Professional Documents
Culture Documents
42 Exam #1
Spring 2004
Open notes, book (1h 20mins)
Problem 1) (15%)
Can you apply Morisons formula and the LTI system theory together?
Consider the equation as a system with an input, the fluid velocity, u(t), and
an output, the force, f(t), on the system. Justify your response.
Note that according to Morisons equation the force, f(t), on a structure is
given as a function of the fluid velocity, u(t):
f (t ) = (m + ma )
u (t ) 1
+ 2 Cd Ap u (t ) u (t )
t
Problem 2) (20%)
An LTI system has an impulse response h(t ) = h0 uo (t t0 ) where uo (t ) is the
dirac delta function.
a) Determine the transfer function of the system, H ( ) .
b) Given an input to this system, x(t ) = xo sin(ot + o ) , determine the
output, y(t).
S +( )
20
15
10
5
0
0
0.5
1.5
Frequency (rad/sec)
(rad/s)
0.5
1.0
1.5
2.0
d =
0.5 rad/s
S+()
m /(rad/s)
12
24
18
6
2
(Problem 4 Continued)
d) Determine the height above which the blades would swipe through the
water less than 3 times per day (on average).
e) Is it sufficient to design the windmill so that the blades just barely skip
the waves? Do you estimate that this would be an efficient design in
general? Explain briefly.
13.42 Exam #2
Spring 2004
Open Book
1) (30%) A buoy used in the local sailing regattas on Boston harbor is left out over
night. At high tide, it is fully submerged due to the short mooring cable length
attached to it.
The buoy can be modeled in two dimensions, as a positively buoyant circle of radius
r, and density, b, connected to the sea bottom by an inextensible mooring cable of
length l. The cable length, at high tides, is such that the circle is fully submerged with
its center a distance d below the average free surface elevation.
In this region a constant current is prevalent, with speed, U, oriented in the positive xdirection, in addition to regular surface waves (propagating in the x-direction). You
can assume for this problem that the surge added mass and damping coefficients and
the excitation force are the same functions of frequency as in the absence of the
current.
a) Give the equation of motion of the circle in surge in the presence of the
current and linear gravity waves incident on the buoy in the positive xdirection. Neglect viscous effects.
b) Determine the natural frequency of the buoy in surge. Is your answer
different for the case when U = 0?
c) Given waves incident on the buoy in the positive x-direction, in addition to the
current, determine the wavelength(s) that will excite resonance in the
surge motion. How does your answer change for waves propagating in the
negative x-direction (against the current)?
Extra Credit (5 points)
d) If the mean drift (steady drag) force acting on the circle, due to incident waves
in the absence of a current, is some function D(o), what is the drift force
acting on the buoy when the current is non-zero in the positive x-direction?
(neglect viscous effects)
z
x
Uc
2) (20%) A 10cm diameter bundle of underwater fiber optic cable is installed across the
channel where a strong current, with magnitude U, is prevalent. This cable is
anchored 10 diameters above the sea floor, where there is little effect on the cable
behavior due to the presence of the sea floor.
a. Modeling the cable as a rigid cylinder free to move in the inline and
transverse directions, determine the frequency of the inline and transverse
cable vibrations, assuming that the cable bundle can be modeled as a cylinder
with a rough surface, for a velocity of 0.5 m/s incident perpendicular to the
long axis of the cable.
b. How do you anticipate your answer would change, over a range of inflow
velocities, if the cable was modeled as a flexible cylinder pinned at both ends.
c. If the cable was mounted 1 diameter from the sea floor how do you expect
your answers in parts a and b would change? Rationalize your answers
physically or mathematically.
Uc
Not to scale!
10D
1D
3) (40%) SHIP MODEL TESTING: You have been asked to evaluate the seakeeping
characteristics of a model ship in the MIT towing tank. The ship model mass is 2500
kg and the hydrostatic restoring coefficient is c33 = 1.4x104 N/m.
You are given results from two sets of basic tests with a stationary ship model located
a distance of 2.4 meters downstream from a wave probe used to measure the incident
wave elevation. The tank is 5 meters deep. The first set of experiments, diffraction
tests, were performed with waves of constant amplitude (a = 5 cm) and varying
frequency with a rigidly fixed model. The second set of experiments, radiation tests,
involved the vessel being forced to move in heave at constant amplitude (|x3|= 5cm)
over the same range of frequencies. For the specific case of = 3.7 rad/s you are
given the following results for the two types of tests:
Diffraction test force in heave: |F3d| = 1200 N with a phase lag of 215 from
the wave probe measurements.
Radiation test force in heave: |F3r| = 1000 N with a phase lag of 160 from the
imposed motion.
Determine:
a. The added mass and damping of the model in heave.
b. The natural frequency of the model in heave.
c. The heave response of the model (amplitude and phase) to waves of
amplitude a = 10 cm and frequency = 3.7 rad/sec.
Next you are asked to determine the seakeeping characteristics of the vessel
advancing at a forward speed U in a given sea state. The ship has length L and the
ratio of full scale ship to model ship lengths is 50:1. The full scale spectrum for
unidirectional incident waves is given by the function plotted below:
Where 1 and 2, and S are dimensional and measured relative to a fixed reference
frame in space.
d. Determine the shape of the wave spectrum you want to generate for tank
tests with your model. Consider carefully the non-dimensional parameters
you need to scale your tests. You will need to scale the testing speed and the
physical characteristics of the waves.
e. A sensor placed on your vessel indicates that the standard deviation of the
heave velocity is some value . Determine the corresponding value for the
standard deviation of heave of the full scale vessel.
Non-dimensional parameters that might come in handy:
Reynolds number
R = UL
Froude Number
Strouhal Number
F =U
gL
S = fL U
4) (10%) A ship is cruising at forward speed, U, with quartering seas incident to the
stern of the ship at 30. The waves have an absolute frequency, o.
a. Determine the combination of U and o such that the ship is stationary
relative to the ambient wave crests.
Extra Credit: (10 points)
b. If the roll resonance period is Tn, determine the range over which U can vary
to ensure that only one ambient wave frequency, o, excites resonance in roll.
