SIMILARITY OF DISTORTED RIVER MODELS WITH MOVAELE BED

THIS PAPER
H. A. Einstein,l
--represents
an effort by the SocIety to deliver
technical
data direct
from the author
to the
reader with the greatest possible speed. To this
end, it has had none of the usual editing required
in more formal publication procedures.
Readers
are invited to submit discussion
applying to current
papers.
For this paper the final
date on which a discussion
should reach the
Manager
of Technical Publications
appears
on
the front cover.
Those who are planning papers
or discussions
for "Proceedings"
will expedite
Division and
Committee
action measurably
by first studying
"Publication
Procedure
for Technical
Papers"
(Proceedings
- Separate
No. 290) .. For free
copies of this Separate-describing
style, content, and format-address
the Manager,' Technical Publications,
ASCE.
Reprints
from this publication may be made on
condition
that the full title of paper, name of
author, page reference
(or paper number), and
date of publication
by the Society are given.
The Society

is not responsible

made or opinion expressed

for any statement

in its publications.

This paper was published at 1745 S. State Street,

Ann Arbor, Mich., by the American
Society of
Civil Engineers.
Editorial
and General Offices
are at 33 West Thtrty-ntnth Street, New York 18,
N.Y.

M. ASCE, and Ning Chien,2 A.M. ASCE

SYNOPSIS
The similarity conditions
derived from the theoretical
I describe the hydraulics and
[ numerical example is added
ticular river.

for distorted river models with movable bed are

and empirical equations which have been found to
the sediment transport in such rivers.
A complete
to demonstrate the method of application to a par-

INTRODUCTION
There are many hydraulic engineering problems for which the basic equations are known but which are geometrically
so complicated that the direct api plication of these equations becomes impossible.
Many such problems can be
! solved today by the use of models which are shaped to duplicate
thecomplicated geometry and in which the resulting flow patterns can be observed directly.
: Such a model permits the prediction of the corresponding prototype flows
: quantitatively only if the exact laws of modelstmtlartty
are known. The pre~diction of the model scales cannot be based on simple dimensional consider-a~tions if the model is distorted and includes the motion of a movable bed. A
" different approach must be used to find compatible systems of model scales
, and distortions.
I
In 1944, the senior author of this paper published a short account(l) of how
compatible systems of scales can be found. He proposed in that paper the der, ivation of model scales from empirical working equations rather than from the
: underlying differential equations of motion. It was pointed out there that equations must be used which are applicable in the same form both to model and
. prototype, and which should preferably have the form of power functions. The
same approach will be used in this paper. Many new developments in the del scription of flow and sediment transport in alluvial rivers will be incorporated.
Before the conditions of similarity can be stated .it is necessary to define
the exact meaning of this term. Let us define that similarity may be said to
: exist
1) if to each point, time and process in one scale, which may be called prototype, a corresponding point, time and process of the other scale, which
!
may be called the model, can be coordinated uniquely;
2) if the ratios of corresponding physical magnitudes between model and
t
prototype are constant for each type of physical magnitude .
I

I
I

~------------------------------------------------~

1. Prof. of Hydr. Eng., Univ. of California, Berkeley, Calif.

2. Asst. Research Engr., Inst. of Eng. Research, Univ. of California,
Calif.

566-1

:Berkeley,

1) If Lr is independent from hr, the model is vertically distorted.

