You are on page 1of 10

OFFSHORE TECHNOLOGY CONFERENCE

6200 North Central Expressway


Dallas, Texas 75206
THIS IS A PREPRJ1'iIT --- SUBJECT TO CORRECTION
1614
Cno i da I Wave Theory for App I i cat ion
To Offshore Structural Design
By
Gerald O. Mallery and George C. Clark, Continental Oil Co.
Copyright 1972
Offshore Technology Conference on behalf of the American Institute of Mining, Metallurgical, and
Petroleum Engineers, Inc., American Association of Petroleum Geologists, American Institute
Chemical Engineers, American Society of Civil Engineers, American Society of Mechanical Engineers,
Institute of Electrical and Electronics Engineers, Inc., Marine Technology Society, Society of
Exploration Geophysicists, and Society of Naval Architects &Marine Engineers.
This paper waS prepared for presentation at the Fourth Am:m8.1 Offshore Technology Conference
held in Houston, Tex., May 1-3, 1972. Permission to copy is restricted to an abstract of not
more than 300 words. Illustrations may not be copied. Such use of an abstract should contain
conspicuous acknowledgment of where and by whom the paper is presented.
ABSTRACT
Cnoidal wave theory has never been rou-
tinely employed for engineering design al-
though several versions of the theory have
been advocated in the literature. The main
reason that cnoidal theory has not been used
is that the computations have been somewhat
complicated. The theory presented in this
paper was programmed specifically for engi-
neering calculations. It can easily be com-
bined with any general purpose structural
mechanics program for simplified application
to offshore structure design. This version
of cnoidal theory was obtained by differen-
tiating Keulegan-Patterson's velocity poten-
tial. It predicts velocities close to labo-
ratory measurements for large amplitude,
shallow water waves. The predicted vertical
velocity and both components of acceleration
differ from previously published results.
INTRODUCTION
The designers of offshore structures
must select a theory to predict water par-
ticle velocities and accelerations (kine-
matics) in order to calculate wave forces.
The choice of a suitable wave theory is
References and illustrations at end of paper.
determined by water depth and design wave
height and period. Unfortunately, there
have been so few measurements of water par-
ticle kinematics, and those that have been
made are over such a small range of wave
parameters that it is often difficult to
know which theory to use in a specific de-
sign situation(l).
In intermediate and deep water, the
oil industry has employed, generally quite
successfully, Stokes' fifth order, stream
function, and linear theories. The research
reported herein was initiated in response to
design engineers' needs for accurate wave
theory to predict forces in large amplitude,
shallow water waves. In a study of such
waves, Le Mehaute, et al(2) state, "While no
theory was found exceptionally accurate, the
cnoidal wave theory of Keulegan and Patter-
son appears most adequate for the range of
wave lengths and depths studied." Dean has
shown that stream function theory fits the
boundary conditions(3) and data(4) quite
well. However, errors had been discovered
in the published equations of Keulegan-
Patterson cnoidal theory (5). Consequently,
it appeared that improvements might be pos-
sible in the fit of cnoidal predictions to
shallow water wave kinematics measurements.


(7)
(6)
(5)
(4)
O'l'r. H;1l..
(3)
(h
z
+ s) K(k)
d + H - 3h
z
- s .
(3)
(Contd) J

,
2d dX3
L I l6d
3(H + s) K(k)
k Z = _H_
H + s
(H + s) E(k)
d - H + :' [ 2 - 3 ]
Elimination of hz, s, and L from the above
equations yield the equation
..:::1.::..6....:d=.-3--:;:k_
z

