12293 AUGUST 1976 WW3

JOURNAL OF
THE WATERWAYS
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

HARBORS AND COASTAL

ENGINEERING DIVISION

SIMILITUDE IN COASTAL ENGINEERING

By Bernard Le Mehaute, 1 M. ASCE

INTRODUCTION

The purpose of this paper is to examine a general engineering approach to

· the concept of similitude in scale model technology with the emphasis on coastal
engineering applications. By engineering approach, it is meant to distinguish
it from the more classical and formal approach that is based on dimensional
analysis, as best presented in Yalin (13). Langhaar (3), and Ledov (4), for example.
In the formal approach, similitude is usually presented as a natural consequence
of dimensional analysis . Then the similitude of Froude, Reynolds, Mach, Weber,
Cauchy, Richardson, etc., are presented.
Actually, the formal approach to similitude based on dimensional analysis
· has never been adopted by engineers in its entirety. For example, scale models
built according to the equality of Reynolds number or the equality of Weber
number have of course never been built, just to point out a practical limit
to the application of dimensional analysis to scale model technology. From
an engineering standpoint, the Reynolds similitude does not exist in coastal
engineering and in general , in scale model technology.
The formal and engineering approaches are not incompatible. However, a
difference of emphasis does exist on the relative importance attributed to various
parameters. Also, the more pragmatic engineering approach allows the engineer
to use the tools of scale model technology beyond the limits allowed by the
formal approach .
This paper is essentially written for the project engineer who has to decide
whether a scale model experiment is worthwhile, to understand the basic principle
at stake, and take engineering decision based on the results of scale model
experiments. The project engineer does not really have to be familiar with
Note .-Discussion open until January I, 1977. To extend the closing date one month,
a written request must be filed with the Editor of Technical Pub!ications, ASCE. This
paper is part of the copyrighted Journal of the Waterways, Harbors and Coastal Engineering
Division, Proceedings of the American Society of Civil Engineers, Vol. 102, No. WW3,
August, 1976. Manuscript was submitted for review for possible publication on November
5, 1975.
'Sr. Vice Pres., Tetra Tech, Inc., Pasadena, Calif.

317

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 318 AUGUST 1976 WW3
detailed laboratory procedures . Very often , he is not even a specialist in hydraulic
engineering but rather a civil engineer trained in a multiple number of di sciplines.
Much misunderstan ding about scale model technology- overconfide nce or
disbelief-is to be linked to a formal approach to the concept of s imilitude.
The ideas expressed in thi s paper are not new . They are based only on common
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

a nd well-established practices amongst laboratory engineers. An attempt at

rationalizing these practices will then be found.
Since most of the contribution on similitude found in the literature follow
the formal approach , the Task Committee on Coastal Engineering Methods of
Investigation has found it pertinent to present this engineering approach to the
profession. Hopefully, it will lead to more detailed presentation s and analyses
on specific problems .

SIMILITUDE AND CHOICE OF MODELS AND SCALES

A scale model must fulfill the following conditions: (1) It must be exact
(i .e., it must reproduce with sufficient exactness the natural phenomenon under
study); (2) it must be consistent (i.e., it must always give the sa me results
under the same conditions to the required precision); (3) it must be sensitive-o r
more exactly, its sensitivity has to be imposed by the needs of the reproduction
of the phenomenon under investigation ; and (4) it mu st be economical , of
reasonable size , and completed within a reasonable time interval.
Scale model technology is based on similarity between two phenomenon s
at different scales . It is an engineering art in which a compromise is made
with the formal laws of similitude , given by dimensional a nalysis, to build a
practical tool with which engineering problems can be solved .
The advantages of scale model versus theory are:

1. Complex boundary conditions cannot be analyzed by analytical means .

Numerical solutions and high speed computers extend the power of analytical
methods used, but scale models are the best analog computer.
2. Nonlinear effects are the greatest source of mathema tical difficulty. It
is true that many problems can often be linearized, but the great advantage
of scale model study is the reproduction of not only the linea r forces in similitude,
but also ~he nonlinear convective inertial force , p 'v ( V 2 / 2). The force , p V
X V X V , may not always be reproduced in similitude , as the distribution
of rotationality, V X V, is often related to the distribution of friction forces ,
which cannot be reproduced in similitude exactly. This scale effect is often
negligible, as will be seen in the following.
3. Fully turbulent phenomenon s are grossly in similitude as also detailed in
the following .

Generally , a scale model does not reproduce all the aspects of the phenomenon
under investigation , but models a few that are of interest in the research . A
knowledge and understandin g of the physical laws that govern the effects studied
permits us to determine the relative influence of "scale effects." (Scale effects
are the error due to the unsatisfactor y reproduction s of some phenomenon s
due to too small a scale.)
The scale is chosen as a compromise between economics on the one hand

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 319

and the technical requirements for similitude on the other hand. From the
economical viewpoint, the cost of scale model experiments increases approxi-
mately as the cube of the scale, A, since the area of the model increases as
1- 2, and the theoretical duration of the test increases as A 112 , but the actual
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