Us
Dispersion Relationship
a. Write the dispersion relationship for linear free surface gravity waves.
b. In the limit where kH >> 1, how does this relationship simplify?
c. Discuss the significance of the dispersion relationship and give a physical
example of why it is important.
Problem 3:
For designing ocean systems we need to understand when extreme environmental
conditions occur. For linear free-surface gravity waves, the velocities and dynamic
pressure varies harmonically in space and time.
a. Given a free surface elevation, ( x, t ) = a sin ( kx t ) , with wavelength, = 24
m, amplitude a = 1m, non-dimensionalize and plot the following quantities for a
fixed point in space (i.e. fixed x position @ x= xo and fixed depth z=zn) but
varying time (the x-axis should be non-dimensional time and y-axis(or axes)
should be the non-dimensional quantities being plotted):
( xo , t ) , u ( xo , t ) , w ( xo , t ) , and pd ( xo , t )
Choose the depth such that zn = n /36 (meters), where n is the integer
corresponding to the month you were born in.
Line up the plots such that you can deduce the relative phase of the variables. You
can plot them on one plot or multiple subplots just make sure to label the
different curves appropriately.
b. Plot the same quantities in part a, but with x varying and at one instant in time
(i.e. @ t=to). The x-axis should be non-dimensional position in space and yaxis(or axes) should be the non-dimensional quantities being plotted.
c. At the depth determined by your birth month, what is the total pressure under the
wave crest, the wave trough, the wave nodal point (the nodal point is the point at
which the wave elevation corresponds to z = 0)?
d. The added mass force on a body is proportional to the fluid acceleration.
Determine when the horizontal added mass force is maximum, minimum, and
zero relative to the wave elevation (i.e. at the crest, the trough, or a nodal point).
e. While we are neglecting viscous forces for the most part in this course they are
still important in certain applications. In general the non-linear, viscous wave
forces are proportional to the square of the fluid velocity. When is the horizontal
viscous wave force maximum under the wave crest, the wave trough or the
wave nodal point? When is the vertical viscous wave force maximum?
Problem 4:
A series of monochromatic (single-frequency) linear free surface gravity waves, with
amplitude, a, and frequency, , is propagating over a spherical instrument dome, with
diameter 0.6 meters, that is fixed to the seafloor by a rigid mount (see figure 1).
a) The instrument dome is fixed halfway from the surface to the bottom. It is fitted
with a pressure transducer to measure the dynamic pressure. The signal being
sent to an observation station has the form P(t) = po cos(wt) where po = 19000
N/m2 and = 1 rad/s. Determine the wavelength and amplitude of these waves.
b) Use the information found in part (a) to determine the average kinetic, potential
and total energy per wavelength of these waves.
c) Determine the added mass force on the buoy in the vertical direction as a function
of time.
a,
Figure 1
H=18m
Problem 5:
a. Use the Euler equation to derive Bernoullis equation for Ideal flow. State your
assumptions. Explain the physical significance of each term in your final
equation.
b. Linearize Bernoullis equation for linear free surface waves. Justify your
linearization by comparing the relative magnitude of the various terms.
Problem 6:
A wave maker generates a packet of monochromatic, linear free surface waves with
amplitude, a = 0.25m, and wavelength, = 4 m, in a still tank. The waves propagate
towards a group of 3 students dangling their feet in the water. The students, student A,
student B and student C, are spaced along the tank at 5meters, 10meters and 15 meters
away from the wave maker, respectively.
a. Assuming deep water, how long does it take, from the time the initial wave was
generated, until each student, A, B, & C, feels the presence of the waves?
b. Can the students use the information about the time it takes the waves to reach
their feet to discern anything about the velocity of the wave packet? What about
the velocity of the wave crests?
c. If not, could they reposition themselves along the tank in order to determine the
velocity of the wave packet and the wave crests? Given that student A is not
willing to move, where would students B & C have to in order to make these
determinations?
1. Determine whether the following systems are linear and/or time invariant.
t +
a.
y (t ) =
u( s)ds
0
t +
b.
y (t ) =
[u( s )] ds
2
du(t ) du(t )
dt
dt
c.
y (t ) =
d.
y (t ) + y (t ) + y (t ) = u(t )
a.
a1 cos t
b.
sin 5t
a2 cos( t + )
2 cos(10t + )
3. Fourier Transform
a. Find the Fourier Transform of f (t ) = u0 (t ) .
b. Given that f ( x ) 0 as x , and the Fourier Transform of f ( x ) is f ( ) ,
df
? (Hint: Use partial integration.)
dx
d2 f
df
?
0 as x , what is the Fourier Transform of
c. Given that
dx 2
dx
4. Transfer Function:
a. Given the following linear system:
mx + cx + kx = f (t )
where input f (t) = Re{Fei t } and response x(t ) = Re{Xei t } , and X and F are
both complex quantities, find the transfer function H().
b. Using the same system, for which you have just found the transfer function, if
the input is f1 (t ) + f 2 (t ) determine the system output, x(t).
5. Convolution
Perform the following convolutions (from page 2.11 in Triantafyllou and
Chryssostomidis, Environmental Description, Force Prediction and Statistics for
Design Applications in Ocean Engineering):
6. A linear time-invariant system has a transfer function H(), show that when the
input is sinusoidal with frequency o, i.e.
f (t ) = f o cos(ot + )
the output is also a sinusoidal function with the same frequency.
1. Find the probability of drawing a five-card hand in which every card is a heart.
(Assume that every possible five-card hand drawn from a standard deck of 52
cards has the same probability of being selected.)
2. Given a standard deck of 52 cards, let
A = {x | x is an ace}
B = {x | x is a heart}
D = {x | x is black}
p(AB)
p(BD)
p(AC)
p(BD)
p(CD)
3. Let the random variable X be the sum of the outcomes of two dice that are rolled
simultaneously.
a. Find the probability density function i.e., (x) = p(X = x).
b. Find the cumulative distribution function i.e., P(x) = p(X x).
c. Using the probability function, find the mean, variance, and standard
deviation.
