.

CHK-17 8/1/97 2:53 PM Page 189

International Journal of Offshore and Polar Engineering

Vol. 7, No. 3, September1997 (ISSN 1053-5381)
Copyright by The International Society of Offshore and Polar Engineers

Numerical Simulation and Spectral Analysis of Irregular Sea Waves

Celso K. Morooka and Irineu H. Yokoo
State University of Campinas, Campinas, So Paulo, Brazil
ABSTRACT
The most used methods of numerical simulation of linear irregular sea waves are studied, including the Wave
Superposition Method and inversions of FFTs. An alternative method applying Walsh functions is also used. The generated
wave records are analyzed applying the well-known FFT Method and the Maximum Entropy Method. Results of the simulations are presented for the Deterministic (DSA) and the Nondeterministic (NSA) models. The results of the simulations are
compared from the point of view of computational efficiency, estimated spectra and statistical properties of the generated
records. Finally, the wave records are used in the motion analyses of a semisubmersible platform in time domain.

INTRODUCTION
In the study of the dynamic behavior of floating structures in
irregular waves, two steps are of great importance: the digital simulations of the sea surface elevations, and the spectral analyses of
the generated sea waves time series. And recently great effort has
been spent to successfully represent ocean waves through computational programs, and several methods of spectral analysis and
statistical calculations have been developed to analyze the wave
records. In the present work, some of these methods of wave simulation and spectral analysis are shown and applied in the study of
the motions of floating structures.
Sea waves are of stochastic nature and, mathematically, are
represented as Gaussian stationary and ergodic processes. The
models used to represent these processes are based on the work of
Rice (1944, 1945), who presented two formulations
(Deterministic Spectral Amplitude and Nondeterministic Spectral
Amplitude) to represent electronic signals as Gaussian stationary
processes from previously specified energy spectra. These methods are largely used in the simulation of irregular sea waves and
are known as the Wave Superposition or the Summation models;
they were applied by Morooka and Yokoo (1992) for the simulations of directional waves. Borgman (1969) applied these models
and introduced the simulation through white noise filtering in
time domain. Shinozuka (1971) has shown a method to simulate
a class of multivariate and multidimensional random processes
with specified cross-spectral densities. Later, Wittig and Sinha
(1975) presented a method of synthesis of random processes using
Fast Fourier Transforms (FFTs) of the spectrum in frequency
domain. This method was improved by Hudspeth and Borgman
(1979) with the application of FFT routines for real numbers.
Simulated or measured time domain wave records can be used
for the estimation of wave spectra. There are several methods for
spectral estimation; the most used are the Traditional Methods
based on Fourier transforms of the signals in time domain. The
Parametric Methods are another well-known family of methods
that include the AR, MA and ARMA. Other methods used in signal processing are the Maximum Likelihood Method (MLM) and
Received March 10, 1995: revised manuscript received by the editors
April 12, 1997. The original version (prior to the final revised
manuscript) was presented at the Fifth International Offshore and Polar
Engineering Conference (ISOPE-95), The Hague, The Netherlands,
June 11-16, 1995.
KEY WORDS: Sea waves, simulation of irregular waves, wave statistics,
floating platform dynamics.

the Maximum Entropy Method (MEM). For Gaussian stochastic

processes the MEM is equivalent to the AR.
In this work, the Wave Superposition and FFT methods are
applied for the simulation of irregular wave records. An alternative technique applying Walsh functions is used similarly to the
Superposition Method. The wave records are analyzed through
the Direct Method, which is the most common Traditional
Method, and through the MEM. The main statistical properties are
calculated from the wave records.

WAVE SUPERPOSITION METHOD

Ocean waves are treated as zero mean stationary Gaussian processes, represented in frequency domain by energy spectra. Rice
(1944, 1945) presented mathematical models to represent electronic noise of Gaussian distribution with zero mean and variance
determined from the white noise spectrum filtered by a specified
spectrum. Later these models were applied to sea waves simulations, being known as the Wave Superposition Method or the
Summation Method.
The first method suggested by Rice represents the sea surface
elevation (h(t)) as a discrete Fourier series:
h (t ) =