duration increases faster. Of course, this rule of thumb has many exceptions,
•. and is not valid if the scale of the model is too small. (In the case of a very
small model, the cost of the operation actually increases as the difficulty of
carrying out reliable measurements increases.) The "best" scale is not the largest
possible one since the scales of models must primarily be dictated by economical
considerations.
Actually, the decisive elements rarely present themselves in a precise manner.
. It usually is almost impossible to evaluate, even roughly , the "economic
expectation" of a model and particularly the variation in this economy with
respect to the accuracy of the tests. Thus, the determination of the scales
of optimum reduction will necessarily be an art rather than a science, but the
value of the results obtained will nevertheless primarily depend upon the overall
knowledge available to the designer. Among other things, it is imperative that
the engineer be acquainted with the laws of similitude because not only will
' they enable him to estimate the degree of precision that he may expect from
the model, but, above all, they will render it possible to have a sound concept
of the latter and to achieve as much as possible a precision compatible with
a given sum available for research.
From an engineering viewpoint, the first stage of the study of similitude
is not concerned with a detailed analysis of the equations, but is essentially
an examination of the problem as a whole and of its physical character (1).
For example, it is impossible to form a sound concept of a model without
· a priori having an idea of the causes and origin of the phenomenon studied.
The overall examination of the phenomenons involved enables us, a priori,
to eliminate basically erroneous methods of approaching the problem. It renders
it possible to submit and broad outline of the boundaries of the model. In
many cases, it also will enable us to determine the principal stages of the study
and occasionally, the number and type of the models required. The detailed
study of the equations of the phenomenons will then define the approximations,
scales, distortions, etc ., and eventually , the number of models required, as
well as the rules of similitude to be applied to each of them. For example ,
in the case of a conventional study of the protection of a port against wave
action , it may happen that the installations as a whole are studied on a scale
of I/ 150, some particularly important structures (such as those at the entrance)
on a scale of I / 75, and the stability of the individual structures on a scale
, of I/ 40. Models of rivers and estuaries may be distorted . A model is distorted
when one of its scales (e .g., the vertical scale) is different from the horizontal
scale. The value of D = µ/A is the rate of distortion. Whenever possible (or
necessary as seen in the following), distortion allows considerable space and
cost savings.
In summary, the problem of similitude is far from simple, and there are
no standard solutions. The engineer has at his disposal a number of complex
· means that he must be able to handle with skill. The better he is acquainted
with the tools, the more effective and economical the solution will be.
Now, from the technical viewpoint, the problem consists of examining when

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 320 AUGUST 1976 WN3
scale effects become important. This leads us to consider the f_orce~ that a_lways
need to be in similitude . Each separate case needs to be exammed m detail.
The law of similitude can be deduced from the law of motion under a differential
form or an integrated form. The dynamics of a system obey the general law,
I+ G + P + F + E + C = O, in which I= the inertial force; G = the gravity
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

force; P = the pressure force; F = the friction force ; E = the elastic force; and
C = the capillary force. These forces could be expressed vectorially, or on
projection axis, in differential form, or an integrated one .

= G' =. .=
On the scale model one has I' + G' + P' + F ' + E' + C' = O; for
similitude , I/ I' G/ C/ C' . Once all these forces are mathematically
expressed, the conditions of similitude are obtained. It is then realized that
complete similitude is impossible. Then it is important to know what can be
neglected and what are the scale effects. Some general rules are now presented.

SHORT MODEL-GENERALIZED fROUDE SIMILITUDE

There are two kinds of scale models for which the law of similitude is deduced,
short models and long models. Short models are either eonsidered as nondissipa-
tive or fully turbulent. Nondissipative short models are the ones where the
flow pattern is essentially governed by inertia force and gravity only in the
case of free surface flow, or inertia forces and pressure gradient only in the
case of flow under pressure .
The ratio of inertial force (dimensionally equal top V 2 ), to gravity or pressure
force (dimensionally equal top g L) is the square of a Froude number, V 2 / gL ,
which is to be the same on a scale model as on a prototype. An example
of a phenomenon which can be modeled in this way is the flow over a spillway,
where boundary layer effects are negligible.
Also, the inertial force and the gravity and pressure forces are always present
in water wave problems, and the Froude similitude applies when viscous and
capillary effects are negligible .
From an equality of Froude numbers

;: I p = ;~ I, , · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·(I)
in which p = prototype and m = model, and the approximate constancy of g
on the earth's surface , one deduces

(::r=~:=L . . . . (2)

in which X. = the geometric scale. Therefore, the ratio or scale of velocities

is '\J">: . Accordingly, the ratio for time or time scale is also YT: since
Tm / Tp = (L 111 / V m) / (LP / VP) = X./YT:-. The ratio of discharges is like the ratio
~f areas _(X. 2 ) ti1:1es ratio .of velocitie~ YT:, i.e., X. 512. The ratio of powers
1s ~he ratio of d1schar~~ times the rah~ of height (if Pm= Pp), i.e., x_1 12. The
ratio of angular veloc1t1es or frequencies, 1/ T, is X. -o /2J, etc. Since the ratio
of pg is unity, the ratio of pressures is X., and the ratio or scale of forces
(area x pressure) is X. 3 , the ratio of powers defined by force times velocity
is also X. 7 12 •

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 321
These relationships also apply to flow under pressure. The Froude parameter
is then replaced by p V 2 I6. p or 6. p / (p V 2 ), in which 6. p = a pressure gradient
dimensionally equivalent to pgL. The value of 6.p/(p V 2 ) is sometimes called
the Euler number. The absolute pressure value is irrelevant as long as the
pressure gradient is in similitude, i.e., Cl.pm= X.6.pP. However, in the case of
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

· cavitation, the similitude parameter requires the atmospheric pressure, pa, to

be reduced to a scale model partial vacuum.
Let us now consider the viscous forces. It is known that the ratio of inertial
force (or more exactly the gradient of kinetic energy) to viscous force is
dimensionally represented by a Reynolds number, VL/v, in which v = the
kinematic viscosity; with the same fluid, an equality of Reynolds number and
Froude number is possible only at scale unity. Thus, Froude similitude, at a
scale smaller then unity, is possible in two cases only: (1) Where viscous forces
are negligible, as in a gravity wave prior to breaking or in a flow over a weir;
or (2) where the flow is very turbulent and the flow pattern to be reproduced
on scale model is short, as in a hydraulic jump, a breaking wave , or a sudden
enlargement. These also are short models, even though dissipative. Indeed,
in these latter cases, the dissipation of energy is mostly due to turbulent
fluctuations and is not due to laminar viscous effect. While these viscous effects
are linearly related to the velocity, the turbulent fluctuations are quadratic,
i.e., proportional to the square of the average velocity , V 2 , as are the inertial
forces. Thus , the ratio of dissipative forces to gravity force in a very turbulent
flow is also a Froude number squared. This situation allows us to use the
"generalized Froude similtude." Of course, the depth at which the air bubble
penetrates in a breaker will be relatively larger in the field than in the model,
as the size of the bubble (determined by capillary effects) is approximately
the same in any air entrainment phenomenon. But the dissipative forces are
in similitude and proportional to V 2 , and the total amount of energy dissipated
is in similitude (X. 7 12 ), even though the fine structure of the flow could be
different.
This is evidenced when one considers , for example, the shearing stress

