Vous êtes sur la page 1sur 7

CRITERIA FOR INCIPIENT MOTION OF SPHERICAL SEDIMENT PARTICLES

By Chi-Hai Ling I

ABSTRACT: Initiation of bed-load transport of uniform spherical sediment particles on a horizontal bed in an
open-channel flow is studied. On the basis of micromechanical and fluid dynamical considerations, two separate
criteria for the initiation of motion are derived: one for rolling and one for lifting. Fluid forces such as drag,
shear lift, Magnus lift, and lift due to centrifugal force are included in the derivation. The formulation of the
lift force is theoretical. No empirical coefficient is used other than the drag coefficient C" which is well
established in the literature. In the low particle Reynolds number regime, where u.dlv oS 1, the dimensionless
stress required to initiate lifting is found to be much higher than that of rolling, the former substantially above
the Shields' curve while the latter is substantially below it. The theory is compared with Vanoni's 1964 data,
with meaningful results. For higher particle Reynolds number, where u*dlv > 10, the dimensionless stress
needed to initiate lifting is closer to that of rolling. The Shields' curve for the most part lies between the two
Downloaded from ascelibrary.org by University Of Miami on 12/29/14. Copyright ASCE. For personal use only; all rights reserved.

theoretical thresholds.

INTRODUCTION THEORY

Shields (1936) has been credited as the first person to ex- For noncohesive uniform spherical sediments in a steady
press the critical shear stress for initiation of sediment motion two-dimensional uniform flow of fluid in an open channel,
in a dimensionless form. His relation has the form the movement of the sediments on the top layer is typified
as shown in Fig. 1, where one sphere is situated on top of
two other spheres in a hypothetical two-dimensional config-
Til _ F (U*d) (1)
('Y, - 'Y)d v uration. Letting the fluid motion be from left to right, the
balance of forces in the normal and tangential directions are
in which To = critical shear stress; 'Ys = specific weight of the
D sin A + (G - L)cos A = N + 2rnr(dS/dt)2 (3)
sediment; 'Y = specific weight of the fluid; d = diameter of
the sediment; u* = shear or friction velocity, being equal to D cos A - (G - L)sin A - T = 2rnr[(d ZS)/(dt Z )] (4)
V(To/p), where p = density of the fluid; v = kinematic vis-
cosity; and F = some function. in which D, G, L, N, and T = fluid drag on sphere 1, weight
Eq. (1) was obtained largely from dimensional considera- of the sphere, fluid lift, normal force from sphere 3, and
tions, but it was subsequently verified by experiments and tangential force from sphere 3, respectively (no reactional
became the well-known Shields' curve. A number of theories forces from sphere 2 is considered, because they are zero
have come to existence to explain (1). Reviews by Vanoni when sphere 1 is about to move to the right); S = angle
(1966) and Graf (1971) included theories by White (1940), between the line segments 0\0 3 and 0203; A = angle be-
and Egiazaroff (1965), among others, who essentially derived tween the normal and gravity forces with A + e = 90°; t =
their theoretical results by balancing fluid drag with the weight time; rn and r = mass and the radius of the sphere, respec-
of a sediment particle. Coleman (1967), Yang (1973), Ikeda
a
tively; and the symbol is used for differentiation to distin-
guish it from the diameter d.
(1982), and Wiberg and Smith (1987) on the other hand in-
The moment equation (McCuskey 1959) about the point
cluded the lift force F L with the drag force and weight in their
of contact A is
momentum or moment balance
(5)
L = C,A[(p/2)V2] (2)
in which M A = moment about point A; re; = position vector
in which CL = lift coefficient; A = cross-sectional area of O\A; ac; = acceleration of the sphere at the point 0\; x =
the sediment particle; p = fluid density; and V = fluid ve- symbol for vector product; Ie; = moment of inertia of the
locity at some reference point between the top and the bottom sphere about 0\; CI. = angular acceleration of the sphere; and
of the particle. k = a unit vector perpendicular to both re; and a(; in the
The present theory starts with both the moment and mo- direction a right-handed screw advances.
mentum equations in the balance of drag, lift, and gravity.
It partially incorporates the previous work of Coleman (1967) SPHERE 1
and Wiberg and Smith (1987). The purpose of this paper is
z
to develop separate criteria for rolling and lifting by treating
the lift force as the sum of the Saffman (1965) shear lift, the
Rubinow and Keller (1961) spin (or Magnus) lift, and the
centrifugal force. The need to use the empirical and uncertain
lift coefficient C L has been eliminated.