This implies, for an undistorted model, that all magnitudes of equal dimensi
must follow the same scale ratio; also, the various laws which interrelate the
2) If the grain size ratio Dr is different from both Lr and hp a third length
variables parameters
and constants of the prototype must apply to the model
scale is introduced and with it a second distortion.
too. The 'results derived Irom the operatian of such a hydraulic model may
3) If Sr is chosen independent from Lr and hr , the model is assumed to be
then be tra.nsferred to the prototype by the use of these scale ratios.
tilted in addition to the other distortions.
A distortion in the physical sense exists if there are two types of variables
4) If the ratio of effective densities of the sediment (jj - if)r is assumed to
or parameters
of equal dimension operative in the same problem which are
be different from the ratio of the fluid densities ftr which is unity, then
physically sufficiently independent such that they can be given different scale
that represents a fourth distortion.
5) A fifth distortion is introduced if the time scale tlr far the time values
ratios.
involved in the determination of velocities and sediment rates is chosen
The applicatian of distortians in engineering is not new. Many engineers
apply daily distort.ions in the presentation of flat sections and profiles.
The
different from the ratio t2r of durations for individual flow canditions,
indicating the speed at which flow duration curves are duplicated.
desirability
of such distortion in hydraulic models has become apparent to
everyone who has tried to design a model of wide and shallow w~ter~ourses ~ 6) A sixth distortion is introduced because of the impossibility to. obtain
suspended-load rates in a model at the same scale at which the bed-Load
which the large hor-izontal dimensions call for a model scale WhIChIS much
small far application to the vertical direction.
Such small water depth would
rates are reproduced, making qB~ different from qT .
very often cause laminar flaw in the model which cannot be used to duplicate
7) A seventh and last distortion permits the ratio of setlling velocities Vsr
the turbulent prototype flow. Many laboratories have built extremely valuabh
of corresponding grains to be different from the ratio of corresponding
models in which the vertical scale is much larger than the horizontal scale.J
flow velocities.
It is impossible, however, to introduce one distortion alone. This becomes
parent by the fact that, for instance, the slopes are distorted in the same deRelationships Describing Alluvial Flows
gree as the vertical heights. It is generally known that the water depths and
the slope enter most of our friction formulas.
The velocity scale must be
It has already been stated that the flow and sediment description of the alchosen to satisfy the Froude condition and will satisfy the friction equation 0 luvial reaches which are to be studied by the model must be accomplished by
ly if the roughness of the various boundaries is also. properly dist~rted.
W~ pdentical formulas in model and prototype. These formulas can be transwill see shortly that in the case of a sediment carrying stream strll more dlSjfOrmedinto relationships between the various ratios and thus represent condrtortions must be introduced to balance their effects in the various equations tions which must be satisfied when the various distortions are chosen. The indescribing flow and sediment motion.
dividual relationships shall be discussed now using the various equations as
they are given in Ref. 2. It is not believed that the choice of these equations in
Namenclature and Distortions
the exact form of Ref. 2 is a strong restriction to the generality of the method.
Even if the reader prefers to substitute different formulas for the ones proIn the following discussion, the symbols with the subscripts, p and M, will posed here, he will find that he is not able to. change the number of equations
indicate quantities, in prototype and model, respectively.
The ratio of correllwhich must be satisfied in a particular problem. From this follows that he will
ponding values in the two scales will be denoted by a subscript r. Thus, the have the same number of conditions and, therefore, the same number of deratio of depth scales, for instance, is defined as hr = hp/hM; Similarly, the grees of freedom in choosing his ratios. He will find that his substitute equafollowing ratios are introduced:
tions will contain the same variables and that the entire difference will be a
Lr
ratio of horizontal Iengths
slightly changed set of exponents. These equations and the underlying criteria
hr
ratio of vertical lengths
are
Dr
ratio. of grain diameters
a. Friction Criterion: It has been pointed out in Refs. 2, 3 and 4 that the
Sr
ratio of slopes, particularly energy slopes
flow in an alluvial channel cannot be described generally by one formula with
Vr
ratio. of horizontal flow velocities
universal constants, but that it must be interpreted as a composite effect, the
(~ - f. ) ratio of sediment densities under water with
assumed to be
arious parts of which follow different and independent laws. It can be shown
s t r: equal in model and prototype
easily that no distorted model is possible if the friction criterion is formulated
tlr
ratio of llydraulic times
las an identity, i.e., if one stipulates that both the grain resistance (surface
t2r
ratio of flow durations (sedimentation time)
rrag) and the bar resistance (shape resistance) must both individually be simiqBr
ratio of bed-load rates (weight under water per unit of time and I~ar. Actually this is not necessary for an overall similarity as long as tile towidth)
fal friction behaves similarly.
It is proposed, therefore, to base the Ir-ict ional
ratio of total-load rates (weight under water per unit of time andsimilarity on the behavior of the entire section for which it assumes the form
Width)
pf a r-ating curve. As most rating curves give approximately a straight Line if
Vsr
ratio of sediment settling velocities
~hetotal discharge Q is plotted against the stages on a log-log sheet, it is asCr
ratio of constants C (generalized Manning equation)
sumed that the prototype and the model channel can be described by an equation
,?,.
ratio of Ri/RT
- values
,pf the form
c
Vz.
( 12 +"')
The large list of variables and scales already implies that a number of dl\
(1)
tortions will be contemplated.
I
0""
5 h

ff

566-2

.f2

566-3

which is a generalized Manning equation. Eq . (1) become~ iden~ical Wil'6t~51r..m.

odels, as it safe-guards similar flow and energy loss around structures
and
Manning equation by taking m " 1/6 and by using the r-elat.ionship n-vD
other channel irregularities .
If it is assumed that the same exponent m can be used to describe both the.
There appears to be a possibility that Froude criterion loses its signifjprototype and the model relationships,
at least for the most important range~.cance in deep rivers when the Froude's number becomes very low
of discharges, the equation can be written between ratios as
F = V2IjD
I). The only effect of velocity changes seems to be in
that case an energy loss which is taken care of elsewhere. No experience
Z
-I J,-/-2,."
0 bot C -z
= LJ
(4 exists on this part, however.
S,.,.
,."
.. v.
Froude law may be written in ratios as

t(

.I

v,.

n,.-llz

The various ratios on the left side of eq. (A) are explained preVlously: ~he~
V,.
= LlF
(B)
value .av equals unity if the similarity is exactly satisfied, but may mdlca~
a small deviation from the exact similarity if such a deviation becomes nee!l
sary for any practical reason.
~Here again stands AF " 1 for exact similarity while a deviation from unity
While V r, Sr' hr and Dr are ratios which will recur in other equations, mfIileasures a possible necessary deviation from the exact solution.
and Cr are the exponent and the ratio of the constants in generalized eq,
c. Sediment transport criterion:
In order to have similar sediment transthe form.
port conditions near the bed it is, ingeneral, necessary that both
and "f/;.
rr
are equal in model and prototype since the two are not connected by a powerV/ jRr5(1
C (Rr / Ks )
((typeequation. Only if the transport rates are restricted to a very narrow
[r.angeof values is it possible to combine the two conditions into one. The
.
!equality of the values
.
where RT is the hydraulic radius of the total section with the bottom width ~,
wetted perimeter and Ks the grain size of the bed representative
for its gr~.
roughness.
If one may assume similar grain mixtures are used in model asi
(4)
exist in the prototype, the ratio of the Ks values equals that of the grain siz\
D. The values C must be determined individually for model and prototype

(l)l

ffl:

:J

the ratio found for a representative

average channel section.
clature of Ref. 2 to 4
RT = AT/fi
( A~ -r Ab" -t- Aw) lfi
(Rb'
R;

p" + R;' Pb -t Rv.""",,)/

Using the no liSpossible for all fractions of a mixture only if the two mixtures are similar,
UCh that the ratios of the i-v~lues become equal to unity. With
equal fn
rOde I and prototype the equation of equal ~A -values may be written as

t..