= p u'v' + µ, a'i", = p /2 (au ) + µ, au

aJ aJ a; . . . . . . . . . . . . . . . . . .(3)
2
T

in which fi = an average velocity; µ, = the viscosity; p = the density; 1 = the

Prandl mixing length; and u' and v' = the turbulent velocity components.
In the case where µ,(afi/ay) << p 12(afi/ay), it can easily be seen that the
shearing forces, T, will be in similitude provided the mixing lengths are such
that /' = X.1 in addition to the similitude of Froude. This condition is fulfilled
in the case of fully turbulent motion occurring over a short distance and presenting
a large velocity gradient (9).
An example will illustrate these considerations. The rate of energy dissipated
by a hydraulic jump or a moving bore, dE/ dt, depends only upon the discharge,
Q, and the depth of water before and after the jump (h I and h 2 ) . The rate
of energy dissipation by an hydraulic jump is (11)

. . . .............. . .. . ....... . (4)

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 AUGUST 1976 WN3
322
independentl y of the scale . Therefore , the ratio [(dE / dt)ln.J ! [(dE /
dt) Ir] = x. s 12 (X. 3 / x. 2 ) = x. 7 12 , which is in accordance with the Froude similitude,
since the rate of energy dissipation is a power dimension.
Thus , a small hydraulic j ump, defined by h;, h;, Q', is in similitude with
0

a large jump defined by h; = X. h 1 , 11; = X. h 2 , Q' = X. 512 Q, at a scale, X.. Indeed ,

Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

the phenomenon is very turbulent. The average flow pattern is the same. The
turbulent fluctuation may statistically be different, but the gross effect remains
the same. It is the reason why stilling basins, for example , could be investigated
by scale models. In the case of breaking waves , scale effects will be negligible,
provided the height of breaker is larger than a critical value which is a function
of the phenomenon s to be investigated. This subject will be further investigated
in the section on scale effects. (It is pointed out that the phenomenon of wave
breaking on beaches is not due to the "drag" of the bottom a s it has been
written in vulgarization books . Otherwise, this phenomenon would be subjected
to scale effects. The total rate of energy dissipation which is a function of
water depth variation, is partitioned between bottom dissipation and breaker
dissipation. Only on a very gentle slope and at a very small scale, a high viscous
bottom dissipation tends to limit the rate of energy dissipation by breaking.)
From the most general viewpoint, it is known that the a verage fluid flow
obeys the Reynolds equation:
au; _ au; a(p + pgz) a 2 u, a __
p-- + p u . --= ------ +µ--- - - ( p u; u;) . . . . . . . (5)
at I axj ax; ax;axj ax,
in which U; = the~verage velocity; and u;
= the turbulent fluctua tion velocity.
It is known that u; = u; and u; = 0 in the case of a perfect fluid (µ = 0) and .
in the case of a laminar flow.
If this equation applies for the prototype, a similar equation should also apply
for the scale model, and for similitude one should have ·

p ~I p U; au i I a(p g z ) I ~ (p u; u i ) I
at axj ax; axj Ill

-1
m l1l Ill

= . (6)

P aai;; Ir PU; ::: Ir a(::;z) Ir -a( p u( u '. )

ax 1. I
p

Since Pm= Pp, g =

g and zm/ zP= X., one may deduce that the scale for
pressure p is also X.. The scale for acceleration , auj at , is unity as is the scale
of g__:_Consequentl y , the scale for time is t'" / t P = ...,;-y;:, and the scale for velocity
is (LI; 1,,,)/(u; Ir)= ...,;-y;:_ It is seen also that a similitude of viscous force is
impossible unless X. = unity, and that a similitude of Reynolds stresses is possible
only if u; u;
= (u;) 2 which is the case of very turbulent flow on short structures .
It is also seen that similitude is possible , as previously stated , in the case of
a perfect fluid (µ = 0, u; = u ;), where the only forces are inertia (linear or
nonlinear) and gravity-press ure. Consequentl y, a first-sc ale effect is due mostly
to the fact that the viscous forces are not in similitude . The value of scale
model experiments depends upon their relative importance , i.e., the thickness
of the boundary layer and its effect on related phenomenon s such as separation
and wakes .
Fortunately, many fluid flows may be successfully considered as that of a

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 323

perfect fluid (µ = 0) or as very turbulent. The Froude similitude or generalized

Froude similitude is then the rule for all free surface short flow patterns. Similar
considerations applied to flow under pressure which permit us to define what
we call "the generalized Eulerian similitude" which is valid for all short flow
patterns under pressure.
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

It is important to note that short models cannot be distorted, as the flow

pattern on scale model would be completely different from what it is on the
prototype. Generally , scale models used in the study of water waves should
be considered as short models and cannot be distorted . Indeed since the wave
velocity is C = (gT /2-rr) tanh (2-rr d/ L) (in which T = the wave period; L = the
wave length; and d = the water depth), the tanh must be the same on the model
as on the prototype, and this is only possible if the relative depth, d/ L, is
the same. Then both time scale and wave velocity scale are like YT. Since
the ratio of wave lengths, L,., /LP, is given by the horizontal scale , >-., it is
deduced that the ratio of water depth, d,.. / dP , should also be>-. . Then the
entire wave pattern is in similitude . There are some exceptions in the case
of long waves and in the case of movable bed scale models. The latter case
is treated in a following section.
Let us examine the case of long wave agitation. In the case of long wave,
tanh (2-rrd/ L) - (2-rrd / L) and L = T v'gd , C = V gd. The model and flow
pattern could then be distorted and the wave agitation pattern remains identical.
The velocity scale is (dm/dP)' l 2 = µ 112 and the time scale scale is

>-.
- - . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
Vµ

This time scale also applies to distorted tidal model and distorted model of
river.