'Hydro., U.S. Geological Survey, Water Resour. Div., Western Re-


gion. 345 Middlefield Rd., MS-496, Menlo Park, CA 94025.
Note. Discussion open until November 1,1995. To extend the closing
date one month, a written request must be filed with the ASCE Manager
of Journals. The manuscript for this paper was submitted for review and
possible publication on May 2. 1994. This paper is part of the Journal SPHERE 2 SPHERE 3
of Hydraulic Engineering, Vol. 121, No.6. June, 1995. ©ASCE, ISSN
0733-9429/95/(J()06-0472-0478/$2.00 + $.25 per page. Paper No. 8386. FIG. 1. Definition Sketch

472/ JOURNAL OF HYDRAULIC ENGINEERING / JUNE 1995

J. Hydraul. Eng. 1995.121:472-478.


Since the magnitude of au is 2rd 2 6/dt 2 and I u = (2/5)mr 2 , in which
(5) can be written as
(15)
d2e 2 d2 e'
(L - G)sin A + D cos A = 2mr -d->
t-
+ -5 mr d- >
t"
(6)
L", = 'lTr'pwV, ( 16)
where 8' = angle of rotation of sphere 1. Eqs. (3), (4), and
and
(6) can be solved numerically when it is known whether sphere
1 is rolling or sliding. If it is rolling, then 6' = 28; and if it (rwF
is sliding, then T = ILrN, where ILr = coefficient of friction L = m --cos A (17)
.. 2r
between the spherical surfaces. Ling and others (1992) studied
a similar problem where the driving force on the sphere is where a" = Saffman lift coefficient, being equal to 1.615:
gravity only. They numerically solved the equations of motion dU/dZ = velocity gradient; r = radius of the sphere; and w
and found that when motion begins, it starts with rolling = angular velocity of the sphere, counting positive clockwise.
because T is usually much smaller than ILrN. In fact, sliding Saffman (1965) found that the spin lift was less than the
is limited to consuming up only about 1% of the kinetic energy shear lift by an order of magnitude and subsequently many
Downloaded from ascelibrary.org by University Of Miami on 12/29/14. Copyright ASCE. For personal use only; all rights reserved.

the sphere carries. Experiments of Halow (1973) showed that authors ignored the spin lift in their momentum analysis.
spherical particles roll and angular particles slide in turbulent However, the value of the Saffman lift coefficient was later
horizontal and inclined pipe flow. We will therefore assume found to be miscalculated, and Saffman (1968) corrected it
rolling in the analysis. from 20.3 to 1.615. This correction puts the spin lift back at
Eq. (6), after incorporating the condition 6' = 26, becomes the same order of magnitude as the shear lift, therefore both
lift forces L, and L", should be retained in calculating total
. 14 d2e lift L.
(L - G)sm A + D cos A = 5 mr dt 2 (7)
The gravitational force on the sphere or the submerged
weight is
The condition for rolling to begin is then d 2 8/dt 2 ~ 0, or
G = (4/3)'lTr'(y, - "I) = ('IT/6)(y, - -y)d' (18)
(L - G)sin A + D cos A 2': 0 (8)
When the first threshold is reached, sphere 1 starts to roll.
For A = 30°, this reduces to
And if the flow produces enough stress, sphere I may roll
V3D+L~G (9) over sphere 3 and sphere 4 or more spheres while picking up
speed. However, the maximum angular velocity sphere I can
Eq. (9) is a condition of initial movement for closely packed achieve is that of the freely rotating particle, or (1/2)(dU/dZ)
two-dimensional arrangement of spheres assuming a hori- (Saffman 1965), when the flow is the only driving force for
zontal bed and identical packing in the y-direction. It needs the sphere. Thus w :s (l/2)(dU/dZ), or W lllax = (1/2)(du/dz).
some modification to be applicable to three dimensional sit- The experiments of Ling and others (1992) have shown that
uations. Coleman's (1967) analysis of closely packed three for a particle Reynolds number less than 500 the rolling sphere
dimensional arrangement of spheres showed the moment still keeps in contact with the sphere under it after collison,
equation as follows: because the energy in the normal direction is completely ab-
sorbed by the other spheres, directly and indirectly, so that
(1,/IJD + L 2': G (lOa)
the normal velocity is negligible after collision. In this paper.
where 12 and I} = lengths of moment arms, with we will therefore assume that sphere 1 keeps in contact with
the spheres below it while rolling until it is lifted. The second
d d threshold, one for suspension or lifting, is the condition be-
L = ~ R and I, (lOb,c)
- 4v 3 . V6 yond which the total lift force on the sphere exceeds its sub-
merged weight in the fluid and lifts the sphere off the bed.
thus (10), after dropping the unequal sign, can be written as
L=G (19)
4/Y2D + L = G (11)
As can be seen from (15) and (16), the shear lift L, is pro-
This condition represents the beginning of sediment move- portional to V, which is equal to V r - Vx' where V r = average
ment and will be considered as the first threshold, or "rolling velocity of the fluid across sphere 1; and V, = velocity com-
threshold," for incipient motion. A second threshold, or "lift- ponent of sphere 1 (at 01) in the x-direction. The spin
ing threshold," will be discussed later. lift L", is proportional to both V, and w, which is equal to
The drag force in (11) can be expressed as 2V,/(d cos A). As sphere 1 accelerates from W = 0 to w =
(1I2)(du/dz) the total lift L attains a maximum value some-
(12) where in between. To find this maximum, let
according to Rijn (1984), where CD = drag coefficient; and V, = VI - V, = BV, (20)
V, = relative velocity between the fluid and sphere 1. When
the Reynolds number is small, CD = 24/(Vr d/v), giving by using the fact that w = 2Vx /(d cos A) we can write