- 0 (8)
3 H TZ g
Equations (7) and (8) represent the
velocity of propagation of a wave height
element with respect to a fixed coordinate
system and correspond to wave staff obser-
vations. An alternate definition is avail-
able corresponding to the velocity of the
wave profile with respect to the center of
inertia of the fluid in the wave (Stokes'
second definition) (10). This velocity is
given by
Equation (8) can be solved using nonlinear
solution techniques for the elliptic modulus,
k. This equation is equivalent to one pre-
sented by Wiegel(9, Eq. 2.163).
respectively
It is necessary to use additional re-
lationships to define the unknown variables
in these equations (hz,k,L) in terms of the
known variables (H,T,d). These defining
equations are
(1)
CNOIDAL WAVE THEORY FOR APPLICATION TO S
+ H
and
u h h
Z
-- ... ----+
Igd d .4d
z
The horizontal and vertical velocities are
given by
v
--=
The surface profile of a wave is de-
scribed by
The equations which were used are pre-
sented in Keulegan and Patterson (8) , and the
readers are directed to that manuscript for
the theoretical development. The nomencla-
ture which is used is basically that used by
Keulegan and Patterson with some minor ex-
ceptions and is described in the table of
nomenclature.
THEORY
In addition to making these improvements
in the theory, we developed this technique
specifically for the design engineer. Lee(6)
has stated, liThe complex equations involved
in cnoidal wave theory have prevented its
wide application for practical purposes."
Iwagaki (7) adds, lilt is very difficult for
engineers to deal with the cnoidal wave
theory for practical application . "
Similar statements appear in nearly every
paper discussing some aspect of cnoidal
theory. The reason for the difficulty in
applying the available cnoidal theories is
that they are generally presented as non-
dimensional graphs and tables of functions.
Consequently, computation of water particle
kinematics at the many positions needed in
the design of a structure is tedious, time
consuming, and subject to human error. Thus,
we wanted to develop a computer solution in-
puting only wave height and period and water
depth to be compatible with the commonly
used structural mechanics programs.
OTC 1614 CNOIDAL WAVE THEORY FOR APPLICATION TO OFFSHORE STRUCTURAL DESIGN I-943
Using this equation in place of Equation (7)
yields
_ 16 d
5
k
2
K2(k) _ 0 ...... (10)
3 H T2 g
where
y = d - (11)
These equations can be solved for the ellip-
tic modulus, k, by techniques similar to
those used for Equation (8).
av
[l }
a
2
h 1 ah ah
-'"
aXdt - Wat ax at
+1.
[ Z']'"h J
3
d- 2d ""t_
av
-zlid
[[ }
a
2
h 1
[:: r
-zs
ax
2
- 2d
2
ax
+1.
[ - ] '4
h
J
3
d 2d 8X4
Using Equation (2) the appropriate terms for
the horizontal acceleration are
a
= Du = au + u au + v au (12)
x Dt at ax az
2 a
3
h
z W (15)
[ 2K(k) [ J: - t ] . k ]
ah 4H () () dn( )
ax .. - L sn cn
To evaluate the velocities and accel-
erations, the first four partial derivatives
of the equation for the wave surface are re-
quired. The derivatives of the Jacobian
Elliptic function can be found in several
standard texts (11,12). For brevity sn( ),
cn( ), and dn( ) represent these functions
wi th an argtmlent 0 f
The local accelerations are given by the
first of Equations (14) and (15).
in the following equations
(13)
Dv av av av
a = - ... -+u-+V-
z Dt at ax az
The particle accelerations are given
by the total derivatives of velocity with
respect to time. The horizontal accelera-
tion is
The vertical acceleration is
= Igd [_ j ::,h] (14)
K2 (k) [
-8H cn
2
() dn
2
( )
- sn'( ) dn'( ) - k'cn'( ) sn'(
a3h
= -64H L cn() dn( ) sn( )
Using Equation (3) the terms for the
vertical acceleration are
[k' [ sn'( ) - cn'( ) ] - dn'( ):J
.944
CNOIDAL WAVE THEORY FOR APPLICAq
(16)
(Contd)
r
a4h
p=
{
-128H~ 6 k2 cnz( ) sn2( ) dn2( )
[[
+ k2 sn2(
L
)
- cn2( )
1
- 7
1
- dn2( ) cn2( ) dn2( )

sn2( ) dn2( )