LoNG MoDEL, S1M1uruoe OF HEAD Loss

By contrast with the formal approach, which consists of presenting a parallel

between Froude and Reynolds similitude, in the following, a parallel is drawn
between what we can call "short model" and "long model." In a short-scale
model, viscous friction is unimportant as compared to gravity and inertia;
therefore, it is governed by Froude similitude, as previously seen. Also, energy
dissipation may result from a fully turbulent condition, as in the case of a
hydraulic jump or a wave breaking on a beach. Boundary layer effects in both
cases are unimportant. The dissipative forces are also proportional to the square
of velocity like the inertial forces. This is the generalized Froude similitude
seen in the previous section.
On the other hand , in a long model, friction has a definite influence on the
flow pattern; therefore, in addition, a similitude of head loss is required. This
head loss is a function of the Reynolds number, but is not determined by the
so-called Reynolds similitude requiring an equality of Reynolds numbers. Simili-
tude for long models requires, in addition to the Froude similitude or Euler

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 AUGUS T 1976 WW3
324
similitude, another conditio n as now shown. The pressur e drop , .1.p in a pipeline
,
is given by

PAi2 : = f ( UvD , ; ) . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . (8)
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

=
in which L the length of the pipe; D the diamete r; U the average =
velocity ; =
and k., / D = the relative roughne ss. This equatio n is equival ent
to

A; = f ( ~D ' :; )
2~~ . . . . . . . . . . . . . . . . . . . . . . . . . . . (9)

in which the head loss, .1.H = Ap/pg.

[In the case of noncirc ular cross sectio~ s,
D could be replace d by a hydraul ic radius, RH(RH = D/4).] Th_e
latter appli~d
to flow under pressur e and free surface flow as well, but evidenc
ed gravity
for free surface flow.
For similitude, one must have

: I"'=: IP and p,1.i21 m =:i2 \P . .......... ... ()Oa)

or A:L=A:\P and ~;\,,,=~;\P .... .... .... . . (10b)

In both cases, one finds

f UD
( - \ ,ks\ ks \ ) or f,,, = f P • • • •
- ) =( fUD
\ 'D - . . . . . ( 11)
V D m rn V P P

This cannot be achieve d in the case of a smooth gallerie s or

channe ls, for
which the friction coeffici ent, f, is indepen dent of the value
of the relative
roughne ss, k,/ D. Indeed followi ng the Euler or Froude similitu
de, the Reynolds
number s are related by

UD \ = UvD \ x X. 3/2 . . . . . . . . . . . . . • . . . . . . . . . . . . (12a) •

V tn P

i.e., UD \ < UD \ . . . . . . . . . . . . . . . . . . . . . . . . . . (12b)

V m V P

Referrin g to the Blasius curve or the Nikura dse or Moody Diagram

s of elemen tary
hydraul ics, it is seen that the friction coeffic ient, f"', can only
be larger than
fp-
In the case of a rocky rough gallery (not lined by concret e walls),
the relative
roughness is large and the bounda ry layer is turbule nt. Then,
it is possible
to adjust by trial and error the scale model roughn ess in order
to obtain the
same friction coeffici ent. A "Froud ian" dischar ge is then obtaine
d, i.e., Q"' =
~PA~ 12 , when the relative roughn esses, (ks/ D) 1:" = (ks/ D) IP,
are theoretically
1dent1cal, and the Reynol ds number , (UD/v ) Im, exceed s a given
value in such
a way that the bounda ry layer be "compl etely rough." Or again,
k 5 u. /v > 70,
in which u. = the shear velocity .

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 325
In any case, the velocity and vorticity distribution which is also a function
of the Reynolds number could never be in similitude exactly. The error is a
scale effect, which is most often acceptable .
One has seen that a water wave scale model is generally considered as a
short model, whether in Froude similitude (wave propagation), or in the general-
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

ized Froude similitude (wave breaking). However , the magnitude of longshore

currents and location of rip currents may , to some extent, depend upon friction
characteristics of the beaches, in which case the study of these phenomenons
would have to be considered as belonging to the categories of long models,
and therefore, may not be studied on scale models . Nevertheless, most scale
model studies have to deal with short coastal structures (like entrance of harbor),
and therefore, the water motion is not too dependent upon the friction coefficients.
The main dissipative mechanism is due to wave breaking. If viscous damping
is too significant, as in the case where the wave h:!!s to travel a long distance
in very shallow water, a correction coefficient, determined theoretically, could
be applied. Very rarely do we have to be concerned with adjustments of roughness
for similitude of energy dissipation, and coastal models can be considered as
short models.
The wave reflection coefficient of a smooth wall is smaller on a scale model
than on the prototype. The opposite tends to apply in the case of wave absorbers
made of ripraps (large rocks). This scale effect is corrected by adding wire
mesh on the scale model wave absorber, or by relatively increasing (distorting)
the size of the model riprap.
As has been pointed out previously, a similitude of head loss is imperative
for models of rivers and estuaries-these are long models. Despite this adjustment,
the vertical velocity distribution being a function of the Reynolds number can
· never be in similitude .
Ship resistance is due to the waves created by the hull displacement, and
to the boundary layer drag and flow separation. The first phenomenon-the
ship wave or Kelvin wave-would exist even in a perfect fluid, and it is a
function of a Froude number only. From this standpoint, the ship hull is a
short model and this effect could be investigated in similitude. But with respect
to boundary layer development, the hull is a long model, and since the hull
is smooth, this force is not reproduced in similitude. One can only obtain an
approximate effect, by introducing a thin wire at the bow of the scale model,
in order to cause some turbulence in the scale model boundary layer. On the
contrary, waves breaking on large armor units covering breakwater core, create
very violent and turbulent flow motion around these blocks. Therefore , it is
to be expected that the corresponding pressure forces will be in similitude.
Studies on stability of breakwater are done according to the law of the
generalized Froude similitude. It is a short model. The study of the permeability
of breakwater requires the reproduction of similitude of permeability and energy
loss by distorting the size of the material of the core of the breakwater (5).
From this standpoint, these are long models, according to our definition: The
stream tubes within the porous core are long. If the similitude of permeability
is not respected, underpressure forces acting on concrete caps are subjected
to scale effects. Therefore, breakwater stability studies must be done at a relatively
large scale for preventing scale effects.
Let us further consider the case of long (fixed bed) scale model of rivers

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 326 AUGUST 1976 WW3
and estuaries. Such models are generally distorted . Otherwise, they would r~quire
very large horizontal space and water depth would be so small that viscous
scale effects, and measurement difficulties will impair the value of the model.
Distortion, of course, induced scale effects, particularly on secondary currents
but it is a satisfactory compromise. The experimenter may chose different vertical
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

scales for the cross section from the vertical scale chosen for the energy slope,
S. In any case, since S = 6.H/ L, the scale for the friction coefficient must
be equal to the scale of the energy slope, i.e., fmlfr = S'"f Sr , which may
lead to the increase of the relative roughness by artificial means.