where IL = viscosity of the fluid. For turbulent flow at large


(13) L. = Bapv"'d2,),
., ., ( )'"
dU
'dZ
. (21)

Reynolds numbers CD is constant in (12).


L - B(I - B) _'IT_ pd 2 V2 (22)
The lift force L is expressed as the sum of the Saffman '" - 4 cos A '
(1965, 1968) force or shear lift L" the Magnus force or spin
lift (Rubinow and Keller 1961) L"" and the centrifugal force
L, = (1 - BF _'IT_ pd 2 V 2 (23)
L, 6 cos A I

L = L, + L", + L .. (14) Introducing the dimensionless velocity

JOURNAL OF HYDRAULIC ENGINEERING / JUNE 1995/473

J. Hydraul. Eng. 1995.121:472-478.


(24) u~
au - -
- (38)
az v
and the dimensionless velocity gradient
or
(25)
(39)
Eqs. (21)-(23) can be written as

( d)
0.5

L, = BaH,H~ "u*
--;;- Td~
'1
(26) Eq. (13) then gives

71'
Lot = B(1 - B) - - HrTd2 (27)
D = 371'J..ldV = 371'J..ld
f
(! U~d)
2 v
4 cos A

L = (1 - B )"' -71'- H'1'T"


d' (28)
= ~ 71' (1 + 2 ~) Td l (40)
Downloaded from ascelibrary.org by University Of Miami on 12/29/14. Copyright ASCE. For personal use only; all rights reserved.

, 6 cos A
and (26) gives
where T = pu~ = bed shear stress. Since L = L, + L", +
L" differentiation of L with respect to B gives the maximum
L at (41a)

(29) L, = B~ (U:d) (1 + 2~) Td 2 (41b)

where at the point where sphere 1 starts to roll, 0 = 0, W = 0, and


B = 1, thus
(30)
D = 371' Tdl. L (U.d)
a, - d'
2 ' , =2 - v T" (42a,b)
F, = 71' -
-- H'l' (31)
- 4 cos A L", = 0 and L, = 0 (42c,d)
and Substituting (14), (18), and (42) into (11) gives the rolling
threshold
F, = - -71' - H'1 (32)
~ Td2 ~' Td2 (U:d) ~ (y,
. 6 cos A
+ = - -y)d 3
Eq. (14) can be written alternatively as (43a)
L = [BF, + B(1 - B)F2 + (J - B)2FJ 1Td 2 (33) or

T
Turbulent Flow at Small Reynolds Numbers (71'/6) (=0.04 for u.d s; 1.0)
(-y, - -y)d (a)2)(u.dlv) + 671'/\/2 v
For small particle Reynolds number u*dlv the thickness of
the laminar sublayer is about 11.6 times the value of vlu* (43b)
(Einstein 1950; Wiberg and Smith 1987). Any particle di-
A numerical study was done to estimate roughly the max-
ameter less than 11.6vlu* (or u*dlv < 11.6) would be com-
imum total lift by using (29) and other related equations, for
pletely submerged in the sublayer. It is safe to say that for
values of B less than or equal to 1 (otherwise sphere 1 will
lI"dlv < 1 the fluid velocity around the particle is linear. Thus
be moving backward). The total lift obtained this way is greater
(34) than or equal to what is physically possible (for example,
some values of B require the sphere to rotate much faster
We now assume that the bed surface is at the bottom of sphere than W max ) , thus giving a lifting threshold from (19) lower
I when it is at rest (as shown in Fig. 1). As sphere 1 rolls than or equal to what it should be. But it is still much hig~er
over sphere 3, its center 0, is raised by the amount 0, where than the rolling threshold of (43). Therefore the followmg
scenario is assumed: sphere 1 will start to roll first when the
5 = d( cos A - cos Ao) (35) stress reaches the rolling threshold, and then accelerate to its
maximum angular velocity W max ' As the stress level increases,
here cos Ao = 0.8164, the lowest position for sphere 1 in the sphere will roll to higher values of W max ' Finally the stress
three-dimensional configuration. Letting Ub and u, be the fluid reaches the lifting threshold and sphere 1 lifts away from the
velocities at the bottom and the top of the sphere where z bed. Hence at the point of lifting, sphere 1 is rotating at the
= 0 and z = d + 0, respectively, we have W max produced by the threshold stress. Yet