Pb

It

(\

(()

1.>8r J,5 -

)-3/2

f:t r:

3e

(C)

Dr = /

-i: Rf," -t R"" P""/Ii

.

No deviation from this equation is usually desired.

The sediment rate q is
The model values of C and m can be determined by a trial-and-error
methop1easured in weight under water.
B
only, as they depend on the choice of the remaining scale ratios.
This procf d. Zero. Sedime~t-Ioad Criterion:
The equality of
-values in model and
dure will be explained in the example of a "Big Sand Creek" model.
prot~~pe 1Soften mt.er~reted by engineers as the condition of similar flow
b. Froude Criterion:
In open channel flows, Froude's law is one of the ,ond1hons at the begmnmg of sediment motion. It can be expreased as equal
criteria because it balances the gravitational forces against the inertia forCj:~ values.
.
This can easiest be expressed by the corresponding energies.
Since the en~ .
content of the flow may be divided into its kinetic and potential parts as velo;
(5)
ity head and depth, both are correlated by the equation
!
.
t

-v;

ConS+onf

Even if in river

er

order to analyze the s~nifi.cance of all the corrections outside the brackets,
t us remember. that
1~ different for the various grain sizes of a mixture.

channels the water surface slope is often very small, one ntr the larger
t

srzes, both m model and prototype,

has the value 1. In or-

remember t~at it. is the graVittatiOnal forllce ,whiClhmainttai~s tthe flow .Furt~er to have Similarity the ratio of
-values must be unity for all ~izes.
more, any rrver improvemen
plan usua y mvo ves cer am ypes 0 fr-iver e~.
l:.
.
strfction work by the construction of training walls, jetties, and groins alon!mc~ d 1S a function of nix, Dr must equal Xr. Referring again to Ref. 2,
the banks. At the constricted sections and especially in the neighborhood of e fmd that
structures,
the elevation of water surface may change rapidly. It is, theref~
advisable to include the Froude law as one of the criteria in designing river!

566-4

I
i

566-5

tor
tor

o7'7A

y =

and
with

A =

KsI.x.

This time ratio must be such that corresponding time intervale are required
by corresponding sediment rates qT to fill corresponding volumes. Expressed
(6)' in ratios this equation can be written for the unit width as

Ll/{J <: 1.8

.<1/~ 7 I.

t>

.x.

and

(H)

y(K.5/~)

:x. (K.s / ';,)

assuming the pore volume of the deposits to be equal in model and prototype.
.
....
The sediment rates qT are to be measured in weight under water. t2r is the
and with (Kslr Dr by definition, it c~ ~e se.en that I~ general ~lmIlar~ty .WI time scale at which the hydrographs must be repeated in the model.
be possible only if fJ = Dr' This COndlh?n WIll perm~t some_ alight devlaho
i. Independent ratios have been introduced in this study for length, height
r
however, since it controls some correcho~s only. WIth br - Dr we see tha1and slope. This implies that the model is not only vertically distorted but in
Xr = 1 and that LI,. "Dr. Thus we can wnte
. addition tilted. As the tilt is applied to the model during construction and as
it is assumed to be proportional to the prototype slope, it can be applied to
(Dlflows only which have at all points water surface and energy line slopes which
are constant with time. This condition is not fulfilled wherever the flow reverses direction, such as under the influence of the tide or in most overbank
'flows. In all such cases no additional tilt can be permitted, which is expressed
in ratios where the value 7. is defined by
by
(81
(I)

or as the ratio of the hydraulic radius referring to the surface drag to the ,e
tire radius RT' This correction must be introduced since RTr = hr' but Rbi ~N = 1 represents
,. hr'
fOf AN from unity.
e. Laminar sublayer criterion:
As 'has just been ~een, . ~r must usuallJ\
equal Dr to achieve similarity.
This can be written In ratios as

Dr

1r"z

h;

1/2

5r

Vz =

LJ~

zero tilt.

A small tilt is represented

Solution of the Similarity

by a small deviation

Equations

(E!

Table 1 gives 9 equations (A) to (I) which are products of powers of the variOUSratios equal to unity. The number sgfve the exponent of the various ratios.
.
.
..
LI
.
.,
\ There are 13 ratios, the exponent m and 46 - values involved. Of these
Agam, a shght daviation
~ may be permitted aspectafly m cases w~ere t~are 10 free ratios. The exponent m and the three ratios must be determined
bulk of the bed is considerably coarser than ~ and, therefore, not dlrec~IYby auxiliary calculations (as shown in the accompanying problem) while the
affected by its value. In the derivation of eq (E) it may be noted that ~ IS 6-values can be chosen.
'
pendent on the surface drag only, and thus must contain
=
L Between the 10 free ratios there are 9 equations to be satisfied, such that
which has a ratio 'lr I/Z fir V.e 5r '12. The viscosity Y is assumed to be equal pnly one of the 10 ca~ be chosen freely if the 6. -values are held rigi~lY to unimodel and prototype, but can be introduced differently if necessary.
~yo.r to some ot~er fixed val~e. If s~me. o~ the ~ -values are permitted to
f. By calculating for some characteristic
flows in model and prototype thClevIatefrom unity by a certam margm, It IS poaaible to choose two or more
ratio qT/qB the average ratio of these ratios qTr/qBr can be determined, iff the fr~e ratios .within certain li~nits...
.
may be called B. This value can then be used to give a general relationship . The mne ~quatlOns are solved m Table II m three different ways: fLrst,
between the two load ratios
,
~Ith the vertical scale hr chosen freely, then with the horizontal scale Lr
fhos~ll, and. finally, with the density of the model sediment chosen. One of the
-I
solutions WIll usually satisfy the particular need in designing such a model.
'tSr If/r 8 = I
(FII'.he.
exact solution is obtained with all 6. -values equal to unity. li some deElahons of 6. from unity are accepted, it is possible to choose more than only
Here in both qB and qT are measured in weight under water per unit of wido/n~sc~le arbitrarily.
This is sometimes necessary when the size of the modand time.
el Is given by the available space, while the sediment density is determined by
g. Hydraulic time tl may be defined as the time which a water particle t~he availabili~y of materials for the p.articular purpose. The des'ign
comlete set of distorted model scales WIll now be demonstrated on a Iicttttoua
to move with velocity v through a distance L
fo~el of Big Sand Creek, the channel of which was used as an example for
V -t
L t =
(csedlment load calculation in Ref. 2.
r
t rr
I