(a cosw t + b sinw t)
n

(1)

n =1

where an and bn = independent random variables that are distributed normally around zero with the standard deviation =
s n = S(w n )Dw ; S(w) = specified wave spectrum; and w = 2pf
= frequency (f > 0).
The Fourier coefficients are calculated by filtering the white
noise spectrum with the specified one. As the white noise spectrum is unitary, the filtered spectrum is equal to the specified
spectrum. The coefficients an and bn are random variables, so Eq.
1 is called Nondeterministic Spectral Amplitude (NSA). The second model for Gaussian signals introduced by Rice is represented
by:
h (t ) =

A cos(w t - e )
n

(2)

n =1

where An = 2 S(w n )Dw ; and e = phase angle distributed at random (0 < e < 2p).

.CHK-17 8/1/97 2:53 PM Page 190

Numerical Simulation and Spectral Analysis of Irregular Sea Waves

190

It can be noted that the amplitude of each component An is

immediately determined from the wave spectrum, and Eq. 2 represents the Deterministic Spectral Amplitude (DSA) model. When
N, Rice has shown that the DSA and NSA models present the
same statistical properties.
The first step when building an irregular wave simulator with
the Superposition Method is to specify the wave spectrum. The
Nyquist theorem must be obeyed for the determination of the
maximum frequency (fmax = 1/2Dt). As Dt is constant, the specified spectrum is restricted to a finite range of frequencies.
The next step is to divide the spectrum in N spectral bands (wn,
wn+1), and for each band a representative frequency (wn) must be
determined. The total variance of each band must be concentrated
in only one representative frequency. In frequency domain, each
spectral band is reduced to a Diracs delta function corresponding,
in time domain, to a sinusoidal component with random phase.
The choice of the representative frequency (wn) is of great importance for the result of the simulation, because the concentration of
the total variance of the band in only one sinusoid with random
phase causes significant loss of information. Then, an infinite
simulation time would be necessary to recover all the information
contained in a continuous spectrum. There are several criteria to
determine the representative frequency of a spectral band.
Usually, nonconstant frequency intervals are used to avoid periodicity in the realizations. In this work, a small random variation
around the arithmetical mean of the band extremes is used.
Another largely used criterion is to take as representative frequency that located in the center of the variance of the band.

FAST FOURIER TRANSFORM METHOD

Borgman (1969) and Wittig and Sinha (1975) introduced the
method of simulation of Gaussian processes using the inversion of
FFTs. The technique is based on the generation of frequency
domain functions corresponding to the Fourier transforms of the
desired process. Then, inversions of FFTs are applied to generate
the records of the desired process is time domain. The main
advantage of this technique is the processing time saving.
However, for the application of FFTs, the time variable (t) must
be uniformly spaced, the same occurring with the frequency, so
the generated time series is periodic because the wave components are harmonic.
Tuah and Hudspeth (1982) presented equations of this method
for the synthesis of wave records using the models of Eqs. 1 and 2
proposed by Rice. In frequency domain, the white noise spectrum
is filtered by the specified spectrum to generate Gaussian random
waves in time domain.
The NSA model given by Eq. 1 can be rewritten applying the
following trigonometric identities:
1/ 2

DSA and NSA models, can be expressed as Fourier pairs with

appropriated functions in frequency domain. The Fourier pairs
have the forms:
2p (n -1)( j -1)
1 N
Cn expi
,

N n=1
N
N
2p (n -1)( j -1)
Cn = h ( j )exp-i
,

N
j =1

h ( j) =

j = 1,2,, N

(6)

n = 1,2,, N

(7)

where the complex Fourier coefficients (Cn) represent the discrete

Fourier transforms of the wave records (h(t)). So, once having
determined the coefficients (Cn), the inverse Fourier transform is
applied to synthesize the wave records. For the application of
FFTs routines, the number of discrete points N must be equal to
the integer 2 elevated to an integer power (N = 2m, m is an integer). As the sea surface elevations are real numbers, by Fourier
transforms properties: C N+2n = C n*, where the superscript *
denotes a complex conjugated. The wave zero mean in time
domain is represented in frequency domain, making C1 = 0. The
periodicity of the generated sequence implies that CN/2+1 must be
real, representing the Nyquist frequency. Both models NSA and
DSA are represented in frequency domain by Eq. 7. However, the
models still maintain their differences in the records synthesized
by FFTs inversion, because the steps followed to determine their
Fourier coefficients are different, as shown in the following.
The complex Fourier coefficients for the NSA model are calculated by:
R
Cn = N n-1 exp(iq n-1 ),
2

n = 2,3,,

N
+1
2

(8)

where Rn and qn are given by Eqs. 3 and 4, and CN+2n = Cn*.