MODELING OF WAVE FORCES ON STRUCTURES

The important case of a cylindrical structure is now analyzed. One must

distinguish between small piles and large piles, characterized by the value of
the ratio of the diameter, D, to the wave length, L: D / L . In the first case
(small D/ L), the wave force is given by the Morison formula as a function
of the horizontal component of velocity u of the incident wave as

(13)

in which C O = drag coefficient ; and CM= the virtual mass coefficient. The
values of C O and CM are supposed to be parameters that are constant in time
but which may vary as functions of Reynolds number, uD/ v, roughness of
the pile , etc. Acutally C O and CM are time-dependent functions [but these
functions are still not well defined (2)], and the wave velocity field, u, is not
free , but modified by the presence of the pile. The coefficients, C O and CM •
are obtained experimentally . The values of CM and C O must be the same on
the model and on the prototype. For a perfect fluid, CM = 2 and C O = 0. In
practice, a wake develops on the lee side and on the front side of the pile
alternately; thus CM ,ic. 2 and C O > 0. In practice the wake effect does not
change the value of CM very much, but strongly affects the drag coefficient,
C 0 . The dependence of C O upon the Reynolds number makes similitude possible
only if uD/v > 2 x 10 5 in the case of uniform flow, but actually u is time
dependent and varies from positive value to negative value alternately. The
effect of roughness remains small, except that it may change the inception
of flow separation and thus influence the wake. From this standpoint, the
development of the boundary layer around the pile is subjected to scale effect
and should be considered as a long model (even though the body itself is short).
Therefore, wave forces on small piles cannot be studied accurately on scale
models unless the ratio of inertial force to drag force is large. This ratio is
represented by the Iversen modulus:

~1
at max
D
. . . . . (14)
u2 Imax
which is actually an inverse Froude number where the gravity acceleration,
g, is replaced by the particle acceleration, au/ at.

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 \/vW3 COASTAL ENGINEERING 327
In the case of very large piles, large D/ L, the problem is one of wave
reflection and wave diffraction. The drag force is small, and in some cases
(like large circular or ellipsoid cylinders) , the problem can be linearized and
treated completely analytically with good accuracy. The solution is given by
a potential function. It is evident that such wave motions around very large
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

piles (D/ L > 0.5) are very well reproduced on scale models even at a small
scale.
Moreover, complex boundary conditions due to complex structural forms and
nonlinear effects, particularly important in the case where the wave breaks
on the structure, are also in similitude (except for scale effect due to air
entrainment), whereas they cannot be analyzed by theory.

DRAG FORCE PREDOMINATES

(NOT IN SIMIL.ITUOE)

"
~~,_,o

!i
..,
:c
... 1

:!;
0

VIRTUAL MASS FORCES

PREDOMINATE
(IN . SIMILITUDE)

o.,oL-.1---'--'--.,__L...ll._._uu,'---'---,J,o·
.o~,-'--'-'---'--'--'-.Lo'--,-.1----''---'--'-~~~
O.lo!-..,
d/ L • DEPTH / WAVE LENGTH
PERIOD T, HC

FIG. 1.-Forces on Cylinders FIG. 2.-Capillary Effects

All these considerations being borne in mind, it is possible to establish the

critical value of H/ D for which inertial forces

. . . . . . . . . . . . . . . . . (15a)

equal the drag force

(FD)max = p CoD I~d u~.x dz (15b)

as a function of the wave height , H , and the relative depth, D/ L, in which

u, at a first order of approximation, is equal to
H cosh m(d + z)
u= k - - - - - c o s (kt - mx) . . . . . . . . . - - - - • • • • • · (16)
2 sinh md

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 AUGUST 1976 WW3
328
The calculations have been performed for two values of the wave height, i.e.,
H-+0 and H=Hb, such that Hb/Lb=0.1 4 tanh (2ndb/Lb) (limit wave
steepness).
For the sake of simplicity, this work has been carried out by assuming C O = I
and CM'= 2 as the most realistic conservative values. The results of these
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

calculations are presented in Fig. I (2). On the high side of these two curves,
the drag forces dominate and there is no way of obtaining a satisfying scale
model investigation except at scale unity. On the low side, the inertial force
dominates in such a way that similitude is more valid. Another significant criterion
used in engineering practice is the "excursion ratio," i.e., the ratio of the horizontal
wave particle amplitude to pile diameter, which for the sake of similitude, has
to remain smaller than unity.
From these consideration s, it is seen that in the case of small D/ L, the
larger the scale, the better. At the opposite, large values of D / L can be studied
satisfactorily even at a small scale. Fig. I can also be used as a guide for
forms of body other than circular pile. For example, if D is considered as
the beam of a ship, it is seen that the study of ship behavior in water waves
can be done in similitude.

MOVABLE BED MODELS

It is beyond the scope of this paper to present a detailed analysis on movable

bed scale models. This very specialized subject has been treated in an abundance
of literature (8). However, it is pertinent to introduce this subject in order
to illustrate the engineering approach to scale model technology by comparison
with the formal approach. Probably it is in the field of movable bed scale
models that the difference between engineering as an art appears best with
the formal scientific approach to similitude.
The first scale model involving sand and water interaction was reported by
Sextus Julius Frontinus who lived from 40 A.D.-103 A.D. He held the position
of administrator of water supply in Rome under Emperor Trajan. He wrote:

. . . Our attention has not been limited to visiting each of these aqueducts .
We have built a scale model where one can observe rivers and ducts
which lie on hill slopes. Layout of such ducts requires great caution for
avoiding obstruction by sand. The scale model presents a great advantage
to simulate not only the overall layout of the duct but also to help us
on what should be done.