u,,..' J~'S1

0: ~

-r

I
h. A different time is the time t2 indicating the duration of individual flo
566-6

I
\

566-7

TABLE I
Model Laws for River Models with Sediment Motion
Eq.

:Lr

-1"'2m

~l

s.

<J.

(!j-ft)r

Br

tlr

~r

t2r

* C*r

13

-1 2m

*
l(r

Ll

. ..=

-2

Ll

Ll~

IN

Friction Eq.

-1

Froude

-2
-3

-1

-1

-3

1 signi:ficance

f
-1

<Jt

a>
a>

-2

D/J

00

-1

-1

-i

~l

I .

.*

B, Cr,

~l

De:finition
o:f tl

"1 r'

Suspended
Load Ratio

-1

Sediment
continuity

No tilt

-1

and m are determined from auxiltary calculations

- vaIues'may be chosen to suit the conditions.

~--.~~--

~-~~~---~~---~'-~~"""""""'"TJl1lI;~----"""""'"

Mode~ Ra'o:Losfor .Open Channel Flows w:L

th

1.~~"'*"~dit""dfl.,hi

Sed:Lment ~!ot:Lon

Solution with known effects of 'leaving' out eCJ.s.(A), (B), (E) and (I)
(a) Choose hr
Scale Ratio

<Jt

a>
a>

Symbol

Horizontal. lengths

4m+l
m+l

Vertical lengths

Flow velocity

'2

Slope

S
r

Sediment size

C
r
2
m+l

"tr
m

i1i+I

13

~.

Ll

A~

-2
m+l

-2m
m+l

L1v
1

m+l

chosen

-3m
mH

-2
m+l

-m
rn+l

2
m+l

2m
m+l

m+l

2in-l
2(lltt-l
)

1
m+1

-1
2(m+1)

-1
m+1

1
m+1

2(mt-l)

<)j-jf}r

3(1-2m)
2(m+l)

-3
mH

3
2(m+l)

3
m+l

2m-1
mH

-3
.2(m+l)

CJ.13r

3fl-j)
2 m+l

-3
m+l

3
2(m+l)

3
m+1

3m
iii+I

2(m+1)

"'rr

3fl-jl
2 m+1

miT

-3

3
2(m+l)

3
iii+r

3m
iii+r

2(mtl)

Hydraulic time

t1r

7m+1
2(m+1)

i1i+I

iii+r

-(m+3)
m+l

-2m
m+l

1
m+l

Sedimentation time

t2r

5m+2'
m+l

m+l

m
m+l

mrr

-1

~
S~diment density
under water

f-----

Bed-load rate per

unit width and tim~
by wt under water
Total-load rate per
unit ,'Tidthand tim~
by ,rt. under water

;?~'::"~~~~4~~:5.~;::;-~~-:"~,~,,_,

"'_"".;..:o'~'

:;:"::'::'~-::.:.::;..:.::::::::::"'~':!"!<

-1

-2

mtl

-(3m+l)

n;:;:r

-3

-3

---"=E~:::=::'",:,:"~~~~:7-:-~
r~:' i;':~:'!.:.='~-';;;:-~~~~~~~~.':~

"'I"~

,1

TABLE II
Model TIatios for Open Channel Flmrs ,-lithSediment Motion
Solution with known effects of leaving out egs. (A)] (E)] (El and (r)
(b) Choose L1"
I

Symbol

Scale Ratio

a>
Cl'l
Cl'l
I

Horizontal lengths

Vertical lengths

Flow velocity

~r

.(l

4~

L\

L\

chosen

r
-m
4mH

m+l
4m+l

-2
4m+l

m+l
2(4m+l)

-1
hm+l

-m
2(4m+l)

Slope

S
r

-3m
4m+l

-2
4m+l

-m
4m+l

Sediment size

2m-I
2(4m+l)

hm+l

3(1-2m)
2(4m+l)

'<:Br

2
4m+l

2m
4mH

-(m+l)

-1

""""4iii+:L

4m+l
-1
2(4111+1)

2(2m+l)

-(m+l)
2(4m+l)

2
4m+l

2m
4m+l

3m
4m+l

-(2ml-l)
2(4m+l)

-2
4m+l

2m+l
4m+l

I-2m
1
2(4mt-l) 4m+l

-6
4m+l

3(2m+l)
2(4m+l)

6
4m+l

2m-I
4m+l

3(2m-l)
2(4m+l)

-3
4m+l

3(1-2m)
2(4m+l)

-6
4m+l

3(2m+l)
2(4m+l)

6
4m+l

6m
4m+l

}(2m-i)
2(4m+l)

-3
4m+l

'1rr

3(1-2m)
2(4m+l)

-6
4m+l

3(2m+l)
2(4m+l)

6
4m+l

6m
4m+l

3(2m-l)
2(4m+l)

-3
4m+l

Hydraulic time

tlr

7m+l
2(4m+l)

4m+l

m
2(4m+l)

Sedimentation time

t2r

5m+2
4m+l

-2
4m+l

-m
4m+l

....