For the simulation of the sea surface elevation applying the
NSA model, it is necessary to generate a pair of random numbers
(an and bn), of Gaussian distribution and zero mean, N(0,1). The
random variables (an and bn) are calculated from an and bn, representing the nondeterministic amplitudes:
1/ 2

an = a n [ S(w n )Dw ]

(9)

1/ 2

bn = bn [ S(w n )Dw ]

(10)

Finally, Rn and qn are calculated, allowing the determination of

the coefficients Cn. Then, the FFT routine is applied to synthesize
the sea surface elevations (h(t)).
For the DSA model, the random characteristic of the sea waves
is represented only by the phase angles uniformly distributed
between 0 and 2p. The amplitudes are totally deterministic, given
by:

Rn = (an2 + bn2 )

(3)

bn
an

(4)

N
Cn = N S(w n-1 )Dw exp(iq n-1 ), n = 2,3,, +1
2
2

(5)

Having determined the Fourier coefficients, the inversion of

FFT is applied to give the sea surface elevations through the DSA
method.
The simulations of waves through Fourier transforms are
equivalent to filtering the Gaussian white noise by the specified
spectrum in frequency domain. The DSA model simulates the
Gaussian white noise spectrum, while the NSA model simulates a

q n = arctan

1/ 2

Then, Eq. 1 can be written as:

h (t ) =

R cos(w t +q )
n

n =1

Applying complex notation, Eqs. 2 and 5, representing the

(11)

.CHK-17 8/1/97 2:53 PM Page 191

International Journal of Offshore and Polar Engineering

random variation around the white noise spectrum. Theoretically,
both models require infinite simulation time to generate the spectra with the desired characteristics. However, the specified spectra
have bands limited by the Nyquist frequency, and the simulation
time is truncated for a finite value. In this case, the NSA model
produces time sequences with greater variation than the DSA
model.

STACKED FFT
The wave simulations are efficiently carried out with the initialization of the complex Fourier coefficients in frequency domain,
followed by the inversion of FFTs to generate time sequences, as
described previously. An important characteristic of the generated
records is that the sea surface elevations are sequences of real
numbers in time domain. Thus, the complex Fourier coefficients
present even real components and odd imaginary components in
frequency domain, and the application of FFTs produces complex
numbers with imaginary components equal to zero in time
domain. Based on this property, Hudspeth and Borgman (1979)
developed a method of initialization of the Fourier coefficients
that uses just half the computer memory for the FFT synthesis
than for the method presented previously.
In frequency domain, the Fourier coefficients (Cn) are modified
to be represented by a new sequence of complex coefficients (Tk)
producing a stacked Fourier pair:
4p (k -1)(l -1)
2 N /2
Tk exp i
,

N k =1
N
N /2
4p (k -1)(l -1)
Tk = h (l )exp-i
,

N
l =1

h (l ) =

l = 1,2,,

N
2

(12)

N
2

(13)

k = 1,2,,

The relation between the coefficients Cn and Tk is not simple,

but Hudspeth and Borgman have shown that the stacked FFT
coefficients (Tk = RTk iITk) can be represented in the range (1
k N/2) in function of the coefficients (Cn = RCniICn) known in
the range (1 n N):
2pk
1
RCk + RCN / 2-k - sin
( RCk - RCN / 2-k )
N
2

2pk
+ cos
( IC + ICN / 2-k )
N k

2pk
1
ITk = ICk + ICN / 2-k - sin
( IC - ICN / 2-k )
N k
2
2pk

- cos
( RCk + RCN / 2-k )
N

N
k = 2,3,,
2

RTk =

1
( RC1 + RCN / 2 )
2
1
IT1 = - ( RC1 - RCN / 2 )
2

And also RT1 =

(14)