1]
-k2sn20cn20 . . . . .. (16)
The mixed derivatives can be evaluated by
the following relationship
an h =_Lanh
.
an-1
T
x at axn.
Equations (16) and
stituted into Equations
. . . . . . . . (17)
(17) can be sub-
(14) and (15) to
develop a single equation for each accel-
eration and Equations (2) and (3) for an
equation for each velocity.
Performing
this substitution, we check the equation
given by Wiegel(9) for horizontal velocity
but are unable to obtain the final equa-
tions for vertical velocity and local
acceleration which he presents. For our
solution, we evaluated the derivatives of
the wave profile using Equations (16) and
(17).
In the program, Equation (8) or (10)
was solved for a value of the elliptic
modulus, k, using the Newton-Raphson
Iterative Technique. In reality, the com-
plementary elliptic modulus, k, was used
because of the wide range of values which
can be obtained for a range of wave prop-
erties (1 > k > 10-40). TWO roots for
Equations (8) and (10) exist, and the root
at the smaller value of k is the desired
value. The physical significance of the
wave obtained from the other root is not
physically apparent. The waves computed
with Equations (8) and (10) had very
similar properties. Below a particular
value of period for a wave of otherwise
constant properties, the real roots cease
to exist. This value is primarily a
)N TO OFFSHORE STRUCTURAL DESICN
OTC 161
function of depth and a weak function of
relative wave height (H/d). The minimum
period can be approximated by
T=7K. .0 . . . . . . . . .. (18)
g
This minimum period increases very slightly
as the relative wave height increases. Be-
low this minimum period, this cnoidal solu-
tion does not exist. The significance of
the minimum period will be discussed in the
next section. The subroutine for the com-
plete elliptic integrals and the Jacobian
elliptic functions were obtained from the
IBM Scientific Subroutine Package (13) and
were modified for accuracy at small values
of the complementary modulus.
Once the elliptic modulus is obtained,
the wave length can be determined from
Equations (4) and (6), and the surface pro-
file can be computed from Equation (l). The
velocities and accelerations of water parti-
cles within the wave can be computed from
the appropriate equations given above.
RESULTS
We have compared this version of
cnoidal theory with other theories and with
data from several publications. Space limi-
tations prevent us from including all of
these comparisons. Consequently, we will
concentrate on a discussion of the range of
applicability of the theory and will compare
our predictions with the data presented by
Le M&haut6 et al(2).
There have been several investigationa
of the region of applicability of cnoidal
wave theories. For instance, Laitone(14)
concluded that it should not be used if the
wave length is less than five times the
depth and is superior to Stokes theory only
if the wave length is greater than eight
times the depth. Masch(15) adds that
cnoidal wave theory describes waves whose
lengths are between ten and fifty times the
water depth. Bretschneider et al(1) state
that cnoidal wave theory is applicable main-
ly if the wave length is greater than ten
times the depth. These references all indi-
cate that cnoidal theory can be extended
beyond shallow water waves, which generally
are defined to have lengths greater than 200
times the depth(16). We conclude that until
comparison with data proves the contrary, a
wave theory may be applied over any range of
parameters where it will converge to a solu-
tion.

-.