Sextus did not know about dimensional analysis. He was not inhibited by
his model's lack of compatibility of the condition of similitude. He did not
know that his movable bed scale model was wrong, according to the rules
dictated by the God of Mathematics . He was an engineer and he just made
it work. To say the least, he saw in his model a number of phenomenon s
w~ich helped him to u~derstand why some quarters of Rome were not provided
with water. The expenmental method has been, and will remain, the basic tool
of the engineer.
The first scale model with a movable bed was built in 1875 by Fargue. It
was a model of Garonne Estuary with fixed banks and a bottom made of sand.

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 329
Reynolds probably built the first fully movable bed scale model, which was
of the Estuary of Mersey. The first model had a horizontal scale of I /31,800
and a vertical scale of I / 960. Later , Reynolds built a larger one with a horizontal
scale of I/ 10,560 and a vertical scale of I /396. In both models , bottom evolutions
were found to agree closely with natural conditions.
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

It was not until after World War II that the state-of-the-art was advanced
to a point where the model become the most valuable tool for investigating
coastal engineering problems . This occurred first at the Waterway Experiment
Station at Vicksburg, Miss ., using sand tracers, and then in Europe at other
hydraulic laboratories. The methodology was essentially based on trial and error
and experience. Any attempt to rationalize it, taking into account all known
laws of sediment transport, led many to the conclusion that movable bed scale
models just do not work, or if they do work, they must be sacrificing some
effects of lesser importance.
The more we know of the laws of sediment transport (and we still have
a lot to learn), the more the scientists of the profession become convinced
that similitude is impossible. However, engineers do not appear to care that
specialists in sediment transport have declared that the movable bed scale models
are wrong; so, just like the bumble bee who goes on flying even though it
has been declared aerodynamically impossible for him to do so, engineers keep
using the movable bed scale model, and most of the time with great success.
For example, the scale model of the Seine Estuary, which was done more
than a quarter of a century ago in Grenoble , has been able to predict a complex
bottom evolution of the meandering process . We now have 20 yr of field monitoring
which proves the the scale model prediction was exact within engineering
accuracy. One may wonder why these discrepancies between the engineering
approach and the formal approach based on analysis still exist.
We cannot deny the fact that engineers have advanced their art to a point
where this methodology has prover. itself to be fully mastered by experienced
laboratories . First of all, in the case of littoral processes, the laws of sediment
transport are poorly known , but a scale model obeys the same laws of physics ,
rendering results closer to nature than their mathematical counterparts; nature
follows the laws of physics whether or not they have been mathematically
formulated.
Mathematical modeling of complex phenomenons cannot really prove that
scale modeling is erroneous because the mathematical models are an oversimpli-
fication of nature. As such , they are probably more misleading than their physical
analogs . Let us refer to the surf zone example.
Sediment transport in the surf zone is so complex that there is very little
hope of developing a realistic mathematical model. The scale model of the
surf zone tends to be a more accurate reproduction of nature than its mathematical
counterpart. The analysis of the law of similitude, based on mathematical laws,
will lead us to an incompatibility between the uprush and the offshore . But
these differences are actually reproduced relatively satisfactorily in the scale
model, despite our lack of knowledge about how nature in the lab duplicates
nature at sea . Both are governed by almost the same law of physics . Is there
any reason to deny the validity of scale model reproductions of natural phenome-
nons in the name of analysis? Shouldn't we believe facts rather than mathematical
construction?

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 AUGUST 1976 WW3
330
Engineers believe facts and they adapt the law of similitude to their convenience.
They have found a way to develop this tool. Engineering is an art. By ignoring
conditions of similitude dictated by a formal approach or by stretching them
to the limit, they have invented what the writer has called in the past "criteria
of similtude."
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

The conditions of similitude are imposed by the law of physics. The selection
of criteria of similitude is the free choice of the experimenter to make his
model work . Let us go back again to history . Harcourt , who was a disciple
of Reynolds and built a number of movable bed scale models using artificial
sediments such as charcoal and pumice wrote:

If I suceed in demonstraing with the model that the originally existing

conditions can be reproduced typically ; and if, moreover, by placing
regulating works in the model, the same changes can be reproduced that
were brought about by the training works actually built, then I am sure
that I can take the third and most important step , namely , of investigating,
with every promise of success, the probable effect of the projects that
have been proposed . . . .

This principle has been the basic guideline ever since . The model is said to
be verified if it reproduces all the features of bottom evolution observed in
the past. Such a model can then be used with confidence to predict the events
in the future . This is a simple and reliable rule, although no rigorous justification
has ever been made .
Nature is very complex . For example, waves change from day to day and
from one to another in amplitude, height, and direction . The engineer who
has at his disposal only a wave paddle will only be able to experiment with
a limited set of conditions. What is the point of being in similitude at a given
location under a given wave condition when the conditions imposed are incompat-
ible with those for another wave on another day? It is obvious that the condition
of similitude can only be a guideline, since storms, swells, and tidal inlet effects
can only be reproduced by an adjustment of these mechanical tools to reproduce
an observed bottom evolution. Then, if the phenomenology does not drasticall y
change when one investigates engineering solutions, the engineer can be confident
that this model is a reliable tool to predict the future.
It is also important to note that in the case of movable bed scale models
that are used to study sediment transport and bottom evolution of rivers, estuaries,
and beaches, distortion is not an engineering trick for reducing the size of
the model and the bottom friction, but is the extrapolation of a natural observed
phenomenon (8). The method to obtain a satisfactory scale model is first to
obey the law of nature, even though this law may not be fully understood .
For example , a small river flowing in its own alluvium can be considered a
distorted model of a large river. This means that the ratio depth to width of
the small river is greater than the relative depth of the larger one . The ratios
of depths, µ, and widths, >-., are approximately •related by the law of Lacey
(>-- 2 = µ 3 ), in accordance with the "regime theory." Similarly, a beach in a
protected area has a relatively steep slope, while a beach in an exposed area
tends to have a more gentle slope . The vertical scale being defined by the
ratio of incident wave heights, a protected beach can be considered as a distorted