Sediment density
under \later

(fs -Jj}r

Bed-load rate per

unit width and time
by ,rt. under water
Total-load rate per
unit llidth and time
by ,rt.under vat.er

.,"

'ct

-1

"""4iittl I+m+l

-2(2m+l) -m
4m+l
4m+l
2
4m+l

"'.'

-(2m+l)

m+l
1
2(4m+l) 2(4m+l)
-(m+l)

-1

4m+l
. .cfi

};

"":

-1
4m+l

"

TABLE II
Model Ratios for Open Channel Flows .,ith Sediment Motion
Solution with known effects of leaving out eqs. (A), (B), (E) and (r)
(c) Choose (j'5-ff)r
I

Scale Ratio

c.n
Cl'l
Cl'l
I

Symbol

(f5 - ft)r

cr

'lr

AF

2m+l
2m-l

4
2m-I

"3

-2
2m-I

2
2m-I

"3

2m-I

2m
2m-I

'3

2(2m-l)

-2
2m-I

2m-I

Horizontal lengths

-2(4m+l)
3(2m-l)

-4
2m-l

Vertical lengths

r'

-mHj
3 2m-I

2m-l

2m-I

Flow velocity

-(m+l)
3(2m-l)

-1
2m-l

2(2m-l)

Slope

S
r

2m-I

Sediment size

,...
....
Sediment density
under Hater

2m

-1
r

Hydraulic time

tll;'

Sedimentation time

t2r

Ll

-1

-1

2
2m-I

-2m
2m-l

2
3

--

chosen

(~-ft)r

Bed-load rate per

unit ,;idth and time qBr
by Irt. under Hater I
Total-load rate per I
unit ,fidth and time
'1rr
by lleight under water

-2

4~

-(7m+l)
3(2m-1)

-3
2m-I

4m+l
2(2m-l)

-2(5l;l\+2)

_6
2m-I

2(m+1)

3(2m-1)

2m-l

2(2-m)
2m-1
(,

-1

2m-I

:5

"3

-3
2(2m-1)
-3
2m-l

"',""""','

'Il""'.

~
The Big Sand Creek Model

hr =
First" the friction conditions in the prototype are summarized in Table ill
based on Table 6 of Ref. 2. With S " 0.001050 and D65 " Ks = 0.00115 ft. the
following values are obtained.

S.r

Lr~

r:

Friction Conditions - Big Sand Creek

"
~

R.r

ft

ft

ft

ft/sec

050

0.86

130

292

0.384

0.00515

1130

0.00198

435

409

1.00

0.76

192

4.44

0520

0.00330

1670

0.00172

870

1110

2.00

050

306

6.63

0.653

0.00236

2660

0.0015).j.1740

3710

300

030

4.40

8.40

0.682

0.00211

3820

0.00144

2610

9160

4.00

0.14

573

992

0.698

0.00197

4980

0.00137

3480 19850

500

0.07

704

1130

0711

0.00186

suo

0.00132

4350 35700

'lp= ~/R.r

R.rSg

VKs

VT

1\/K;>

vZ

From these ratios results

S. =
M

Q.Z5

-1.147

ISO

0.25

0.827

(11)

z{iiM+ij _

_0.113

,,/41

-150.

0,/jZ7

o,3z6

the following model

P
.Is,.,
=

5""a/c
v /r c.e ,

/.0

The model h~draulics of the first approximatioI!. is then given by Table IV. R'
values used In Table IV correspond to those used in the prototype using nIb
= 37.3.
(""r
Table IV

A similar hydraulic pattern for the model is now obtained by trial and error. First the length scale Lr = 150 is chosen on the basis of available room
Then the prototype friction is plotted in Fig. 1 as ~
against ~/Ks
(~
Sg)
(o-symbol) :md the poin~s for. the higher discharges approximated by the line~
A(ifA(I)
WIth the relationshtp
"
r

Hod e 1
I

e;

v2
RTSl

il

Next,.th. model resrstance

Is

o,31;:!

(9)

= 21.05 ( Ks )

estimated to ',allOWthe line B(lfB(ll with th.1

v2

.3G176

RTSI
From these tworelattonshtps

7
(Rr)O.3
Z
-

Y(r =

0.0134 0.0425
0.0268 0.0601

D65

- First Approximation
x

Ll

C/VI

1 and all Ll -values

C = 0.827

I,

566-12

Log1OI\

v
,-

y'

0.00289

1.22

1.59

0.00221

1.872

0.458

302

0.00204

172
2.44

1.46

0.00241

2.135

0.738

151

270

1.27

0.00277

2376

1.160

072

51.0

298 1.18
3.44 1.14

0.00298

2520
2.630

151
1.82

050

0.00309

3.86

0.00320

2712

2.10

030 240

0.0536 0.0850
0.0805 0.1042

0.00118

0.1072 0.1202

0.00102

0.00145

0.1342 0.1348 .0.00091

ft

1.10

(12.27 A)

ft/s

u*
16.80

870
038 150

t"

(9) and (10) one may find the values of C

=5.55

ft

Table IV
(cont'd)

(10\

Ks

m:

Assuming

ft

u*
ft/$,

Hyd rauli cs

equation

Cp = 4.595

1\

".,= 0./86

"

1\"

1\

f~/s

ft

ft

stage

0.0273 0.0055

0.0189

4.010

0.0273 0.0055

0.0323

4.034

0.0227 0.0038

0.0574

4.086

0.0174 0.0022
0.0121 0.0011

0.0827

4.172

0.1083

4.260

0.0087 0.0006

0.1348

4340

28.0

= 0.00.4'"~

0.00109'

37.3

-(Z,"+')

'1r

-3

0,.