OTC 1614 CNOIDAL WAVE THEORY FOR APPLICATION TO OFFSHORE STRUCTURAL DESIGN
1-945
As given Zn Eq. 8, we have found that
this theory will converge for periods greater
than approximately
74-.
This corresponds to deep water wave lengths
greater than about eight times the depth.
Thus, if a structure were to be designed for
100 feet of water, the design wave period
would have to be greater than about 12.3
seconds. In 50 feet of water, the minimum
period would be about 8.7 seconds, and in
25 feet of water, about 6.2 seconds.
The fact that the equations will con-
verge to a solution does not necessarily
imply, however, that the results represent
a realistic description of a freely propa-
gating gravity wave. Since the wave height
has little effect on the convergence range,
for instance, this theory will compute kine-
matics for completely unrealistic waves
based on a wave height to water depth crL-
terion. Unfortunately, breaking wave
criteria have not been firmly established.
The commonly quoted height to depth ratio
of 0.78 is misleading. The ratio is
actually a function of bottom slope, but
the wave theories are generally developed
assuming a horizontal bottom. It seems
reasonable then, to look for other sta-
bility criteria to assess when a wave be-
gins to break and limit the region of
applicability based on the stability cri-
teria. Two possible stability criteria
that have been proposed(l7) are the kine-
matic stability parameter (KSP) and the
dynamic stability parameter (DSP).
The KSP is the ratio of the maximum
horizontal particle velocity, Urn,under the
crest relative to the phase speed, u.
KSP=: . . . . . . . . . . . ..(19)
It is argued that if KSP is greater than
1.0, the water particles are outrunning the
wave form, and consequently, the wave be-
gins to break.
The DSP is the ratio of the maximum
vertical acceleration, Dv/Dt, at the sur-
face to the acceleration of gravity, g,
DSP =
-~= . . . . . . . ... .(20)
g Dt
If the DSP is greater than 1.0, the water
particles will leave the surface of the
wave and breaking will occur. Figure 1,
analogous to that given by Dean(17)~ shows
the breaking criteria for this cnoldal
theory. As was found before, the KSP is the
governing criterion. however, by this cri-
terion, the cnoidal wave breaks when the wave
height to depth ratio is only 0.68, whereas
the commonly quoted values range from 0.73
to 0.87, and Dean found 1.0 for stream func-
tion theory. It is interesting to note that
on a horizontal bottom, Divoky et al(18)
were unable to generate waves higher than
about .6 of the water depth without break-
ing occurring. In investigating wave defor-
mation, Street and Camfield(19) reported
waves only up to .59 to .65 of the water
depth on a horizontal bottom. These measure-
ments are in good agreement with cnoidal
theory. However, measured maximum crest
velocities in breaking waves have generally
been significantly lower than the phase
speed(18), which leads one to question the
validity of the KSP as a breaking wave
criterion. We do not, however, recommend
applying this cnoi.daltheory where crest
velocities are predicted higher than the
phase speed.
The best test of a theorys applica-
bility, of course, is to compare it with
measured data. The problem is to get good
wave measurements with which to compare the
theory. Shallow water wave kinematics
measurements have been reported by Le
M6haut4 et al(2), Iwagaki(7), Iwagaki and
Sakai(20), and Wiegel(9). In this paper,
we have chosen to compare our reeults with
Le M6haut6s data becauee many other
theories have been compared with it and
because the results have been verified by
independent experimentation(5,21). In
addition, we have compared our results with
other shallow water wave data, and the con-
clusions reported herein apply equally well
to the remainder of the data. Figures 2
through 9 show the results of our compari-
son with Le M6haut6s horizontal velocities
beneath the wave crest. For a discussion
of other theories fit to the data, the
reader is referred to the original paper.
We will confhe our discussions to the
Keulegan-Patterson (K-P) cnoidal theories.
We found I.tdifficult to compare this theory
to Le M6haut6s version due to the effect of
mass transport currents on the data and on
the two versions of the theory. During the
experiments, the wave tank was completely
closed allowing no mass transport for the
steady state condition. The K-P equations
presented herein, however, do predict a
significant mass transport in the direction
of wave propagation. The mass transport can
1-946
CNOIDAL WAVE THEORY FOR APPLICATION TO OFFSHORE STRUCTURAL DESIGN OTC 161
-. .-.-- _.- .
be reduced to zero and the governtng equa? Since we did not have acceleration
tions still he satfsfied by subtracting a data available$ we compared our predictions
current, uniform with depth, from the with those previously published(23). These
velocity profile predictions. Le M&haut~, pu~l?shed tables of functions yielded the
et al, rather than solve the equations by an same horizontal water particle velocities
iterative technique as done here, used mea- that we obtained. However, our vertical
sured values of the trough depth in their velocity and two local acceleration con-
version of the theory. One cannot, from ponents differed from theirs, as was to be
their paper, compute how much, Tf any, of the expected.