4. Let the random variable x have a cumulative distribution function P(x) defined by
x<0
0,
2
0 x < 12
x ,
P( x) = 1
1
2 x,
2 x< 2
1,
2 x
0 x <1
otherwise
1.
2.
(t ) = A cos( t + )
where and are constants and A is a random variable with probability
density function
p A (a) =
1
e
2
a2
2 2
3.
U (t )
LTI
Y (t )
where U(t) is the input random process and Y(t) is the output random process.
a. Show that Y (t ) = h(t ) U (t ) , where h(t) is the impulse response of the
LTI system.
g ( x)p
( x )dx .)
4.
(t ) = Ai cos(i t + i )
i =1
where Ai , i , and i are the wave amplitude, frequency, and random phase
angle (with uniform distribution from 0 to 2 ) of wave component number i.
The amplitudes Ai corresponding to each frequency i may be found from a
known wave energy spectrum, S ( ) , where S (i ) = 1 2 Ai .
5.
LTI
Y (t )
where X(t) is the input random process and Y(t) is the output random process.
a. If X(t) is stationary, show that
where C > 0 and k > 0 , and the impulse response of the system,
h(t ) = e at S (t )
where S(t) represents the Heaviside unit step function, find the
autocorrelation of the response Y(t).
(Hint: Use the Wiener-Khinchine Relations)
Figure 1
a. Find the variance of the wave elevation.
b. Find the average energy flux through a plane of constant x for the given
seastate.
c. Suppose a pressure probe is fixed 10 meters below the surface. Determine
the variance of the linear hydrodynamic pressure signal recorded by the
probe.
d. Determine the average frequency of upcrossings of the plane z = 3.5 m (in
upcrossings/second).
e. Suppose a wave probe is moving with constant velocity U = 2 m/s in the
direction of the ambient waves. Sketch the spectrum of the wave elevation
recorded by the probe.
mx + cx + kx = f (t )
N
where
10 ,
10,
+
S ( ) =
30 10 ,
0,
0 <1
1 < 2
2 <3
otherwise
Determine the minimum deck clearance, h, required for an offshore platform such
that the deck is flooded less than 20 times per hour. Assume that the presence of
the platform has no effect upon the ambient waves.
4. Ocean Waves:
a. Discuss the evolution of water waves generated when wind blows over the
ocean surface from developing seas to decaying seas. What are the main
physical mechanisms for wave production by wind? Think in terms of the
boundary conditions at the air-sea interface and basic fluid mechanics.
b. Sketch a typical spectrum for developing, fully-developed, and decaying
ocean waves on one plot. Discuss briefly why the spectra differ in shape.
c. Explain what significant wave height is and why it is so widely used to
describe ocean waves and sea states.
d. Why is the Bretschneider Spectrum so widely used? What are its
limitations?
Due: 4/1/04
Part I: The goal of this exercise is to analyze the Bretschneider spectrum, a continuous
spectrum, using Matlab or another computational tool. Using the Bretschneider spectrum
given in class:
S + ( ) =
m
1.25
1.25
e
4
4
m
5
Where the significant wave height is and m is the modal frequency of the spectrum.
1. Plot the Bretschneider spectrum for four significant wave heights between 1 and 3
meters for fully developed seas.
2. Consider one significant wave height chosen in part a.
a. Compare the spectrum in part 1 with the JONSWAP spectrum for the
same significant wave height.
b. Compare the B-S spectrum for developing, fully-developed, decaying sea
states. Note that the spectrum shape can be altered by changing the modal
frequency.
3. Using one case from part a:
a. Calculate the moments of the spectrum using numerical integration (like
trapezoidal integration or similar)
b. Determine the spectrum bandwidth, .
c. Calculate the 1/Nth highest wave height where N = 10, 50, and 100.
Part II: A ship is being designed to operate in the conditions you have analyzed above.
The transfer function between the incoming waves and heave motion (heave transfer
function) H() is given by
n2
H ( ) =
2 + 2i n + n2
where n is the structures natural frequency (rad/s) and is the structural damping ratio.
2
a) Plot H ( ) = H ( ) H * ( )
2
Plot this spectrum. Use a structural damping ratio = 0.3 and a n = 1.0 rad/s.
c) How will the response change if the structural frequency is different illustrate
your explanation with plots of the response at a higher or lower natural frequency.
2. Consider a slender cylindrical buoy of diameter d and draft T fixed in deep water
subject to ambient plane progressive waves of amplitude A and frequency (see
Figure 2):
d. Calculate the moment acting on the cylinder about the seafloor mount.
e. Find the location of the center of force along the height cylinder (relative
to the seafloor).
f. In the early morning the waves have lower amplitude but similar
frequency. Repeat parts (a) through (c) for the case where the wave
1.) Floating cylindrical instrumentation buoy is moored in deep water. Given: the
wavelength of the incident wave is larger than the dimension, B, and B > b; also the
wave amplitude is less than h and b.
a. Determine the heave forces on the buoy due to the incident waves.
b. Write out the full equation of motion for this buoy if it is allowed to heave freely.
You should approximate the added mass and hydrostatic coefficients based on the
geometry, but can leave the damping as B33.
c. Give the natural frequency of the buoy in heave.
d. Sketch a qualitative plot of the heave amplitude to wave amplitude ratio, |x3/a|, as
a function of kT (k = wavenumber and T = height of lower buoy section).
Discuss possible ways to use this plot in your design assuming that your design
goal is to minimize heave.
e. Using the force determined in part a, and hydrodynamic coefficients from part b,
the total force on the body can be found by adding in the diffraction force,
estimated as follows:
F(t) = FI(t) + FD(t)
FD (t ) = ma
d
w( x = 0, z = (h + T / 2), t)
dt
Find the value of b that minimizes the heave motion at frequency = 0.7 rad/s,
given the following buoy geometry:
B=5m
T = 10 m
h=4m
T1 = T + h = 14 m
2.) Consider the following cylindrical buoy (see Figure 2) in ambient plane progressive
waves of amplitude A and frequency . Let the mass of the buoy be m and the center
of gravity be located at z = 43 T . Assume deep water.
a. Given that d (the diameter of the buoy), find the surge excitation force
and the pitch exciting moment, neglecting viscous effects.
b. Find the added mass and linear restoring coefficients Aij and Cij i,j = 1, 3, 5.