_/.,47

. 0.1)27

_o.~Z

4",+,

r:

cfs.

(,

C ~

Ib

T]~

r:

2,"-1

1.50

-"..,

C ~

Z{4MH)

0.68

n,,4..,,+-'
\..

-z

-3m

D. = L

-ITl

C,. ~

= Lr 4",+,

Table III

-"

-2

rn-f-t

R.r

R.rSg

ft

'7'

Ib/I\r=

Ks

'1.,.,

0.020 0.0129 568

0.0375 0.0093 10.64
0.075 0.0076 213
0.1240 O.007~ 352
0.1650 0.0068 46.4
0.2000 '0.0061 56.8

566-13

RT

>(p

'?r

0.670

0384

0573

0714

0520

0730

0716

0.653
0.682

0911
1.050

0.650
0.650

0.698

1.073

0.672

0711

1.060

The values of the wall friction, of stage and RT are determined by a trial and .
error method given in Table V and figure 2. First, the model bank roughness
must be chosen. With the prototype banks rather rough, it is assumed that the
Manning formula applies in model and prototype.
The wetted perimeter of the;
banks was assumed in the prototype calculation (Ref. 2) to equal twice the wa.. '
ter depth. In order to make the wall friction assume an equivalent part of the
total friction in model and prototype, we write
(12)
(13)

Determination

l\,

of ~

Stage

R
w

Table V
by Trial and Error Method
P

Pb

ft

:ft

ft

ft

ft'

0.0189

0.0332

(4.000
(4.050
(4.100

050
0.86
1.22

0.07
017

0.0323

0.0672

(4.000
(4.050
(4.100

050
0.86
1.22

0.0574

0.1340

(4.050
(4.100
(4.150

0.0827

0.l985

0.1083

0.134B

PR

Pb~
ft2

\vw

At

ft2

ft2

.00945
.0162
.02285

.0023
.00565

0.07
0.17

.01615
.0278
.0392;

.0047
.0114

0.0162
0.0325
0.0508

0.86
1.22
157

0.07
017
0.27

.0493
.0700
.0900

.0094
.0228
.0362

0.0587
0.0928
0.1262

(4.100
(4.150
(4.200

1.22
157
195

017
0.27
037

.1010
.1300
.1612

.0338
.0536
.0735

0.1348
0.1836
0.2347

0.264

(4.200
(4.250
(4.300

1.95
236
2.85

037
0.47
057

.2110
.2560
3085

.0978
.1240
.1508

03088
03800
0.4593

0326

(4300
(4350
(4.400

2.85
336
387

057
0.67
077

3840
.4530
5220

.186
.218
.251

0570
0.671
0773

AT= Ab Pt; -t Rw P"" for various fictitious values of the stage. These AT
values for each Rb value are connected by a dashed line .. The intersection
with the solid AT curve indicates the actual stage at which the Rb value occurs.
These values of RT are used in Table IV to determine 1,- for the first approximatlon. Also RTSg/V2 is calculated and plotted in Fig. 1 against RT/Ks' It
may be seen in this graph that the first assumed line B(l) - B(l) was somewhat too high and too steep to describe the model points.
Thus. a second approximation of a line A(2) - A(21 for the prototype and parallel to It B(2) - B(2) for the model was tried. The ine A(2) - A(2) does not
cover the two smallest prototype discharges, is therefore not correct but at
the larger flows. This is a minor defect, however, as most of the transport
and of the bed changes occur at the medium and high flows.
The second approximation is based on A(2) - A(2) for the prototype friction
R'

= 44.5 (

0.0095
0.0185
0.0285

=52.6

From these one obtains

I'YJ

The same formulas

7r =

0.1.45

hr

- (
w

(15)

Rr)o.Z~O

(16 )

/.0

10r

L~ -

/so

(17)

give

=4Z.25
= O.;UjZ

Dr = o.eee:

(Hl)

(r5-!!)~ = 40
6.5

V.r: --

This results

in the following model dimensions

5,.., =

as quoted in the first approximation

5r

(141
noN.= ~

'

Vr -

It can thus be calculated

s.zeo

:.s

and on B(2) - B(2) for the model friction

.L:
R""r
l/3 S. 'IZ
,
noM
r.
we ge t nWr by
Introducing previously determined values and ,ar = 1:
In the Manning equation for the banks

K"')

- 2.305 and
~

V"w
t.4865vz

J~-

w,.,=

0.050 = 0 0217
2.305
.

Fig ', 2 gives first the curves of Pw' Pb. ~ and AT as functions of the stag
all denved from corresponding prototype curves and Similarity scales (solid
curves), For each calculated value of Rb Table V'then determines values of
566-14

0.0039>4 ff.

035..,= 0.00''2 ~

and

vi\o~
.9.34

06.5..,=

0.00;372

f'sM

1.04/

;JT-/c.c-.