191
WALSH FUNCTIONS
The development of digital computers leads to the application
of orthogonal functions in the solution of numerical problems.
Sine and cosine orthogonal systems are of fundamental importance in these theories, and the Fourier analyses are the most
important mathematical tools in signal processing and analysis.
Another well-developed family of orthogonal functions with the
evolution of computers (including hardware) is that of the binary
functions, which represent series of rectangular pulses or square
waves. The Walsh functions form an ordered group of rectangular
waves with only two possible amplitude values: 1 or +1. The
Walsh functions can be divided in odd and even functions:
WAL(2 k,t ) = CAL(k,t ),
WAL(2 k -1,t ) = SAL(k,t ),
k = 1,2,, N /2
These functions define two new Walsh series similar to sines
and cosines. The CAL and SAL functions are of simple representation (specially for hardware) and can easily represent stepped
functions. However, the trigonometric functions present better
accuracy when representing smooth functions. More details about
Walsh functions and their applications are presented by
Beauchamp (1984).
The simulation models represented by Eqs. 1 and 2 can be
applied through substitution of the sines and cosines by SAL and
CAL functions, as suggested by Totsuka (1993). However, the direct
summation of rectangular functions produces stepped waves. This
problem can be attenuated through the representation of sines and
cosines with summations of Walsh functions with precalculated
amplitudes, as proposed by Kitai (1975). This technique produces
wave records close to those produced by sines and cosines.

ESTIMATION OF WAVE SPECTRUM

The sea surface elevation records (h(t)) obtained through real
measurements, and numerical or experimental simulations, can be
used to produce the wave spectra. The methods of spectral estimation are frequently studied in signal processing as based on
Fourier transforms of time domain functions. In the well-known
Direct Method, the FFT is applied to the wave record (hj) and the
spectrum is determined with:

(15)
PPER ( f ) =

(16)
(17)

Through inversion to time domain, the discrete sequence contained in both real and imaginary parts represents a random process with the same length as that generated from the nonstacked
method. The advantage is that the stacked FFT requires just half
the computational memory of the method presented before.

(18)

N
1
Dt h j exp( -i 2pfjDt )
NDt j =1

(19)

With the development of FFT routines the Direct Method

became computationally very efficient and the most popular
method of spectral analysis.
In spite of its qualities, the Direct Method presents some problems inherent to the FFT calculations. The application of FFT
routines requires that wave records be truncated on a finite value
N. This is equivalent to applying a rectangular window on the
records, where the windowed values are the original records for 1
j N, and to assuming zero values for j < N. In frequency
domain it is the same as making the convolution of the Fourier
transform of the signal with the transform of the window. If the
power of the signal is concentrated on a certain band of frequency, the convolution will cause a spreading of power over adjacent
frequencies. This phenomenon is known as leakage and is inher-

.CHK-17 8/1/97 2:53 PM Page 192

Numerical Simulation and Spectral Analysis of Irregular Sea Waves

192

ent to the spectral estimation via FFT. The effects of leakage can
be reduced with the application of windows of nonuniform
weights on the wave records. In this work the Hanning window
was used:
h (t ) =

1
2pt
1- cos

2
T

(20)

Statistically more consistent results can be obtained with the

application of some kind of averaging or smoothing of the estimated
spectra. Welch (1967) suggests the division of the records in blocks
and the application of averaging of the spectra obtained from each
block for the final estimation of the spectrum ( S ( f ) ). Welch suggests also that the extremes of the blocks can be overlapped to avoid
any loss of information due to windowing of records.

where apk are the predictor parameters and sp2 is the prediction
error power. The determination of the order of the predictor, the
coefficients apk and the constant sp2 are the crucial points of the
maximum entropy spectral estimation.
Andersen (1974) presented an algorithm to determine the coefficients apk based on the method proposed by Burg, through the
recursive equation:
apk = ap1k appap1pk
where ap0 = 1and apk = 0 for k p, and:
N-p

a pp =

1+ a pk exp( -i 2pfkDt )
k =1

t =1

N-p

2
pt

2
pt

+ b

(23)