The heave added mass may be approximated by the added mass of a sphere of
equal diameter.
c. Determine the natural frequency of the buoy in heave.
d. Determine the natural frequency of the buoy in the coupled surge-pitch.
e. State the equations of motion for the system.
Now consider the following cylindrical buoy tethered to the sea floor by a cable
(Figure 3).
Assume that the weight of the buoy is less than the buoyant force, and that the
cable tension is P. The linear restoring matrix due to the cable is given as:
CCABLE
k11 0
= 0 k33
k
51 0
k15
0
k55
f. Find the added mass and linear restoring coefficients Aij and Cij i,j = 1, 3, 5.
g. Find k11 and the natural frequency of the buoy in surge.
13.42:
c
A.H.
TECHET
MATLAB will be a useful tool to know throughout the course of the semester. The rst homework
can be simplied using MATLAB and the rst lab will ask each lab group to use MATLAB prior
to the lab session to generate a waveform to study. Throughout the term there may be other
assignments which call for the use of a software package such as MATLAB. The specic use of
MATLAB is not required.
Running MATLAB on the MIT server is quite simple. At a workstation you can use the software
pull-down menus at the top of the screen or at a server prompt type add matlab press return,
then matlab.
There are some common functions that are used frequently in MATLAB. Several of these are
listed below with descriptions of their use. If you are looking for a certain function and want to
know how to use it type help command at the prompt, replacing command with the function
name. It is also useful to look at the Using MATLAB manual that ships with the PC version of
MATLAB if you have access to one. There will be a copy available at each of the lab sessions in
the Towtank. Mathworks has extensive information on their web site about the functions and using
MATLAB http://www.mathworks.com. If a problem requires complex MATLAB code, a guideline
will be included in the homework. If you choose to use a program other than MATLAB that is ne.
In MATLAB it is simple to create a script that performs a series of commands similar to a C or
FORTRAN code. Edit a le with a .m extension in the working directory. Type commands in order
you wish them performed and save the le. Note: a script is dierent from a MATLAB function in
that variables dened and calculated in a script are global, but within a function variables are only
local unless passed in and out (help function).
who: lists all variables in the MATLAB workspace. Another option is whos, which is similar
in eect to ls l in UNIX, lists the variables and their size. To look at the contents of a
variable just type the variable name at the MATLAB prompt omitting the semicolon at the
end of a line.
cd dir: changes to the desired directory, where your script is located, unless you include the
working directory in the path you must be in the correct directory to run a function or
script.
path: displays all the directories that MATLAB searches in for the functions called. If your
function is not in one of these directories you must change to that directory before calling a
function.
gure(1): make gure 1 the current plot; future plotting commands will be directed into the
window for gure 1. This is useful when you have multiple gure windows open and want a
script to plot in a window dierent from the current gure. Call the gure(#) command
before the plot command.
subplot(3,1,2): breaks the current gure window into a three by one array of distinct plotting
regions and sets the current region to be the second of the three. You can change the
current region with something like subplot(3,1,3). Can also type subplot 312 dropping
the brackets and commas. subplot 324 will make a gure with three rows and two columns
1
c
A.H.
TECHET
of plots (six total), the third number refers to which of the six plots will be made active.
Counting is from left to right, top to bottom.
hold: hold the current contents of the active gure; this means that any subsequent plot
commands will draw on top of whatever is already there rather than starting from scratch.
A second hold command will release the gure. It is also possible to type hold on and
hold o to toggle the hold state.
clf: clears the current gure window. Also removes any hold and/or subplot sectioning. This
does not toggle the hold command if hold is on it will stay on.
plot(x1, y1, -, x2, y2, , x3, y3, :): plot three curves dened by the xi and yi vectors,
each with a dierent line style (in this case, solid, dashed, and dotted). The text arguments
after the y vector for each curve can also be used to dene marker style (circle, x, *, etc.)
and curve color. Type help plot for a list of color and marker codes.
legend(curve 1, curve 2, curve 3): assign and draw the legend for curves 1 (the solid
line), etc. You can click on the legend with the mouse and move it around once it has been
drawn.
axis([xmin, xmax, ymin, ymax]): change the limits on the plot axes. V = axis will return
the axis limits for the current gure.
xlabel(time (s)): set the label on the x axis of the current graph (and current suplot if ap
plicable) to the given text. Same holds true for ylabel(H (m)). Certain greek symbols can
be used in plots, such as by typing similar commands used in LaTeX. xlabel(\ omega)
for example.
gtext(message text): allows you to place the text given in the argument onto a graph using
the mouse. Helpful for labeling the interesting features of any given plot.
ginput: allows you to click on points in the graph with the mouse. When you hit return you
will leave input mode and the x and y coordinates for the points that you clicked on will
be displayed. To record the values you click type [x, y] = ginput(2); for two values (any
integer number will work).
zoom: toggles the ability to zoom in or out on the current graph. When zooming, clicking on
the left mouse button zooms you in around the point you clicked on; clicking on the right
button zooms you out. zoom on; zoom o.
orient tall: set the orientation of the printed version of the plot to be full page portrait. Other
options are portrait (the default) and landscape.
print -deps lename.eps: send the current gure to the le lename.eps using the current
page orientation. There are other formats you can use listed under help print.