mass transport has been accounted for in
their predictions. Consequently, we have We have not included data on surface
reported our results both with and without profile, since our results are essentially
the mass transport included for comparison the same as obtained previously(2,22,23).
with Le M6haut6s results. In the figures, Cnoidal wave surface profiles have generally
the curves labeled K-P* are our results matched shallow water wave data very well.
with mass transport included. Those labeled
K-P** have no mass transport. Le M6haut6s In order to use this theory for design,
predictions are labeled K-P. As can be the engineer must calculate forces from the
seen from the figures, with no mass trans- predicted water motion. This is normally
port, we tend to predict velocities too low done using the well-known Morfson Equation
at the bottom and too high at the crest. which includes empirically-derived drag and
With mass transport included, our profiles mass coefficients. The drag and mass co-
shift to the right and are very close to the efficient will vary depending on the wave
data at the bottom but even higher at the theory or more precisely on the predicted
surface. With no mass transport, this velocities and accelerations. Unfortunately,
theory has a relative average deviation from we could find no large amplitude, shallow
the data that is essentially the same as water wave force data. For smaller waves,
Le M6haut6s version for the eight cases. where force data was available, this
If we include mass transport, our results theory so nearly matched Stokes fifth order
are about 18 percent worse than his. The theory that the predicted coefficients did
standard deviations of this theory from the not differ appreciably. However, if these
data are .290 with no mass transport and coefficients were applied to this theory for
.340 if mass transport is included. waves nearer breaking, the predicted forces
near the surface under the crest would be
The majority of the variance occurs too high. Obviously, in order to do a more
just beneath the wave crest where the theory accurate job of predicting forces for large
consistently predicts horizontal velocities amplitude waves in shallow water, we need
too high. As the relative wave height (H/d) wave and force data in that range.
Labora-
increases, the discrepancies become larger, tory and field studies will have to be
We may be able to significantly improve on undertaken to obtain this data.
these results; however, with slight changes
in Keulegan-Pattersons solution to the CONCLUSIONS
boundary value problem. They assume some
terms in the differential equations can be The main conclusions reached in this
approximated by the linear shallow water study may be summarized as follows:
approximation. They also discard all terms
above the order (H/d)2. Hendrickson 1. This theory was programmed speci-
has formulated an approach that appears
fically for oil industry applica-
more rigorous and, for one case at least, tion to the design of offshore
predicted lower crest velocities than this structures and does not rely on
theory. graphs, tables of functions, or
measured data as have previous
It was pointed out in the previous cnoidal theories. Use of IBM
section that vertical velocities and the Scientific Subroutines for the
two components of local acceleration re- elliptic Integrals and Jacobian
ported herein differ from those previously elliptic functions have made
presented in the literature. Figure 10 previously tedious calculations
shows a comparison of measured and pre-
routine and accurate. However,
dieted vertical velocity at the point of
there is still a need for shallow
maximum vertical velocity in the wave.
water wave force data to use in
Our predictions (labeled K-P*) fall very
computing drag and mass coeffi-
close to the data and have a relative
cients.
average deviation of 35 percent from Le
M&haut&ts predictions (labeled K-P).
OTC 1614 (XiO~DALWAVE THEORY FOR Aj?PLIcATzoNTO OFFSHORE STRUCTURAL DESIGN
I-947
2. For the eight waves presented here, x = distance from crest
this theory predicts horizontal z = distance from bottom
velocities close to measurements lo = velocity of wave crest with respect
under the crest. The fit is excel- to fixed coordinate
lent near the bottom. Near the
UC
= velocity of wave with respect to the
surface, for the large amplitude, center of inertia (Stokes second
shallow water waves presented, the definition)
horizontal velocity predictions
tend to be too high. ACKNOWLEDGEMENTS
3. The predicted maximum vertical We would like to thank Continental Oil
velocity matched measurement Company for permission to present this paper.
extremely well and had a relative Also, we extend our appreciation to Bob Dean
average deviation of 35 percent (University of Florida) and Dave Dfvoky
from prevf.ouslypublished results. (Tetra Tech), who made several helpful sug-
gestions, and Jim Hendrickson (Science
4. The predicted accelerations, though Engineering Associates), who gave us some
not compared with data, also dif.
results of a cnoidal wave theory on which
fered from previously published he had worked.
values.
REFERENCES
5. This theory predicts that on a
horizontal bottom a wave will 1. Bretschneider, C. L., G. S. Pick and
break if its height is 0.68 of J. 1. Collins, Gravity Wave and Wave
the water depth.