t =1

As discussed above, the application of FFTs on the estimation

of spectra through the Traditional Methods requires truncation of
wave records that produce inconsistent results. The application of
modeling on the estimation of the spectra allows the assumption
of more realistic hypotheses for the process out of the measured
range. So, the application of windows can be avoided and more
coherent results obtained.
Among the most used methods of spectral analysis are the
Parametric Methods. Their application is based on the extrapolation of the wave records for infinite time, assuming that the
extrapolated values are expressed by: a weighted sum of past values plus a noise term (auto-regression - AR), a moving average
(MA) of the past values, or a combination of both processes
(ARMA). From the point of view of modeling, the spectral analysis is based on three steps: selection of the extrapolation model in
time domain (AR, MA or ARMA), estimation of parameters, and
spectral estimation through the substitution of the parameters for
the equations of the chosen model.
Van den Bos (1971) showed that for Gaussian random processes the AR estimation is equivalent to the Maximum Entropy
Method (MEM) proposed by Burg (1967, 1968). In this work the
parametric model will be studied from the point of view of the
maximum entropy, which is more commonly used on sea wave
processing.
Following Burg, the MEM is based on the extrapolation of a
finite segment of the autocorrelation function for unknown values.
In this way, the distortions on the autocorrelation function can be
avoided through the extrapolation of the finite wave records for
infinite length. The method proposed by Burg is based on the
extrapolation of these values so that the sequence in time domain
characterized by the autocorrelation function will have maximum
entropy. So, in a given sequence with p+1 known values of the
autocorrelation function {Rxx(0), Rxx(1),, Rxx(p)}, the value
Rxx(p+1) is determined so that the sequence is the most random.
The spectrum obtained with this method is called the maximum
entropy spectral estimation (PMEM(f)).
The maximum entropy spectrum is obtained by maximizing the
entropy under the constraints that the known values of the autocorrelation function satisfy the Wiener-Khinchine relations. The
solution is found by the Lagrange multiplier technique:
s p2Dt

2 bpt bpt

(b

MAXIMUM ENTROPY METHOD

PMEM ( f ) =

(22)

where bpt = bp-1t - a p-1 p-1bp -1t

bpt = bp -1t +1 - a p-1 p-1bp-1t +1

(24)
(25)

are initialized by b1t = ht and b'1t = ht+1.

The prediction error power is also calculated recursively from
the predictor coefficients:
2
s p2 = s p2-1 (1- a pp
)

(26)

where the first term is given by:

s 02 =

1 N 2
ht
N t =1

(27)

For the application of the method proposed by Andersen it is

necessary first to find the order of the process (p); that is equivalent to finding the number of coefficients necessary for prediction.
Several methods have been proposed in order to determine the
number of coefficients necessary to represent the autoregressive
processes. One of the best-known is the final prediction error
(FPE) introduced by Akaike (1969), and defined as:
N + ( p +1)
FPE[ p ] = s p2

N - ( p +1)

(28)

where the selected order p is that for which the FPE is minimum.
In Marple (1987) some other methods can be found, but no consensus has been reached about their efficiency. Holm and Hovem
(1979) suggest a test with the partial autocorrelation function
coefficients:
Rxx (t ) =

R (t - k )a
xx

tk

(29)

k =1

where Rxx(0). Houmb and Overvik (1981) call the partial autocorrelation coefficients of PAKF and suggest that their values can be
compared with the variance of the partial auto-correlation, given
by:

(21)
s2 =

1
N

(30)

.CHK-17 8/1/97 2:53 PM Page 193

International Journal of Offshore and Polar Engineering

Fig. 1 Irregular wave simulation by DSA

193
records were simulated using the Wave Superposition, FFT inversions and Walsh functions. The specified wave spectrum is the
ITTC with Hs = 7.5 m and Ts = 10 s. Fig. 1 shows part of a simulation with the DSA model with Wave Superposition and with
Walsh functions. It can be noted that the simulation via Walsh
functions produces a record very close to that produced via Wave
Superposition.
Fig. 2 shows the results of the simulations via Wave
Superposition and via Walsh functions applying the NSA model.
The differences between the records are due to the substitutions of
the sine and cosine functions by the summations of Walsh functions.
Fig. 3 presents the simulated record using the DSA model via
FFT inversion. The application of the FFTs has shown it to be the
fastest method. The Wave Superposition simulations have spent
twice the computer time of the FFTs, while the Walsh functions
were up to 20 times slower. The application of the stacked FFT
has produced wave records exactly equal to the standard FFT and
without any advantage in computation time. However, the stacked
FFT routine requires about half the computer memory.
Wave Statistics

Fig. 2 Irregular wave simulation by NSA

Fig. 3 Irregular wave simulation by DSA

The routine tests the values of PAKF and when 3 out of the 5
last tested values are greater than s, the order of the process is
determined.