(36) 2Vx 2(1 - B)Vf 2( 1 - B)H ,11*


W = -d-c-o-s-A = d cos A (44a)
d cos A
or
and

(37) (44b)

and therefore setting W = W max gives


474/ JOURNAL OF HYDRAULIC ENGINEERING / JUNE 1995

J. Hydraul. Eng. 1995.121:472-478.


B = 1 - (::;J cos A (44c)
quartz sand with a mean diameter of 0.102 mm. The 80%
confidence interval was between 0.086 mm to 0.120 mm and
the specific gravity was 2.65. The rate of movement were
at the point of lifting. classified into four groups: in ascending order the groups are
The lifting threshold from (14) and (19) is therefore "negligible," "small," "critical," and "general." A compar-
[BF, + B(1 - B)F~ + (1 - BFF3lTd~ = ('iT/6)(y, - -y)d 3 ison of (43) and (45) with Vanoni's data is shown in Fig. 2.
(45a) Indeed, all of Vanoni's data lie between the two theoretical
thresholds.
or
T ('iT/6) Turbulent Flow at Large Reynolds Numbers
(45b)
(-Y, - -y)d BF, + B(1 - B)F~ + (1 - B)~F,
The equations to be used for establishing the two thresholds
where F Fl , and F3 are from (30)-(32), with for turbulent flow at large Reynolds numbers are the same
" as those at small Reynolds numbers. But since the velocity
H, = V, = ! u*d (1 + 2~)d distribution is different, the expressions for V, and au/az in
Downloaded from ascelibrary.org by University Of Miami on 12/29/14. Copyright ASCE. For personal use only; all rights reserved.

(45c)
u* 2 v the equations will also be different. For particle Reynolds
numbers greater than 30, the flow over the sediments is in
Ho = au .!!- = u*d (45d)
the completely rough regime, for which the velocity distri·
- az u* v bution is (Schlichting 1960)

and u/u* = 8.5 + 5.75 log zlk, (46a)

o or, alternatively, as in Wiberg and Smith (1987)


d= cos A - cos Ao (45e)
ulu* = 2.5 In zlzo (46b)
from (37), (39), and (35). Calculations show that the maxi-
mum lift for the linear velocity profile occurs at cos A = 1,
where k, = equivalent sand roughness of Nikuradse's exper-
or when sphere 1 is on top of sphere 3.
iments on rough pipes; and Zo = roughness parameter being
The rolling and lifting thresholds shown in (43) and (45)
are the theoretical minimum for rolling and lifting, respec- equal to k)30. According to Rijn (1984) who quoted works
tively, of incipient motion under idealistic conditions for uni- by Kamphuis (1974), Hey (1979), Mahmood (1971), and Gladki
form particles. Existing experiments under very idealistic con- (1975), this sand roughness k, should be set equal to about
ditions are hard to obtain for comparison. However, Vanoni two to three times the diameter of the sediments when (46)
(1964) made two series of experiments to study incipient sed- is used in turbulent flow over sediments. Here k)d = 3, which
iment motion in turbulent flow at small Reynolds numbers, corresponds to dlz o = 10, will be used for well-sorted sedi-
each with a single sediment possessing a narrow 80% confi- ment. The effects of this ratio on the thresholds will be dis-
dence interval for the size distribution and thus making the cussed later in the section headed "Comment."
sediment size reasonably uniform. So these experiments are In this paper, VI and au/az will both be derived by taking
compared with the theory. the simple average of u and aulaz.
In one experiment Vanoni (1964) used glass beads with a Consider the velocity profile around the sediment particle.
mean diameter of 0.037 mm. The diameters of 80% of the The logarithmic profile of (46) is valid for z > 100zo . An
beads were between 0.029 mm to 0.046 mm. The specific extension of this profile to z = Zo is used as an approximation
gravity was 2.49. For the other experiment he used natural following Wiberg and Smith (1987). Thus the average relative
velocity Vr between the flow and the particle and the velocity
gradient aulaz can be calculated in terms of the ratio d/z o

o
RATE Of TRANSPORT
NONE V, =
1 {=~d'&
d J=~" udz =
1 {'H&
d J& u*
( z)
2.5 In ~ dz
.. NEGLI GIBLE
+ SMALL