[p,f ] = spectrum(x, 512, 0, [], 200): generate a power spectrum for the data in vector x,
using 512 point FFTs, with no overlap between windows, the default Hanning window and
given that the sampling rate was 200 Hz. In general increasing the size of your FFTs will
increase the resolution of your spectrum (the number of points for which you actually get
information back); it will not increase the maximum frequency for which you get information
back. That is controlled by the Nyquist frequency which is dened as one-half of your sample
rate. Having FFTs longer than the length of x does not make sense and you should make
them small enough such that several windows will actually slide along x so that you will get
good averaging. Its traditional (and faster) to use FFT lengths which are a power of two.
semilogy(f, p(:,1)): plot the power spectrum that you just got from spectrum with a loga
rithmic y axis and a linear x axis. The second argument simply means that you only want
to plot the rst column ((:,1) = all rows, rst column, (1:100,2) would be rst 100 rows,
second column, etc.).
loglog(f, p(:,1)): plot the power spectrum on a log-log scale.
[b,a] = butter(10.0, 50/100): dene lter coecients for a tenth order Butterworth low
pass lter with cut-o frequency of 50 Hz if the Nyquist rate is 100 Hz.
xf = ltlt(b, a, x): lter the data in vector x with the lter dened by coecients b and
a.
13.42:
help plot: get help on the plot command. Very helpful for getting the many details of plot
and all the other commands that are not discussed here.
%: indicates a comment. All text to the right of this marker is ignored until the next line
break.
;: at the end of a line stops the results from scrolling across the screen. Good for math with
big arrays and matrices.
:
mean, median, std, min, max, sqrt, sort, grid, semilogx, semilogy, sin, cos, tan, acos, asin, atan, log,
log10, exp, load, fread, fwrite, fopen, save
c
A.H.
TECHET
ylabel(y(cm) );
title(sin(\theta) );
%
%
%
%
Calculate y = sin(theta).
SAMPLE CODE 2
vec1 = 10 : 10 : 200;
vec2 = [2 : 4 : 20 400 : 2 : 440];
mat1 = [2 : 4 : 20; 400 : 2 : 440];
y1 = vec1 vec2 ;
y2 = vec1. vec2;
size(y1)
size(y2)
length(y2)
%
%
%
%
%
%
%
%
%
%
%
%
Vector 1
Vector 2
b. Sea spectra for developing, fully-developed, and decaying sea states are
shown in Figure 3.
b. = 0.6969
c. 1/ N = 0.5145, 0.6822, 0.7429 for N = 10, 50, 100.
4. a. See Figure 4.
Figure 4 - H ( )
b. See Figure 5.
c. See Figure 6.
Problem #2
13.42 Homework #1
Spring 2005
Out: Thursday, February 3, 2005
Due: Thursday, February 10, 2005
a) 1+ i 3
b)
1+ i 3
3+i
c) -5 - 5i
d) i(1 + i)e ip / 6
b)
z = roe i(q +p )
c) z = roe -i (q -p )
d) z = roei (q + 2p )
Problem 4: Given a cylinder floating horizontally on the free surface and a sphere
floating on the free surface. Both objects are in static equilibrium when submerged to
half diameter depth. Show that for both objects the hydrostatic restoring coefficient in
HEAVE (C 33 ) is approximately equal to C33 rgAwp , where Awp is the area of the object
at the water plane in static equilibrium.
d/2
Sphere,
diameter = d
d/2
Cylinder,
diameter = d,
length = L
Problem 5: Determine whether the following systems are Linear and/or Time- invariant.
Show your work.
t+a
a.
y (t) =
u( s )ds
0
t+a
b.
y (t) =
[u( s )] ds
2
t-a
du( t ) du(t )
dt
dt
c.
y (t ) = a
d.
a y( t ) +b y ( t ) + g y( t ) = u( t )
13.42 Homework #2
Spring 2005
Out: Thursday, February 10, 2005
Due: Thursday, February 17, 2005
Problem 1: There are two LTI systems: system 1 has transfer function H1 (? ), input u1 (t)
and output y1 (t); and system 2 has transfer function H2 ( ? ), input u2 (t) and output y2 (t).
We set the input to system 1 to be a pure cosine:
u1 (t ) = u0 cos(w0t)
And the input to system 2 to be equal to the output of system 1:
u2 (t) = y1 (t)
3 cos( 3w0t)
b) sin( a 0t )
3 sin( a 0t +y )
df
? (Hint: Use partial integration.)
dx
df
d2 f
0 as x , what is the Fourier Transform of
?
dx
dx 2
Problem 4: Answer the following questions. You do not need to provide formal proofs
in each case, but you should make sure that you understand the concepts underlying each
result.
a) Given a stable, linear time invariant (LTI) system with transfer function H(? ),
what is the steady state response (i.e., after the initial transients have died down)
of the system when the input is sinusoidal with frequency ? and amplitude u0 ?
Problem 5: Convolve the following inputs, x(t), with h(t) to find y(t). Show the answer
graphically, and explain your steps for part a) in detail:
a)
b)
c)
Problem 6: Fourier Transforms: Given a graph of x(t), draw the graph of x(w ) :
x(t )
x ( w )
a)
b)
c)
x (t ) =
1
0
if 1
x <
2
if 2
x <
3
if 3
x <
4
otherwise
(Hint: Graph the function and decompose it into a combination of two simpler signals.)
Problem 8: Linear Waves: For designing ocean systems we need to understand when
extreme environmental conditions occur. For linear free-surface gravity waves, the
velocities and dynamic pressure varies harmonically in space and time.
a. Given a free surface elevation, h ( x, t ) = a cos ( kx - wt ) , with wavelength, l = 8 m,
amplitude a = 0.2m, non-dimensionalize and plot the following quantities for a
fixed point in space (i.e. fixed x position @ x= xo and fixed depth z=zn ) but
varying time (the x-axis should be non-dimensional time and y-axis(or axes)
should be the non-dimensional quantities being plotted):
h ( xo , t ) , u ( xo , t ) , w ( xo , t ) , and pd ( xo , t )
Choose the depth such that zn = 2 n l (meters), where n is the integer
corresponding to the month you were born in.
Line up the plots such that you can deduce the relative phase of the variables. You
can plot them on one plot or multiple subplots just make sure to label the
different curves appropriately.
b. Plot the same quantities in part a, but with x varying and at one instant in time (i.e.