RESULTS
Wave Simulation
Following the formulations shown above, irregular wave

For the statistical analyses of the wave records, simulations

were carried out with 8192 points spaced at 0.5 s, for the ITTC
wave spectrum with Hs = 7.5m and Ts = 10 s. The DSA model
results are shown in Table 1.
The obtained values are very close to each other except for the
zero upcrossing periods of the Walsh simulations, which present
lower values than the other simulations. The explanation for this
problem can be found by examining Figs. 1 and 2, which show
that the Walsh simulations produce stepped waves, changing the
zero upcrossing periods.
The values obtained for the skewness and kurtosis indexes are
very good when compared to the Gaussian distribution. For the
skewness index, recall that values close to zero indicate good
symmetry and absolute values close to unity indicate high asymmetry; for the kurtosis index, values close to 3 indicate distributions close to the Gaussian (mesokurtic), greater values indicate
concentration of data near the mean (leptokurtic), and smaller values indicate flatter distribution (platikurtic). Fig. 4 illustrates the
histogram for wave surface elevations simulated by DSA with

X
XMAX
XMIN
s
HMEAN
H1/3
H1/10
T0
TP
SK
K

(m)
(m)
(m)
(m)
(m)
(m)
(m)
(s)
(s)

SUPER
0.318E-5
6.212
6.305
1.872
4.623
7.179
8.881
9.648
6.943
0.0430
2.887

WALSH
0.369E-3
6.403
6.918
1.934
3.750
6.820
8.973
6.011
1.964
0.0173
2.879

FFT
0.589E-6
6.604
6.935
1.875
4.604
7.164
8.878
9.422
6.030
0.0011
2.862

Table 1 Input data mean value X , maximum value of input data

XMAX, minimum value of input data XMIN, standard deviation s,
mean wave height HMEAN, mean of 1/3 of highest waves H1/3,
mean of 1/10 of highest waves H1/10, zero upcrossing period T0,
peak to peak period Tp, skewness SK and kurtosis K. Wave
heights are zero upcrossing defined.

.CHK-17 8/1/97 2:53 PM Page 194

Numerical Simulation and Spectral Analysis of Irregular Sea Waves

194

Fig. 4 Histogram for wave surface elevation simulated by DSA

via Superposition, and Gaussian probability density function

Superposition, and the Gaussian probability density function. The

comparison of the two shows a good agreement.

SPECTRAL ANALYSIS
Complementing the statistical analyses, the simulated wave
records were used to estimate the wave spectra, which can be
compared with the specified one (ITTC with Hs = 7.5m and Ts =
10 s). By applying the Direct Method with Hanning spectral windowing, averaging 15 blocks of 1024 points with overlapping of
half a block, the following spectra were estimated. From the estimated spectra, the moments are calculated and can be compared
with those from the specified spectrum, shown in Table 2.

M0
M1
M2
M3
M4

(m2)
(m2/s )
(m2 /s2)
(m2/s3 )
(m2/s4 )

3.513
2.156
1.536
1.351
1.549

Table 2 Moments of specified wave spectrum

Fig. 5 Estimated wave spectrum (DSA with Superposition)

Fig. 6 Estimated wave spectrum by MEM-FPE (DSA with

Superposition)
Fig. 5 shows the estimated spectrum of a DSA simulation via
Wave Superposition, and the specified spectrum. It can be noted
that there is an excellent visual match between the spectra, which
is confirmed by the spectral moments.
The MEM was applied using both FPE and PAKF criteria for
the determination of the order of the processes, using records of
1024 points. Trial and error procedure was applied to define the
order of the process used in the present analysis. The order was
obtained when the best fit between output and input spectra is
reached. Fig. 6 presents the estimated spectrum with the MEMFPE, which presents an exaggerated peak; however, if compared
with the targeted results, a reasonable visual and a good spectral
moments agreements could be observed. The application of the
FPE on the determination of the order of relatively long processes
(1024 points) produces a high number of coefficients (IP = 155 in
this case), elevating processing time.
Fig. 7 shows the estimation with the MEM-PAKF, which produced a spectrum of lower order (IP = 47) with some variations
around the peak frequency. This estimation also presented good
values for the spectral moments. The processing time in both FPE
and PAKF estimations were of the same order as the Direct
Method, although the MEM used shorter records. This is due to
the necessity of calculating the coefficients before the application
of the MEM estimators.