'b
)( CRITICAL
o GENERAL = 2.5u* { (1 + ~) [In t (1 + ~) - 1]

. ..~~~ .
_ 0, Zo (In ! . . 0, Zo _
1)}
..
I

..•.. ~.. d Zo d (47)


~'? '.-.~~
_,-r- ... ". # and
,,>'Q"+ •••• ~

.~
. -.. ++ ~ ''1'"
~.t·.
au = .! {=-d+& au dz
d J= - &
=
0" •••••••........ az az d
ROLLINS THRESHOLD EO. (131 ........ _-
= 2.5 -u* { In -d ( 1 + -
d Zo
0) -
d
In (!!...~)}
Zo d (48)
where the notation 8, Zo means 8 or zo, whichever is larger.
Substituting (47) and (48) into (12), (24), and (25), gives

FIG. 2. Equations for Roiling and Lifting Thresholds Compared (49)


with Data from Vanonl (1964)

JOURNAL OF HYDRAULIC ENGINEERING / JUNE 1995/475

J. Hydraul. Eng. 1995.121:472-478.


LEGENDS
o GLASS BEADS (VANONIl
.. SAND lVANDNIl
~ AMBER (SHIELDSl
(50) x LIGNITE (SHIELDS)
(~ ~ LIGNITE (SHIELDS)
and ").
v BARITE 1SHIELDSl
~ • STEEL SHOT (WHITEl
He = 2.5 {In -d(1 + -8) - In (d
--8, -zo)} (51) .. \~ ~
'\ (')
.. SAND (WHITEl
Zo d Zo d
..
I
"~ to Tto
The rolling threshold for the incipient motion of sediments .
.... o'~.
"40
·r·····... ~
~ .......• r.
in turbulent flow at large particle Reynolds numbers can ~hA ~ .. EO. 1531
therefore be obtained by first setting 0 = 0; L", = 0; L, =
0; and B = 1 in (14), (41), and (49)-(51), and then substi·
\"", ~ ;~ ..~~~~.~.~_ ..
Downloaded from ascelibrary.org by University Of Miami on 12/29/14. Copyright ASCE. For personal use only; all rights reserved.

tuting into (11) EO. (43) '--. + ,,'7"."-"


ROt · ··... . " .
d)O.5 LING THRf: ..
4 1T C
08 D
H'ITd'• + aH
., I
H~5
- (~
v Td 2 = ~
6 ("V.
I.'
- "V)d
1
J SHOW rlr-" ..
• (57)

(52a)

or
..
u d/"

(1T/6) FIG. 3. Comparison of Various Thresholds of Incipient Motion


(52b)
(-Y, - -y)d o.5 (U*d) -0.5 7T C H2
H ,H
(X le +,!"iD! where
v 2v2
Values of CD are obtained by using the formula of Schiller
and Naumann (1933)
C
K
1
= - In(Kz,i); K = 0.4 (54b,c)

V d)O.6H7] (54d,e)
Cn = 24/(Vf dlv) [ 1 + 0.150 ( -;- (52c)
v

for up to Vrdlv = 1,754, which corresponds to u*dlv = 500. and


The value of a., is set equal to Saffman's 1.615 for all particle
Reynolds numbers in this study (up to 500). Calculation of k,
Zo = 30 (54!)
a., with (15) and (46) using the experiments of Chepil (1961)
gives a, = 0.655, 1.334, 1.146, and 2.78 for u*dlv = 150,
198,522, and 2,584, respectively. The average of these four
values is 1.48, which is very close to 1.615.
The lifting threshold for the incipient motion of sediments
in turbulent flow at large particle Reynolds numbers is [(33)]
(7T/6)
(53a)
(y, - y)d
where
(53b)

while F I , F2 , and FJ are calculated by using HI and H 2 from


(50) and (51). The maximum lift for the logarithmic velocity
profile occurs at cos 'A = 0.92. Since the F-values are different,
the notation T2 is used here to distinguish (53) from (45).
The thresholds for turbulent flow at both small and large
particle Reynolds numbers [(43), (45), (51), and (52)] are
compared with the Shields diagram in Fig. 3. Because the
.( _ 0.33~*d+ _ 1)] _[e -O·JJ!lu.dlv)(Sldl!