@ t=t o ). The x-axis should be non-dimensional position in space and y-axis (or
axes) should be the non-dimensional quantities being plotted.
c. At the depth determined by your birth month, what is the total pressure under the
wave crest, the wave trough, the wave nodal point (the nodal point is the point at
which the wave elevation corresponds to z = 0)?
d. The added mass force on a body is proportional to the fluid acceleration.
Determine when the horizontal added mass force is maximum, minimum, and
zero relative to the wave elevation (i.e. at the crest, the trough, or a nodal point).
e. While we are neglecting viscous forces for the most part in this course they are
still important in certain applications. In general the non-linear, viscous wave
forces are proportional to the square of the fluid velocity. When is the horizontal
viscous wave force maximum under the wave crest, the wave trough or the
wave nodal point? When is the vertical viscous wave force maximum?
-6
x 10
eta*
u*
w*
Pd*
4
3
2
1
0
-1
-2
-3
-4
0
10
15
20
25
t*
-6
x 10
eta*
u*
w*
Pd*
-5
10
12
x*
13.42 Homework #3
Spring 2005
Out: Thursday, February 17, 2005
Due: Thursday, February 24, 2005
Problem 1: Probability Review Problem. Two fair dice are rolled simultaneously with
the following defined events:
A = {Sum of roll is even}
B = {Sum of roll is odd}
C = {Doubles are rolled}
D = {Sum of roll is 8 }
Find the following probabilities:
a) P( A D)
b) P( B C )
c) P(C D)
d) P ( B D)
e) P ( A C )
Problem 2: Probability Review Problem. Find the probability of drawing a full house in
a five-card hand. (Assume that every possible five-card hand drawn from a standard
deck of 52 cards has the same probability of being selected.)
Problem 3: Let random variable X be the total sum of the outcomes of 3 flips of a fair
coin when the value of 1 is assigned to heads and the value of 0 is assigned to tails.
a) Find the probability mass function (i.e. the discrete version of the probability
density function), p X ( x) = P( X = x) .
b) Find the cumulative distribution function, FX ( x) = P( X x) .
c) Using the probability function, find the mean, variance, and standard deviation.
Problem 4: Let the random variable X have a cumulative distribution function FX(x):
0
1
x+ 1
4
4
1
FX ( x) =
4
2
x
if x < 1
if
if
if
if
1 x < 0
1
0 x<
2
1
x <1
2
1 x
1
a) P( x )
2
b) P( x 0)
1
c) P(0 x < )
2
3
d) P( x )
4
e) P( x 1)
Problem 5: Let Y be a random variable with the following probability density function:
2 x
f X ( x) =
0
if 0 x < 1
otherwise
(t ) = A cos(t + )
where and are constants and A is a Gaussian random variable with zero mean.
a) What is the probability density function of A?
b) What is the dynamic pressure, pd(t), under this wave elevation?
c) Determine whether pd(t) is a stationary process.
(t ) = Ai cos(i t + i )
i =1
Where Ai , i , and i are the wave amplitude, frequency, and random phase angle (with
uniform distribution from 0 to 2) of wave component number i.
The amplitudes Ai corresponding to each frequency i may be found from a known wave
energy spectrum, S ( ) , where S ( ) =
1 2
Ai .
2
a) Using the wave energy spectrum given in Figure 1, generate 10 realizations of the
random process (t ) . Let t = [0, 90] seconds. (Hint: Use the MATLAB function
rand to generate a realization of the random variable i .)
b) Compute the ensemble averages (mean and variance at t = 30 seconds, and
compute the temporal averages (mean and variance) for each realization.
Compare the results.
Problem 8: Give short answers to the following questions.
a) What causes the vast majority of sea waves?
b) Why do storm waves have a limiting wavelength?
c) What is significant wave height?
d) Why is significant wave height used so extensively in design practices?
13.42 HW3
SOLUTIONS
SPRING 2005
0:90;
[1 2 3 4];
[sqrt(24) sqrt(36) sqrt(28) sqrt(16)];
j=1:10
i=1:4
phi = 2*pi*rand(1);
%This randomly generates a separate phi value for each of the four
%waves.
WAVE(i,:) = A(i)*cos(w(i)*t+phi);
%This wave equation iterates four times and generates waves for each of
%the frequencies and their corresponding amplitudes.
end
S(j,:) = WAVE(1,:)+WAVE(2,:)+WAVE(3,:)+WAVE(4,:);
%This statement generates my wave elevation as a sum of the four waves
%generated above in each iteration. This loop iterates 10 times, created 10
separate
%realizations of the wave elevation equation with random phase shifts.
%
%
Mean_Ensemble = MEAN(S(:,30));
Variance_Ensemble = VAR(S(:,30));
Mean_Temporal(j,:) = MEAN(S(j,:));
Variance_Temporal(j,:) = VAR(S(j,:));
end
subplot(5,2,1), plot(t,S(1,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,2), plot(t,S(2,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,3), plot(t,S(3,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,4), plot(t,S(4,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,5), plot(t,S(5,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,6), plot(t,S(6,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,7), plot(t,S(7,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,8), plot(t,S(8,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,9), plot(t,S(9,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
subplot(5,2,10), plot(t,S(10,:))
XLABEL('Time (s)')
YLABEL('Wave Elevation (m)')
Mean_Ensemble
Variance_Ensemble
Mean_Temporal
Variance_Temporal
Example Output from Above Code (Note: It changes every run due to random
realizations of phi)
HW37
Mean_Ensemble =
1.7494
Variance_Ensemble =
54.5025
Mean_Temporal =
0.0968
-0.0551
0.1645
0.0659
0.0887
-0.1158
-0.0079
-0.0449
-0.1013
-0.1384
Variance_Temporal =
52.7553
54.5923
51.9517
53.4987
52.7898
53.2102
51.6055
51.3866
52.4914
52.8724
diary off
Comments:
The ensemble statistics are computed over 10 data points, S(:,30), which are
each of the wave elevation realizations at time t = 30 seconds.
The temporal statistics are computed 10 times, for 10 separate wave elevation
realizations, over 91 data points which represent a time range of t = [0:90].