Force Theory, MeasuremeW and Data


Analysis:
.
State of the Art, Nesco

NOMENCLATURE Report SN-9~~~h, 1965).


All units must be consistent. The 2. Le M6haut6, B., D. Divoky, and A. Lin,
elliptic modulus, elliptic functions, and Shallow Water Waves: A Comparison of
integrals are dimensionless. Theories and Experiments, Proceedings,
Eleventh Conference Coastal ~.
a = horizontal acceleration
x
-Ch. 7, 1968.
a = vertical acceleration
CI?() = Jacobian elliptic functions, 3. Dean, R. G., Relative Validities of
cosine amplitude Water Wave Theories, Proceedings of
d = still water depth the ASCE Conference on Civil Engin~r-
. .
dn( ) = Jacobian elliptic functions, ~in the Oceans, San Francisco, Calif.

delta amplitude (Sept., 1967), pp. 1-30.


E(k) = complete elliptic integral of
second kind 4. Dean, R. G. and B. Le M6haut6,
= acceleration of gravity
Expertiental Validity of Watsr Wave
: = height of free surface above (below) Theories, unpublished manuscript.
still water level
hl = distance from still water level to 5. Dean, R. G,, Personal Communications,
wave crest 1969-1972.
h2 = distance from still water level to
wave trough 6. Lee, C. Y., Velocity and Acceleration
H = wave height (hl + h2) Fields in Cnof.dalWaves, M. S. Thesis,
k = elliptic modulus Dept. of Civil Engineering, University
k = complementary elliptic modulus of Texas, (May, 1966).
(k =~) 7. Iwagaki, Y., Hyperbolic Waves and
K (k) = complete elliptic integral of Their Shoaling, Proceedings, Eleventh
first kind
Conference on Coastal Eng~,
L = wave length Ch. 9, 1968T
sn( ) = Jacobian elliptic functions,
sine amplitude
8, Keulegan, G. H. and G. W. Patterson,
t = time
Mathematica~ Theory of Irrotational
T = wave period Translation Waves, Journal of Re-
U = horizontal velocity search, National Bureau of Standards,
v = vertical velocity 24, 1, (Jan. 1940), pp. 47-101.
I -948
CNOIDAL WAVE THEORY FORAPPLICATION TO OFFSHORE STRUCTUIVILDESIGN
OTC 1614
Conference on Coastal Engineering?
Ch. 8, 1968~
9.
10.
11.
12.
13.
14.
15.
16.
17.
Wiegel, R. L., Oceanographical ~-
neering, Prentice Hall, Inc.,
Englewood Cliffs, N. J., 1964,
pp. 40-53.
Stokes, G. G., On the Theory of
Oscillatory Waves, Mathematical and
Physics Papers, I, Cambridge Univer-
sity Press, 1880.
Abramowitz, M. and I. A. Stegun, Ed.
Handbook of MathematicalFunctions,
U.S. Bure~ of Commerce, 1964,
p. 567-626.
Erdelyi, A., et al, Higher Trans-
cendental Functions, Vol. 2., McGraw
Hill Book Co., Inc., 1953.
Scientific Subroutine Package, IBM
Publication #H20-0205-3, 1968.
Laitone, E. V., Limiting Conditions
for Cnoidal and Stokes Waves,
Journal of Geophysical Research,
67(4), (April, 1962), pp. 1555-1564.
Masch, F. D., Cnoidal Waves in
Shallow Water, Proceedings, Ninth
Conference ~ Coastal Engineering,
Ch. 1, 1964.
Kinsman, B., Wind Waves, Prentice
Hall, Inc., E~w~liffs, N. J.,
1965, p. 133.
Dean, R. G., Breaking Wave Criteria;
A Study Employing a Numerical Wave
Theory, Proceedings, Eleventh
18.
19.
20.
21.
22.
23.
24.
Divoky, D., B. Le M6haut6 and A. Lin,
Breaking Waves on Gentle Slopes,
Journal of Geophysical Research~
vol. 75 {~9,(March 20, 1970)s
pp. 1681-1692.
Street$ R. L., and F. E. Camfi.eld,
Observations and Experiments on
Solitary Wave Deformation,
Proceedings, Tenth Conference on
Coastal~, Ch. 19, 1%6.
Iwagaki, Y. and T. Sakai,
Horizontal Water Particle Velocity
of Finite Amplitude Waves,t
Proceedings, Twelfth Conference ~
Coastal Engineering, Ch. 19, 1970.
Divoky, D., Personal Communications,
1971-1972.
Hendrickson, J. A., Personal
Communications, 1971.
Masch, F. D. and R. L. Wiegel,
Cnoidal Waves, Tables of Functions,

Council on Wage Research, The
Engineering Foundation, University
of California, 1961.
Agershou, H. A. and J. J. Edens,
Fifth and First Order Wave lforce
Coefficients for Cylindrical Piles,
Proceedings, Santa Barbara Specialty
Conference on Coastal Engineering,
ASCE, Ch. 1~1965
- J
NU
I I I I I !
,-
.
NIv
,:
_.
0 0
N]u
:\
1 1 1 1 1=-1 I

You might also like