Fig. 7 Estimated wave spectrum by MEM-PAKF (DSA with

Superposition)

.CHK-17 8/1/97 2:53 PM Page 195

International Journal of Offshore and Polar Engineering

In general, all three methods of signal processing produced
good estimations for the spectra, in part because the DSA model
produces sequences of relatively small variation. The Direct
Method produced very good estimation as expected when applying data windowing and block averaging of long records (8192
points). The MEM does not produce estimations as well as the
Direct Method does, because it uses only part of the information
(1024 points), being recommended more in cases when only short
records are available.
Application Example
An application of numerical simulations of irregular waves is
the time domain motion analysis of floating structures. As an
illustration, some results obtained by applying the simulated wave
records in the motion simulation of a semisubmersible platform
developed by Morooka and Vardaro (1992) will be shown. The
source and sink model is used to determine the first- and secondorder forces, and the equations of motions of the platform are
integrated applying the Wilson-q method.
Table 3 presents the floating platforms main dimensions. Fig.
8 shows the sway motion of the semisubmersible in a beam irregular sea simulated through DSA with Wave Superposition. Table
4 shows the statistics for the sway motion for both cases DSA
and NSA simulations with Superposition. It can be noted that the
NSA produces greater variations on the motion results. From a
careful observation of the wave statistics of simulation results
through NSA and DSA, it is noted that random characteristics are
better obtained in the NSA than in the DSA case, in which differences on the motion simulation results arise. Further investigations are needed to clarify this phenomenon.

195

X
XMAX
XMIN
s
HMEAN
H1/3
H1/10
T0
TP

(m)
(m)
(m)
(m)
(m)
(m)
(m)
(s)
(s)
SK
K

DSA
0.656E-2
2.546
2.420
0.786
1.889
2.837
3.502
13.63
10.97
0.0555
2.823

NSA
0.905E2
2.865
2.968
0.966
2.195
3.185
3.827
16.75
10.62
0.0106
2.642

Table 4 Statistics of sway motion

wave generation in a tank basin.

For spectral analysis, the traditional Direct Method was used
for long records, producing good results when applied with block
averaging and data windowing. For short records when averaging
is not possible, the MEM was applied, producing reasonable
results.
The wave simulation methods produced good statistical results.
The importance of the quality of the simulated records was shown
in the application example, which has shown that the greater variation of the NSA model was transferred for the motions of the
platform. More detailed investigation is needed on simulation of
platform motions by using the wave simulation methods here discussed.

REFERENCES
DISPLACEMENT
LENGTH
BEAM
DRAFT

40000 ton
115 m
61.5 m
23.5 m

Table 3 Semisubmersible platform data

CONCLUSIONS
The present work discusses the main methods of irregular
waves simulation and wave records analysis, for DSA and NSA
models. The method of the inversion of FFT has shown itself to
be computationally more efficient, being two times faster than the
Wave Superposition. The Walsh functions have not been competitive with the other methods; however, the study of these functions is suggested by references to application in hardware for

Fig. 8 Simulated sway motion (DSA via Superposition)