Therefore, the temporal statistics are more consistent from one execution of
the m-file to the next.
Wave elevation, with a uniformly distributed phi over a 2 interval, is a
stationary, ergodic random process. Therefore the ensemble statistics should
equal the temporal statistics. Over multiple m-file executions, there will be
more variability in the ensemble statistics, but they should approximate the
temporal statistics over multiple iterations.
13.42 Homework #4
Spring 2005
Out: Thursday, February 24, 2005
Due: Thursday, March 3, 2005
x 2
f X ( x) =
0
if 0 x 10
otherwise
a) Find .
b) Find P( X > 5) .
c) Find the mean, X .
d) Find the variance and standard deviation, X2 and X respectively.
Let random variable X be the number of days it rains in vicinity of a certain weather buoy.
Suppose, from historical record, that X is normally distributed with a mean of 60 days
and a standard deviation of 15 days.
a) Find the probability distribution function, f X (x) .
b) Find the cumulative distribution function, FX ( x) = P( X x) .
c) What is the probability that next year the number of rainy days will be between 40
and 70?
d) What is the probability that there will be more than 30 rainy days next year?
[Hint: Use either a computer program such as Mathcad or Excel (Analysis ToolPak AddIn) to calculate the erf or use standard normal distribution tables.]
Based on historical weather records, there have been an average of 5 rainstorms per year
over the last 30 years in vicinity of a particular weather buoy. Assuming that the
occurrence of rainstorms is a Poisson process:
( x, t ) = A sin(0t + k0 x)
where 0 is a constant and ( x) = k0 x . (x) is a random variable which is uniformly
distributed from to .
13.42 Homework #5
Spring 2005
Out: Thursday, March 3, 2005
Due: Tuesday, March 8, 2005
Problem 1: Consider the following discrete spectrum of unidirectional seas in deep water.
12
10
S ( ) =
1
3
<
4
4
3
5
if <
4
4
5
7
if <
4
4
7
9
if <
4
4
9
11
<
if
4
4
otherwise
if
f (t ) = f i cos(i t + i )
i =1
Where i is a uniform distributed random variable from 0 to 2. The input function, f(t),
is Gaussian with zero mean.
a) Write the equation of motion of the ship in heave.
b) Find the transfer function of the system.
c) Find the mean of the heave response.
d) Derive an expression for the variance of the heave in terms of the input spectrum,
S f ( ) , i=1,,N.
13.42 Homework #6
Spring 2005
Out: Thursday, March 17, 2005
Due: Thursday, March 31, 2005
Problem 2:
A platform is to be installed in a location where the maximum storm for which
operations are allowed is described by a Bretschneider spectrum with significant wave
height = 5.75 m and peak frequency m = 0.68 rad/s.
The elevation of water near the platform, denoted by p(t) was found to be affected by
the diffracted waves. Note that the transfer function, H(), between the elevation p(t)
and the undisturbed incoming waves at a different location, (t), was found to have
absolute value:
H ( ) =
(0.4 - 3 ) 2 + 0.25
(0.4 - 3 ) 2 + 0.15
Determine the height h at which the deck must be installed above the average water level
in order to have less than two times per hour green water on deck.
S + ( ) =
m
1.25
1.25
e
4
4
m
5
Where the significant wave height is and m is the modal frequency of the spectrum.
1. Plot the Bretschneider spectrum for four significant wave heights between 1 and 3
meters for fully developed seas. (see attached table at end and also posted online
which may be easier to read)
2. Consider one significant wave height chosen in part a.
a. Compare the spectrum in part 1 with the JONSWAP spectrum for the same
significant wave height.
b. Compare the B-S spectrum for developing, fully-developed, decaying sea
states. Note that the spectrum shape can be altered by changing the modal
frequency.
3. Using one case from part a:
a. Calculate the moments of the spectrum using numerical integration (like
trapezoidal integration or similar). Note what happens with the 4th moment
does the integral converge? Compare these values with the approximate
values given in class.
b. Determine the spectrum bandwidth, . (Use the approximate value for M4)
c. Calculate the 1/Nth highest wave height where N = 10, 50, and 100.
Problem 5: A ship is being designed to operate in the conditions you have analyzed
above. The transfer function between the incoming waves and heave motion (heave
transfer function) H() is given by
n2
H ( ) =
2 + 2i n + n2
where n is the structures natural frequency (rad/s) and is the structural damping ratio.
2
a) Plot H ( ) = H ( ) H * ( )
2
Plot this spectrum. Use a structural damping ratio = 0.28 and a n = 1.1rad/s.
c) How will the response change if the structural frequency is different illustrate
your explanation with plots of the response at a higher or lower natural frequency.
Problem 6: Use the relationship between frequency and encounter frequency to explain
how the response of a ship in heave would change from the stationary ship to a ship
moving forward in head seas (head seas are those moving towards the ship bow).
Source: "Wave and Sea Scale for Fully Arisen Sea." Table 2.3, pp. 112-113 in,
Berteaux, Henri O. Coastal and oceanic buoy engineering. Woods Hole, MA: The Author, 1991.
13.42 Homework #7
Spring 2005
Out: Thursday, March 31, 2005
Due: Thursday, April 7, 2005
a) Derive expressions for the heave and sway Froude-Krylov exciting force, taking
into account the presence of the pier.
b) Find the wavelengths of the ambient waves for which the heave and sway FroudeKrylov exciting forces disappear.
13.42 Homework #8
Spring 2005
Out: Thursday, April 14, 2005
Due: Thursday, April 21, 2005
Problem 1: Consider the naval vessel in Figure 1 which is transiting in head seas at a
speed of U = 10 m/s.
Its waterline (intersection of the hull with the calm free surface) is described in Figure 2.
The cross-section of the ship at any given location along the x-axis is a semi-circle of
radius R(x) = B(x)/2. Refer to Figure 3. Assume that the ship is wall-sided above the
waterline.
The non-dimensional added mass and damping coefficients in heave for semi-circular
ship sections are given in Figure 4.