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 331

scale model of an exposed beach. In both the case of the river and the beach,
the choice of distortion becomes a stringent condition to be respected quantita-
tively; however , the natural law determined by statistical observation of natural
phenomenons needs to be modified in the case of a river model due to the
fact that scale models generally do not use the same material as the prototypes
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

in order to satisfy other conditions of similitude . In the case of beaches, a

natural distortion based on the use of sand both in the scale model and in
the prototype is unavoidable. However, the use of sand is not always recommend-
ed because it lends to too large a distortion and, subsequently, to large-scale
effects. Lighter materials (pumice, coals, plastic) are preferred (6).
Similitude of wave refraction is also obtained on a distorted scale model
when the ratio of wave lengths is related by the ratio of vertical depths ,
L"' / LP = d ml d P = µ, and which is obtained with T,,. / T,, = \I'"µ. Such similitude
is used in a movable bed scale model of beaches. It is pertinent to point out
that since the only requirement of a movable bed model is a reproduction of
bottom evolution, it is not necessary that this be achieved through exact similitude
of water motion . As a matter of fact , as already pointed out , it is a case where
the importance of properly adjusted criteria of similitude far exceeds the
importance of condition of similitude . Since the model is distorted, a similitude
of wave refraction and wave breaking only is aimed at. These conditions are
the most susceptible of producing satisfactory Iongshore current and sediment
transport distribution. This is achieved by keeping the ratio of wave lengths
and wave heights equal to the vertical scale, µ . Based upon this condition ,
the following wave characteristics are preserved in the model : (1) Wave steepness
(since H / L remains identical); (2) refraction pattern and angle of refraction
with bottom contours (since d / Land tanh 2-rrd/ L remains identical); (3) breaking
angle of wave crests with shorelines, if the distortion is not too large; and
(4) breaking depths. Also, the scale for longshore current and mass transport
velocities is approximately µ 112 • Therefore, the ratio of scales of wave particle
velocity to current velocity is approximately unity .
The wave diffraction pattern on a distorted model is not in similitude. Similitude
of turbulent fluctuations and sediment transport by suspension does not exist.
Generally , the sediment tends to be transported mostly by bed load on the
scale model. A similar bottom evolution could be attempted after a lengthy
calibration of the scale model based on prototyped observations as previously
explained .
All these concepts were fully understood in the past. However , they must
be reemphasized now, since doubts may be generated by the new findings in
the laws of sediment transport. The engineer thinks pragmatically about this
issue. The problem is not so much whether the model is valid or erroneous,
but whether the model leads to more successful results than could be achieved
if the engineer did not have this tool at his disposal. The latter premise has
never been substantiated in the literature.

SCALE EFFECTS

The influence of capillary effects has to be minimized on scale models of

water waves for two reasons: (1) The wave celerity is a function of surface
tension ; and (2) surface tension introduces a wave damping effect. Let us consider

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 AUGUST 1976 WW3
332
the first problem and try to determine the suitable minimum wave pe~iod . for
scale experiments. It is known that the wave velocity (first order of approximatio n)
is given by
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

gL' A 2'TT] 2'1Td

C 2 = [ - - + - - tanh-- . . . . . . . . . . . . . . . . . .. .. . (17)
2'TT p L' L'
in which L' = the wave length ( C = L' T); T = the wave period; g = the gravity
acceleration; d = the water depth; p = the density ; and A = the surface tension
(74 dynes/cm for air-water interface at 20° C). If Lis the wave length assuming
A= 0, L' = L[I + (AL/ L)] . After some straightforwa rd calculation , one
obtains the scale effect defined by AL / L. Fig. 2 shows these results, in which
AL / L is given as a function of T and d. It is seen that AL / L is smaller
than 1% when T > 0.35 sec and d > 2 cm. In practice, it can be stated that
the lower limit for limiting scale effects is T > 0.35 sec .
Let us now consider the wave da mping phenomenon by viscous effects. The
wave damping phenomenon is strongly affected by scale; viscous damping is
mostly due to boundary dissipation. The wave height, H, is a function of the
horizontal distance (12), x, as H = H O e - "' ', in which A = the damping coefficient.
In a rectangular tank of width b a nd depth d, A = A b + A"'. The value of /1 b
is due to the solid wall a nd is

2
11 =2m[~J
11 2
mb+ sinh md . . .. . . . . . . . . . .. . . . . . . (IS)
b b 2k 2md + sinh 2 md

in which m = the wave number and equals 2'TT / L; k = the frequency and equals
2'TT /T; and v = the kinematic viscosity . The value of A"' is due to a contaminated
film on the free surface which , at the maximum, can be

. . . . . . . . . . . .. . . . . . . .. 09 )
11 2 2
A, = 2m ( ~ ) mb cosh md
. b 2k 2md + sinh 2 md
From these formulas, it can be verified that the viscous damping is not negligible
in many wave tanks where the water depth is smaller than 2 cm , and this
water depth should be considered as a reasonable lower limit for scale model
studies .
One important scale effect is due to the fact that in the wave tank one usually
~ses fre shwater instead of seawater. The slight difference in density, which
1s approx 3% , changes the wave forces accordingly. Let us consider , for example ,
the case of a rockfill breakwater. The minimum weight, W , of the rocks or
armor units covering the breakwater is approximatel y given by the Hudson
formula:

· · · · · · · · . . . . . . . . . . . . . . . . . (20)

As . presently ~sect by engineers, K,.. is a stability coefficient =3 for rocks ;

P_h •.s. the density of the armor units; and p , is the salt water. Therefore, for
s1mil1tude:

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 333

. . . . . . . . . . . . . . . . . . . . . . . . (21)
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

in which p = the freshwater density, and since W111 / WP= H;,,/ Ht= X.3, one
must have