Akaike, H (1969). Fitting Autoregressive Models for Prediction,

Ann Inst Statist Math, No 21, pp 243-247.
Andersen, N (1974). On the Calculation of Filter Coefficients for
Maximum Entropy Spectral Analysis, Geophysics, Vol 39, pp 6972.
Beauchamp, KG (1984). Applications of Walsh and Related
Functions, Academic Press.
Borgman, LE (1969). Ocean Wave Simulation for Engineering
Design, J Waterways and Harbors Div, ASCE, Vol 95, No WW4,
pp 557-583.
Burg, JP (1967). Maximum Entropy Spectral Analysis, Proc 37th
Meeting Soc Explor Geophysicists, reprinted in Modern Spectral
Analysis, D G Childers, ed, IEEE Press, 1978, pp 34-41.
Burg, JP (1968). A New Analysis Technique for Time Series Data,
NATO Advanced Study Institute on Signal Processing with
Emphasis on Underwater Acoustics, reprinted in Modern Spectral
Analysis, D G Childers, ed, IEEE Press, 1978, pp 42-48.
Holm, S, and Hovem, JM (1979). Estimation of Scalar Ocean Wave
Spectra by the Maximum Entropy Method, IEEE J Oceanic Eng,
Vol OE-4, No 3, pp 76-83.
Houmb, OG, and Overvik, T (1981). Some Applications of the
Maximum Entropy Spectral Estimation to Ocean Waves and Linear
Systems Response in Waves, Appl Ocean Res, Vol 3, No 4, pp
154-162.
Hundspeth, RT, and Borgman, LE (1979). Efficient FFT Simulation
of Digital Time Sequences, J Eng Mech Div, ASCE, Vol 105, No
EM2, pp 223-235.
Kitai, R (1975). Synthesis of Periodic Sinusoids from Walsh Waves,
IEEE Trans Instr Meas, Vol IM-24, No 4, pp 313-317.

.CHK-17 8/1/97 2:53 PM Page 196

Numerical Simulation and Spectral Analysis of Irregular Sea Waves

196

Marple, SL (1987). Digital Spectral Analysis with Applications,

Prentice Hall.
Morooka, CK, and Vardaro, E (1992). Time Domain Simulation of
Horizontal Motions of a Moored Floating Platform in Waves, Wind
and Current (in Portuguese), SOBENA92, Rio de Janeiro, pp 129155.
Morooka, CK, and Yokoo, IH (1992). Behavior of a SemiSubmersible Platform with Marine Riser in Directional Waves, 2nd
Int Offshore and Polar Eng Conf, San Francisco, ISOPE, Vol III, pp
628-635.
Rice, SO (1944, 1945). Mathematical Analysis of Random Noise,
Bell System Tech J, Vols 23 and 24, reprinted in Selected Papers on
Noise and Stochastic Process, Dover Pub, 1954, pp 1-50.
Shinozuka, M (1971). Simulation of Multivariate and Multidimensional Random Processes, J Acoust Soc America, Vol 49, No
1, pp 357-368.

Totsuka, Y (1993). A Study on Numerical Techniques of

Multidirectional Waves Simulation (in Japanese), Master Thesis,
Nihon Univ.
Tuah, H, and Hundspeth, RT (1982). Comparisons of Numerical
Random Sea Simulations, J Waterway, Port, Coastal and Ocean
Div, ASCE, Vol 108, No WW4, pp 569-584.
Van den Bos, H (1971). Alternative Interpretation of Maximum
Entropy Spectral Analysis, IEEE Trans Info Theory, Vol IT-17, pp
493-494.
Welch, PD (1967). The Use of Fast Fourier Transform for the
Estimation of Power Spectra: A Method Based on Time Averaging
over Short, Modified Periodograms, IEEE Trans Audio and
Electroacoustics, Vol AU-15, pp 70-73.
Wittig, LE, and Sinha, AK (1975). Simulation of Multicorrelated
Random Processes Using the FFT Algorithm, J Acoustic Soc of
America, Vol 58, No 3, pp 630-634.

Proceedings of the 4th (1996)

Pacific/Asia Offshore Mechanics Symposium
ISOPE PACOMS-96
Pusan, Korea, October 31-November 2, 1996
Hydrodynamics, Offshore Engineering, Coastal and Environmental Engineering, Materials and
Welding, Fatigue and Fracture, and Reliability.
A single volume (ISBN 1-880653-27-3), 325 pp., US$60 (member; US$50.) is available from ISOPE,
P.O. Box 1107, Golden, Colorado 80402-1107, USA (Fax 1-303-420-3760).

To receive a Membership Application Form, fill in the information below:

Name (First and Surname)
Co./University
Mailing Address

Also, I would like to receive a Journal sample copy for examination.

Mail or fax to: ISOPE Membership Dept., P.O. Box 1107, Golden, Colorado 80402-1107, USA
(Fax +1-303-420-3760)