completely rough regime starts at about ksu* Iv = 70 (Schlichting


. (_ 0.33u*d ~ _
1960), or equivalently u*dlv = 23, the region where the par-
ticle Reynolds numbers fall between 1 and 30 is considered
v d 1)]} (55)
the transitional regime. For the velocity profile in this region and
Wiberg and Smith (1987) found that the formula of Reichardt
(1951) agrees excellently with measurements made in that
region and also fits well in both the small and large Reynolds
number range. This formula will be used to find the two
thresholds
= 2.5 In (1 + 0.4 u*:+) - C [1 - e 1111 1.6)(".<1 'Iv)

u = u* [1. In(l +
K
KZ+) - C (1 - e -z+/II.6 - ~ e -o.J}z+)]
11.6
__1_ (U*d+) e -'U}IU.d+IV)] _ {2.5 In
(54a) 11.6 v

476 / JOURNAL OF HYDRAULIC ENGINEERING / JUNE 1995

J. Hydraul. Eng. 1995.121:472-478.


·(1 + 0.4 U:d~) - c [1 - e -(I/11.6)((u.d/v)(&/d)J
The criterion for the incipient motion depends on how
tightly packed the bed is, and the direction of the flow, both
of which would change the angle A. How far apart the top
- 1:.6 (U:d ~) CO 33IlU.,IIV)(S/dlJ]} (56a)
layer of loose spheres (sphere Is) are situated also affects the
result because the ratio k)d will change. Schlichting (1960)
studied experimentally the values of equivalent sand rough-
where ness for a large number of roughnesses arranged in a regular
d+ = d[1 + (old)] (56b) fashion. He found for the same sphere diameter d the equiv-
alent sand roughness k, increased substantially when the spheres
With the foregoing dimensionless velocity and velocity gra- are put closer together. The ratio k)d varied from about 0.2
dient, the two thresholds from (11) and (19) can be obtained to about 4. One may expect similar behavior to hold true for
by using the expressions for D, L" L m , and L c in (12), (18) the configuration in this paper even though the bed is not flat
and (26)-(28). like Schlichting's. Thus the value of k,ld = 3 suggested by
The rolling threshold is Rijn (1984) and used in this paper may correspond to some
average situation. A smaller value such as k,ld = I that
Downloaded from ascelibrary.org by University Of Miami on 12/29/14. Copyright ASCE. For personal use only; all rights reserved.