(p p
bp -

Ps
s )
3
(p
Pbm

bm -

P
r
- - - - - . . . . . . . . . . . . . . . . . . . . . . . . (22)
p
If p 111 = p P, it is seen that the error on W due to the use of freshwater could
be 10%-15%, if no correction for this were made .
The solid friction forces are generally relatively larger on a scale model than
a prototype, as in the case of a concrete cap on rockfill breakwater or of
a ship against a fender. The structural strength of material is generally not
reproduced in similitude, for example, if the armor units covering breakwaters
were in complete structural similitude, one should be able to crush them in
the hand. One-fiftieth scale models Dolosses or Tetrapods falling from a height
of, e .g., I m do not break while their prototype counterparts falling from 50
m certainly do .
The Cauchy similitude or similitude of elasticity, could be made compatible
with the Froude similitude (7) . This similitude is required to study moored ship
mooring, for example , since the elastic forces due to mooring lines have to
be in similitude with other forces . Also, the solid friction of the ship against
the fenders has to be made compatible. (This is done by making the fenders
very slippery , smooth, and oily).
The Cauchy similitude is obtained either from the Boussinesq equation for
elastic material, or from the equation of the phenomenons to be studied, through
inspectional analysis. Generally, it is sufficient that the Poisson ratio be the
same on the model as in the prototype, and the modules of elasticity, E 111 ,
be such that E"' = Ev XX. .
Shock waves due to underwater explosions are not in similitude as the
compressibility of water is not scalable. The same reasoning applies to slamming
or impact phenomenons . The free surface is "harder" on scale models.
Air compression effects, such as due to the slamming of a wave against
a vertical wall or buoyant tanks subjected to pressure variations could not be
investigated in similitude, unless the scale model is operated in a partial vacuum.
Air entrainment such as found in hydraulic jumps or in a breaking wave does
not present the same concentration distribution at a small scale as in the prototype.
This effect is significant in the study of wave forces [Skaldnev and Popov
(10), found that this scale effect is significant for wave heights smaller than
50 cm.]
The choice of scales is the historical result of experiences that may have

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 AUGUST 1976 WW3
334
been forgotten. Numerous laboratories have, in the past, carried out systematic
tests at different scales to find out that : (I) The use of a larger scale does
not pay; and (2) the use of a smaller scale will be subjected to too much
scale effects. For example, harbor scale models for wave agitation studies have
been done at scales I / 300, I / 150, and I / 100, to find out that 1/ 150 is typically
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

the most economic and still the most technically satisfying scale .
Typical scales used in engineering practice are: (I) Breakwater stability / rock.fill
cofferdam, 1/ 3001 / 50; (2) wind-wave penetration in harbor , 1/ 100-1 / 150 (the
water depth cannot be smaller than 2 cm and the wave period has to be > 0.5
sec); (3) spillways, bottom outlet, water power structures-typically, I / 50-1 / JOO ;
(4) river , estuary-distorted-typically, 1 / 100 vertical and I / 800 horizontal; (5)
beaches, shoreline processes-distorted-typically, I / 100 vertical and 1 / 300
horizontal ; and (6) ship dynamics problems-typically, 1/ 100.

CONCLUSIONS

The state-of-the-art in scale model experiment has now been established for
a long time . This paper is an attempt to present the point of view of the practitioner ,
which somehow differs from the formal approach based on dimensional analysis .
Engineering ingenuity has extended scale model technology beyond and often
aside from the formal rules dictated by dimensional analysis, leading to widely
accepted world-wide practices that differ little from one place to another.
Experience in handling the scale models , particularly the movable bed scale
models, permits us to think about solutions that may not come to mind without
their use. This exercise develops sophistication in our approach to problem
solving and thus contributes to the advancement of the state-of-the-art.
Scale model technology, of course , has its drawbacks . It is often prohibitively
time consuming and expensive when compared to mathematical modeling and
therefore, many projects have to be designed without recourse to this tool.
However, the writer doubts that many engineers would choose not to employ
scale modeling where time and budget would allow.
The intuitive knowledge gained by personally experiencing the elements is
what permits us to discover " shortcuts" which often seem like serendipity!

ACKNOWLEDGMENT

This paper is present under the auspices of the ASCE Task Committee on
Coastal Engineering Methods of Investigation : Choule Sonu, Hugo Fischer, and
Bernard Le Mehaute, Chairmen.

APPENDIX. -REFERENCES

I. Birkoff, G., Hydrodynamics: A Study in Logic, Fact and Similitude , Dover Publications,
New York, N.Y., 1950.
2. Goda, Y., " W,ave Forces on a Vertical Circular Cylinder," Report No . 8, Law and
Harbor Technical Research Institute , Japan, 1969.
3. Langhaar, H . L., Dimensional Analysis and Theory of Models, John Wiley & Sons,
Inc ., New York, N.Y., 1957.
4. Ledov, L. I. Similarity and Dimensional Methods in Mechanics , Academic Press,
New York, N.Y ., 1959.

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335
 WW3 COASTAL ENGINEERING 335
5. Le Mehaute, B. , "Permeability of Rockfill Breakwaters to Periodic Gravity Waves, "
La Houille Blanche, No. 6 (Translated into English by Associated Technical Services,
Inc.), 1957.
6. Le Mehaute, B., "Theory, Experiments, and a Philosophy of Hydraulics ," Journal
of the Hydraulics Division, ASCE, Vol. 88, No. HYl, Proc. Paper 3040, Jan., 1962,
Downloaded from ascelibrary.org by Texas A&M University on 12/17/24. Copyright ASCE. For personal use only; all rights reserved.

pp. 45 .
7. Le Mehaute , B., "On the Froude-Cauchy Similitude," Santa Barbara Coastal Engineer-
ing Specialty Conference, Chap . 14, 1965, pp . 327-343.
8. Le Mehaute, B., "A Comparison of Flu vial and Coastal Similitude," Proceedings
of the Conference on Coastal Engineering, Washington, D.C ., Sept., 1970.
9. Le Mehaute, B. , An Introduction to Hydrodynamics and Water Waves, Spinger-Verlag,
Berlin, Germany, Apr., 1976.
JO. Skaldnev, M. F., and Popov, I. Y., "Studies of Wave Loads on Concrete Slope
Protections of Earth Dams," Paper No. 7, Symposium on Wave Action, Delft, the
Netherlands , March 24-28, 1969.
JI. Stoker, J. J. , Water Waves, Intersciences Publishers, Inc ., New York, N.Y., 1957.
J2. Van Dorn, W. G., "Boundary Dissipation of Oscillatory Waves," Journal of Fluid
Mechanics, No. 24, 1966, pp . 769-779.
13. Yalin, M. S. , Theory of Hydraulic Models, The Macmillan Co., New York, N.Y.,
1971.

Journal of the Waterways, Harbors and Coastal Engineering Division, 1976, 102(3): 317-335