(19)

A new calculation of the hydraulics on this basis gave the (x) points of Fig. 1
which still follow with sufficient accuracy eq. (16). The results of eq. (18) and
(19) are thus assumed to be usable together with
566-15

r :-

,-i

L,.

150

So far, the equations A, B, D, E and I have been used; eq. G gives

1r

= t.; v,.. -I =

and eq, C gives

~
j

I'

2.3. I

specific gravity of 1.045 can be obtained. What would be the effect of su~h a
. deviation from the required values?
This can easily be found by the enoree
of one of the b. -values which appears to be least critical in the particular
11 problem for the absorption of the deviations.
The system of equations is
solved again for the assumed values of Lr and (fs - ff)", introducing the chosen !:::. as additional variable.
One finds then that an entirely new system of
scales will result together with the magnitude of the deviation which the chosen L'l.-value must undergo to satisfy the remaining equations.
Reliability

of the Method

The reader has by now probably received the impression that all the complications involved in the method have had no other effect but to make the system of similarities more and more unreliable.
Such a statement is definitely
The value of qTr is determined by eq, F which calls for the knowledge of the
unfair. But what the method does is to show that a distorted river model is in
ratio B. These are determined in Table VI which lists first the calculated
the best case an acceptable compromise which will permit the solution of cerprototype total load Z 'r Qr in tons per day, as determined in Table 8 of
tain problems which otherwise can not be solved except by experimentation
in
Ref. 1.
the prototype, which is under all conditions more expensive. This method of
designing a model has one great advantage over all other such methods which
Table VI
are available at the present: It permits the prediction of its reliability at
least in a qualitative way, with respect to the choice of ). -values it gives such
Determination of ratios B
deviations even quantitatively.
Fb (X ,'T6lr)p (Z':r1"r)p (-ZOT't"')M (Z;':T1rt/(E'~ 1-s)p t2 1 year
What are the most important reasons for the loss of similarity?
p
duplir
1. In the description of the channel friction the modified Manning equation
Lbs . under 6:.,.'lT>"1/(E,,1~M
Lbs , under
tonS/day
ft.
cated
water per
can not be expected to describe with equal m-values both model and prototype.
water per
in hours
; B
sec. ft.,
sec. ft.
conditions over the entire range of discharges.
It will be necessary,
there1.89
1.)6
4650
fore,
to
permit
deviations
in
the
friction
relationship
for
the
less
Important
0.00113
09042
100
05
flow conditions. This fact is most important if a large r ange of flow conditions
2.62
1.89
3350
0.00555
0.420
1.0
3190
exists in the problem area.
1145
502
363
0.0184
2.61
2.0
33600
2. If flows must be considered in more than one channel, it becomes even
8.15
1015
0.0)86
590
156000
9.11
30
more difficult to find a friction equation which describes all flows.
11.28
118
8.15
3. The total load rates must be used to determine the time ratio t2 for flow
0.0685
4-.0
538000
223
durations. Since the bed-load rates must be made similar to obtain similar
l4-.65
10.6
598
0.1090
46.2
143)000
50
bed configurations and since the ratio between bed-load rates and total-load
rates changes with the stage, a sliding time scale appears to become necesThe deterFrom a similar table the corresponding-model
values have been calculated an sary in many cases, especially if the range of discharges is large.
mination of this scale becomes somewhat indeterminate if flowin more than
are given in the next column. The ratio of (ri(31Je)pl(:E~8 ~&)M is equal to
one channel is important.
qBr since corresponding iB values are equal to prototype and model.
4. If the deposition of sediment in low velocity areas, such as overbank or
Eq. F and H permit now to determine t2r as
ill reservoirs,
is an important phase of the problem under investigation. the
634Q
wash load must" be introduced into the model. Its characteristic
is not given
8
by the ability to be transported as bed load, however, but by its ability" to stay
ill suspension permanently
in the flow of the overbank areas.
Because of these difficulties it is absolutely- necessary to verify any such
This time scale changes much with the stages and it appears necessary to ex model and its scales. Such a verification consists in the reproduction of a
tend the flood stages percentagewise over the low stages, a fact well known to knownprototype development in the model by similar flows. Only if such a
the practical experime_nter.
i1verification is possible and successful can the model be depended upon for the
prediction of future developments.
It is common experience that the gr-eatest
Significance of the ). -Values
difficulty of most model studies of this type is the gathering of the necessary
prototype information for the construction, the operation, and particularly,
the
No use has been made so far of the.6. -values.
Their Significance may bes verification of the model. But it is felt that this method of designing model
be seen starting Ir om the solution found as the second approximation.
In this scales is able to shorten the time consuming trial and error method of finding
system a material is required for the model sediment of a specific gravity of the proper model scales so much that more effort can be spent on a. re lfahle
1.041. Let us assume no such material is available, but a material with the
verification and the determination of the underlying river Informatlon,
1

566-16

566-17

friction factor in the Manning's equation

ACKNOWLEDGMENT
This work represents results of research carried
gineers, Missouri River Division, Omaha, Nebraska,
University of California.

out for the Corps of En

under contract with the ~-Ib

friction factor

(Manning) of the banks

wetted perimeter

of the bed

wetted perimeter

of the banks

flow discharge
REFERENCES
1. Conformity Between Model and Prototype:
Vol. 109, 1944, p. 134.