(Tr/6) Wiberg and Smith (1987) used as a standard case would cor-
('y, - -y)d
H H~' (~
d)-o., + ~ C HZ (57)
respond to the case in which the moving spheres are less
0:
, 1 v- 2\11 D I densely distributed, and the thresholds from (56) and (57)
would produce lower values (by a factor of about two) for
where HI and H 2 are obtained from setting 0 = 0 in (55) and the dimensionless critical shear stress than those shown on
(56) and the lifting threshold is Fig. 3 because of an increase in the values of Vflu* and (aul
dz)(dlu*). Wiberg and Smith (1987) showed the same kind
(-rr/6) of effects from different ratios of k)d. Viewed from a longer
(58a)
(y, - -y)d time frame, this "compactness" of the top layer loose spheres,
where which can assume any value between "no sphere" to "tightly
packed," is an indication of where it is in the stages of con-
(4~J cos A
tinuous deposition to or erosion from the bed. This is perhaps
B = 1 - (58b) another reason for the scattering of measurements of critical
shear stress.
and F I , F2 , and F3 are from (30)-(32), with HI and H 2 cal- The effect of turbulence has not been included in this pa-
culated from (55) and (56). The value of cos A is set equal per. It is believed that the incipient motion of sediment par-
to 1 for u*dlv :s 3, and 0.92 for u*dlv ~ 3. ticles is sensitive to the fluctuating fluid forces. Therefore
Eqs. (57) and (58) are plotted as shown in Fig. 3. For small further research is needed to study this influence of turbu-
particle Reynolds numbers, they coincide with the curves rep- lence on incipient motion.
resenting (43) and (45); for large particle Reynolds numbers,
they approach those of (52) and (53). It is felt that (57) and CONCLUSIONS
(58) are adequate for use as threshold for all particle Reynolds
numbers. The lifting threshold and the rolling threshold are Two thresholds have been derived as criteria for the incip-
far apart at small Reynolds numbers and get closer and closer ient motion of spherical sediment particles. The rolling
as the Reynolds number increases. This perhaps explains the threshold gives the minimum dimensionless shear stress re-
belief that a unique relation for critical shear stress cannot quired to start the bed-load transport in the form of rolling
be obtained for u*dlv < 3.5 [Graf (1971), page 98], because and the lifting threshold gives the minimum stress for sus-
there is such a wide range between the two thresholds where pension. The Shields curve along with some representative
a critical shear stress may be subjectively and differently de- classical data lie mostly between the two thresholds; and Van-
fined. The Shields' curve lies for the most part between the oni's (1964) data fall completely between the two thresholds,
two thresholds except when the particle Reynolds number showing an increase of rate of transport as the data points
exceeds about 100. Measurements by Chepil (1961) and Cole- move from the rolling threshold toward the lifting threshold.
man (1967) seem to suggest higher values of O:s and CD for Since the experimental definition of a "critical shear stress"
large values of u*dlv than the ones used in this paper. When is subjective and not unique, because one may define it
higher values of O:s and CD are used, the rolling threshold anywhere between rolling and lifting, it may be useful to
would certainly go below the Shields curve for u*dlv > 100. experimentally establish the two thresholds rather than one
However, more experimental data are needed to establish vague "critical" threshold in the study of sediment trans-
that. port.
Although the derivations in this paper are based on an
COMMENT idealized situation with a deterministic approach, further ex-
tension into statistical analysis that may include particle shapes,
The lift force used in deriving the lifting threshold for large sizes and its distribution as well as bed geometry is possible.
Reynolds numbers is based on the maximum that is physically
possible for a sphere moving over a tightly packed bed made ACKNOWLEDGMENTS
of other identical spheres. In the three-dimensional config- The writer is very grateful to Dr. Jonathan Nelson, who suggested
uration (Coleman 1967), Sphere 1 would first roll over the the inclusion of the centrifugal force and commented on the treatment
valley formed by the two spheres whose line of connection of the velocity profile of the flow, among other things; and Dr. John
is perpendicular to the direction of the flow, and then over Dingler for his editorial comments and suggestions. Comments from two
a single sphere around its peak, then the valley and the peak anonymous ASCE reviewers are also appreciated.
again for a number of cycles until sphere 1 is lifted. An as-
sumption is made that when there is enough stress to move APPENDIX I. REFERENCES
the sphere, it will accelerate until the angular velocity of the Chepil. W. S. (1961). "The use of spheres to measure lift and drag on
sphere reaches the free-rotating value of (1I2)(aulaz) and then wind-eroded soil grains." Proc., Soil Sci. Soc. of Am., 25(5). 343-
remains roughly constant. 345.

JOURNAL OF HYDRAULIC ENGINEERING / JUNE 1995/477

J. Hydraul. Eng. 1995.121:472-478.


Coleman, N. L. (1967). "A theoretical and experimental study of drag Wiberg, P. L., and Smith, J. D. (191\7). "Calculations of the critical shear
and lift forces acting on a sphere resting on a hypothetical stream stress for motion of uniform and heterogeneous sediments." WaleI'
bed." Proc., 12th Congress, Int. Assoc. for Hydr. Res. (IAHR), Fort Resour. Res., 23(1\),1471-141\0.
Collins, Colo., Vol. 3, 11\5-192. Yang, C. T. (1973). "Incipient motion and sediment transport. "J. Hydr.
Egiazaroff,1. V. (1965). "Calculation of nonuniform sediment concen- Div., ASCE, 99(10), 1679-1705.
trations." 1. Hydr. Div., ASCE, 91(4), 225-247.
Einstein, H. A. (1950). 'The bed-load function for sediment transpor- APPENDIX II. NOTATION
tation in open channel flows." Tech. Bull. 1026, U.S. Dept. of Agr.
Soil Conservation Service, Washington, D.C. The following symbols are used in this paper:
Gladki, H. (1975). "Discussion of 'Determination of sand roughness for
fixed beds.'" J. Hydr. Res., 13(2). cross-sectional area of sediment particle (sphere);
Graf. W. H. (1971). Hydraulics ofsedimenllransport. McGraw-Hili Book
Co .. Inc., New York, N.Y. VrlV,;
Halow, J. S. (1973). "Incipient rolling, sliding and suspension of particles drag coefficient;
in horizontal and inclined turbulent flow." Chemical Engrg. Sci., Vol lift coefficient;
21\, 1-12. drag force on sediment (sphere);
Hey, R. D. (1979). "Flow resistance in gravel-bed rivers." J. Hydr. Div., symbol for differentiation;
ASCE, 105(4), 365-379. particle or sediment diameter;
Downloaded from ascelibrary.org by University Of Miami on 12/29/14. Copyright ASCE. For personal use only; all rights reserved.

