strong function of state of
turbulent motion.
l
T = u v (17)
1 1
dU
(u 2 ) 2 ~ ( v 2 ) 2 ~ l , (18)
dy
ME637 3 G. Ahmadi
U U
T = l 2 , (19)
y dy
or
U
T = l 2 . (20)
y
G.I. Taylor recovered the mixing length hypothesis based on a vorticity transport
formulation and von Karman used a similarity analysis for estimating the mixing length
as
dU
dy
l= 2 (21)
d U
dy 2
The mixing length theory has dominated the field of turbulence modeling for
more than half a century. It is now known that the mixing length hypothesis works for
turbulent flows, which are characterized by single length and velocity scales. The reason
for its effectiveness is simply due to dimensional requirements!
2
U
0 = y
2 2
, (23)
y
or
dU u *
= . (24)
dy y
ME637 4 G. Ahmadi
Integrating Equation (24), it follows that
U 1
= U + = ln y + c , (25)
u *
or
+1 + + u *y
U = ln y + B , y = , ( 30 < y + 300 ) (26)
where B 5 .
Very near wall, in the viscous sublayer, turbulence fluctuation becomes small and
the viscous stress becomes dominant. As a result,
dU
0 = (27)
dy
or
2 dU dU +
u* = , =1 (28)
dy dy +
Hence,
U + = y+ ( 0 < y + 5 ). (29)
30
U+
20
U + = 2 .5 ln y + + 5 . 5
10
U + = y+
12 30 300
Schematics of turbulent velocity profile near a wall.
ME637 5 G. Ahmadi