A Symposium, Trans. ASCE,

.qB
qT
~

2. The Bed-Load Function for Sediment Transportation

in Open Channel
R'
Flows, by H. A. Einstein, U. S. Department of Agriculture, Technical Bul ;;'b
letin 1026, Sept. 1950.
Rb.
3. River Channel Roughness, by H. A.Einstein
ASCE, Vol. 117, 1952, pp. 1121-1146.

and N. L. Barbarossa,

TranS't--R.r
R

4. Linearity of Friction in Open Channels, by H. A. Einstein and R. B. Banks"S

Association Internationale d'Hydrologie SCientifique, Assemblee Generalel
de Bruxelles, 1951, pp. 488-498.
tl

AT

total area of a cross

V
V
x

section

load in cross section

hydraulic

rate

radius with respect

to the grain

hydraulic

radius for channel irregularities

hydraulic

radius of the total section

hydraulic

radius with respect

to the bank

sedimentation

time

shear velocity with respect to the grain

shear velocity for channel irregularities
horizontal

time

flow velocity

settling velocity of a sediment particle

parameter

for transition

smooth - rough

Aw

part of the cross

A'

cross-sectional

area pertaining to the grain

pressure

cross-sectional

area pertaining

the thickness

qT/qBr
constant in the generalized

grain size

~/RT

D35

grain size of which 35 percent is finer

kinematic

D65
g

grain size of which 65 percent is finer

"hiding factor"

gravitational

density of the fluid

vertical

fraction of bed material

~
~

fraction of bed load in a given grain size range

intensity of shear on representative

fraction of total load in a given grain size range

intensity of shear for individual grain siz e

section pertaining to the banks

total-load

slope

*
List of Symbols

corresponding
total sediment

hydraulic

5. Laws of Turbuient Flow in Open Channels, by G. R. Keulegan, U. S. Nation t2

al Bureau of Standards, Rept. 1151, 1938, p. 738.
u'

APPENDIX

bed-load rate in weight under water per unit of time and width

to irregularities

characteristic

the apparent
Manning's equation

acceleration

lengths

grain size of a mixture

correction

in transition

smooth - rough

of the laminar sub-layer

roughness diameter

deviation from the Similarity

law "a"

viscosity
of grains in a mixture

density of the solids

in a given grain size range

intensity of transport

roughness diameter

horizontal lengths

exponent in the generalized

for individual grain size

Subscripts

Manning's equation
566-18

indicating ratio
refers to prototype
refers to model
566-19

particle

4.3 I

If

t-

u,
c

~.2

l.Ll

~
c.n
a:>
O"l
I

t-

Cf)

4. I

--AT

= Pb Rb+

,...

Pw Rw

4.0

IL

I II?

0.1

0.2

0.3

0.4

0.5

AT

0.6

07

in SO. FT.

0.5

/.0

1.5

2.0

2.5

3.0

3.5

Pb
Fig.2

Big

Sand

Thp

Creek

0.8

and

Model-Trial

Determination

Pw

in

FT.

and

Of R~

Error

Solution

For

4.0

II

AMERICAN SOCIETY OF CIVIL ENGINEERS

OFFICERS

FOR

1955

llQOCIEIEIDI N4GS

PRESIDENT
WILLIAM

ROY GLIDDEN

'\,(~CE'PRESIDENTS
Term ex-pires October, 1956:
FRANK
L WEAVER
LOUIS R, HOWSON

Term expires October, 1955:

ENOCH R, NEEDLES
MASON G, LOCKWOOD

AMERICAN

Term

expires

October,

1955:

expires

October,

WILLIAM

CHARLES
B. MOLINEAUX
MERCEL
J ' SHELTON
A. A. K. BOOTH
CARL G. PAULSEN
LLOYD D. KNAPP
GLENN
W. HOLCOMB
FRANCIS
M. DAWSON

OLIVER

Term

1956:

S. LaLONDE.

expires

JEWELL

JR.

October,

FREDERICK

W. HARTWELL

GEORGE

CIVIL

ENGINEERS

H.

DECEMBER, 1954

PAULSON

S. RICHARDSO~

C. SHEDD

SAMUEL

B. MORRIS

DON

ERNEST

W. CARLTON

GRAHAM

M. CORBETT
P. WILLOUGHB

LA WRENCE

F. DAWSON

1957: .

M, GARRELTS

THOMAS

RAYMOND

SOCIETY
OF

DIRECTORS
Term

SEPARATE No. 567

VOLUME 80

A. ELSENER

PAST,PRESIDENTS
Members
WALTER

L.

of the Board
DANIEL

HUBER

V. TERRELL

FLOW EXPANSION
EXECUTIVE

TREASURER

SECRETARY

WILLIAM

N.

ASSISTANT

CHARLES

CAREY

E. TROUT

ASSISTANT

SECRETARY

PROCEEDINGS

RECOVERY IN FLUIDS

TREASURER

CARLTON

E. L. CHANDLER

AND PRESSURE

by Harold Tults

S. PROCTOR

OF THE SOCIETY

HAROLD
T. LARSEN
Manager of Technical Publications
DEFOREST
Editor of

A. MATTESON,
JR.
Technical
Publications

Assoc.

HYDRAULICS DIVISION

PAUL A. PARISI
Editor of Technical Publications

{Discussion open until April 1, 1955}

Copyright

COMMITTEE
SAMUEL
JEWELL
GLENN
ERNEST

W.
W.

B. MORRIS,

M. GARRELTS.

1954

Printed

ON PUBLICATIONS

by the
in the

AMERICAN

United

SOCIETY

States

OF

CIVIL

0/ America

Chairman

Headquarters

Vice-Chairman

HOLCOMB

OLIVER

CARLTON

DON

W. HARTWELL

of the Society

33 W. 39th St.
New York 18. N. Y.

M. CORBETT

PRICE $0.50 PER COpy

ENGINEERS