Ikeda, S. (1982). "Incipient motion of sand particles on side slopes." J. weight of sediment (sphere);
Hydr. Div., ASCE, 101\(1), 95-114. dimensionless velocity V,lu*;
Kamphuis, J. W. (1974). "Determination of sand roughness for fixed dimensionless velocity gradient (aulaz) (dlu*);
beds." 1. Hydr. Res., 12(2), 193-203.
Ling, C. H .. Jan, C. D., Chen, C. L., and Shen, H. W. (1992). "Nu-
equivalent Nikuradse sand roughness;
merical simulation of a sphere moving down an incline with identical lift force on sediment (sphere);
spheres placed equally apart." Proc., 9th Conf. on Engrg. Mech., centrifugal force;
ASCE, New York, N.Y .. 764-767. Lm Magnus force or spin lift;
Mahmood, K. (1971). "Flow in sand bed channels." Water Mgmt. Tech. L, Saffman force or shear lift;
Rep. No. 11, Colorado State Univ .. Fort Collins, Colo. 12 , I, lengths of moment arms;
McCuskey, S. W. (1959). An introduction to advanced dynamics. Ad- m mass;
dison- Wesley Publishing Co., Inc., Reading, Mass., 3. N normal force on sediment (sphere);
Reichardt, H. (1951). "Vollstandige darstellung der turbulenten I' radius of sediment (sphere);
geschwindig-keitsverteilung in glatten leitungen." Z. Angew. Math.
T tangential force on sediment (sphere);
Mech., Berlin, Germany, 31(7), 201\-219 (in German).
Rijn, L. C. van (191\4). "Sediment transport. Part I: bed load transport." t time;
1. Hydr. Engrg., ASCE, 110(10), 1431-1456. u longitudinal flow velocity;
Rubinow, S. r.. and Keller, J. B. (1961). "The transverse force on a u* shear or friction velocity;
spinning sphere moving in a viscous fluid." J. Fluid Mech., 11(3),447- V fluid velocity at some reference point;
459. V, average velocity of fluid across spherical sediment;
Saffman, P. G. (1965). "The lift on a small sphere in a slow shear flow." Vr relative velocity between fluid and sediment (sphere);
1. Fluid Mech., 22(2), 31\5-400. V, velocity component of sphere in the x-direction;
Saffman, P. G. (1961\). "Corrigendum, the lift on a small sphere in a X, Z longitudinal, vertical coordinate;
slow shear flow." 1. Fluid Mech., 31(3), 624.
z" zero velocity level;
Schiller, L., and Naumann, A. (1933). "Uber die grundlegenden be-
rechnungen bei der schwerdraftaufbereitung." Zeitschrift Des verein 0,.\_ Saffman lift coefficient;
delllscher Ingenieure, Dusseldorf, Germany, Vol. 77, 318-320 (in Ger- specific weight of fluid;
man). specific weight of sediment (sphere);
Schlichting, H. (1960). Boundary layer theory. McGraw-Hili Book Co., angular displacement of sphere from horizontal;
Inc., New York, N.Y. angle between normal force and gravity;
Shields, A. (1936). "Application of similarity principles and turbulence cos - 1 (0.8164), maximum for A:
research to bed-load movement." Rep., W. P. Ott and J. C. van viscosity of fluid;
Uchelen, translators, California Inst. of Technol., Pasadena, Calif. kinematic viscosity of fluid;
Vanoni, V. A. (1964). "Measurements of critical shear stress." Rep. No.
p density of fluid;
KH-R-7, California Inst. of Teehnol., Pasadena, Calif.
Vanoni, V. A. (1966). "Sediment transportation mechanics: initiation critical shear stress with different expressions of HI'
of motion; progress report of the Task Committee on Preparation of H 2;
Sedimentation Manual." 1. Hydr. Div., ASCE, 92(2), 291-314. T" critical shear stress;
White, C. M. (1940). "The equilibrium of grains on the bed of a stream." W angular velocity; and
Proc., Royal Soc., London, England, A 174,322-338. aulaz velocity gradient in the z-direction.

478 I JOURNAL OF HYDRAULIC ENGINEERING I JUNE 1995

J. Hydraul. Eng. 1995.121:472-478.

Vous aimerez peut-être aussi