For a ship operating in head seas, the important modes of motion are heave (3) and pitch
(5). Surge is non-zero but generally small for slender ships and in ambient waves of
small steepness.
a. For incident waves of frequency o = 0.5 rad/s, compute the added mass and
damping coefficients Aij and Bij where i, j = 3,5. Assume deep water. It is
suggested that you compute the coefficients at sections spaced 5 meters apart
and that you create a table in Excel for your calculations (this will save you
time in Part f).
B
b. Given that the center of gravity of the ship is located at (0,0,zG) where
zG = -3/4 * R(0), find the restoring coefficients Cij for i, j = 3,5.
c. Determine the magnitudes of the Froude-Krylov heave excitation force |F3|
and pitch excitation moment |F5|, where F3(t)=Re{|F3|eit} and
F5(t)=Re{|F5|ei(t-/2)}. (Hint: Recall that the ambient wave elevation in the
ship-fixed coordinate system is (x,t)=ARe{eikx+it}.)
d. Find the transfer functions for the following linear systems:
e. Given that M55 = 1.5 * 109 kg*m2, find the transfer functions for the
uncoupled heave-pitch equations of motion.
f. Repeat the above (a-e) for o = 0.75 rad/s.
g. Now given the ambient wave spectrum in Figure 5, plot the spectra of heave
and pitch response.
Problem 2: Consider the cylindrical buoy in Figure 6 in ambient plane progressive waves
of amplitude A and frequency . Let the mass of the buoy be m and the center of gravity
be located at z = 43 T . Assume deep water.
Now consider the following cylindrical buoy tethered to the sea floor by a cable
as depicted in Figure 7. Assume that the weight of the buoy is less than the
buoyant force, and that the cable tension is P. The linear restoring matrix due to
the cable is given as:
CCABLE
k11 0
= 0 k33
k
51 0
k15
0
k55
13.42
HW8
Solutions
Spring 2005
Project scenario:
A naval vessel is transiting in head seas at speed, U =
Deep water.
Density, rho =
Gravity, g =
10 m/s
1000 kg/m^3
9.81 m/s^2
Encounter frequency:
omega_e =
0.754842 rad/s
Breadth
m
x*Breadth x^2*Brdth
m^2
m^3
Sectional
Area, S
m^2
omega_e^2*R
a33
over g
over rho*S
a33
x*a33
x^2*a33
0
11290
32201
59199
95819
95819
95819
95819
95819
59199
32201
11290
0
684474
0
282252
644026
887991
958186
479093
0
-479093
-958186
-887991
-644026
-282252
0
0
0
7056312
12880530
13319862
9581858
2395464
0
2395464
9581858
13319862
12880530
7056312
0
90468051
Observed
0
5
10
15
20
20
20
20
20
15
10
5
0
160
0
125
200
225
200
100
0
-100
-200
-225
-200
-125
0
0
0
3125
4000
3375
2000
500
0
500
2000
3375
4000
3125
0
26000
0
9.8
39.3
88.4
157.1
157.1
157.1
157.1
157.1
88.4
39.3
9.8
0
0.000
0.145
0.290
0.436
0.581
0.581
0.581
0.581
0.581
0.436
0.290
0.145
0.000
1.15
0.82
0.67
0.61
0.61
0.61
0.61
0.61
0.67
0.82
1.15
A33 =
B33 =
A35 =
B35 =
A53 =
B53 =
A55 =
B55 =
5m
3422372
3133220
54989379
-34223725
-54989379
34223725
1052981521
1000238628
b33
x*b33
x^2*b33
0
11561
35571
58692
82999
82999
82999
82999
82999
58692
35571
11561
0
626644
0
289015
711422
880385
829992
414996
0
-414996
-829992
-880385
-711422
-289015
0
0
0
7225378
14228436
13205768
8299921
2074980
0
2074980
8299921
13205768
14228436
7225378
0
90068967
Observed
Here , I find the added mass and damping coefficients (3D) for the ship, Aij and Bij for i,j=3,5
The equations to calculate these coefficients were taken from Faltinsen page 56.
Since I have discrete values, I will use the trapezoidal rule as a substitute for integration.
h=
b33
over rho*omega_e*S
800 m^2
1.56
1.2
0.88
0.7
0.7
0.7
0.7
0.7
0.88
1.2
1.56
z_g =
-7.5
C33 =
7848000
C35 =C53 =
0
C55 =
1456709517
Formulas for restoring coefficients were taken from Faltinsen page 58.
To find C55, I need to find GM_L or KB.
To find KB, I will integrate the underwater volume vertically using waterplanes at 1 m intervals.
Based on geometry, half breadth at each waterline, z, is SQRT(10^2 - z^2)
The area of each waterplane is then 2*HalfBreadth(which is a fn of z)*'Effective Length'
The 'Effective Length' is the length of the waterplane if you cut off the aft triangular portion and
flip it over and put it with the forward triangular portion to make an equivalent rectangular Awp.
z
m
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
Midships "Effective
HalfBr (m) length" (m)
10.00
40
9.95
38
9.80
36
9.54
34
9.17
32
8.66
30
8.00
28
7.14
26
6.00
24
4.36
22
0.00
20
Awp(z)
Trapezoidal
m^2
Multiplier
800.00
0.5
756.19
1
705.45
1
648.68
1
586.57
1
519.62
1
448.00
1
371.35
1
288.00
1
191.79
1
0.00
0.5
Sum =
Here, h =
1m
Volume (m^3) = h*Sum F(M) =
M_volume (m^4) = h^2*Sum F(V) =
KB = M_vol/Vol =
z_B =
3.74 m
-3.74 m
F(M)
400.00
756.19
705.45
648.68
586.57
519.62
448.00
371.35
288.00
191.79
0.00
4915.65
4916 m^3
18375 m^4
Lever
Arm
0
1
2
3
4
5
6
7
8
9
10
F(V)
0.00
756.19
1410.91
1946.04
2346.28
2598.08
2688.00
2599.48
2304.00
1726.12
0.00
18375.09
13.42
HW8
Solutions
Spring 2005
Project scenario:
A naval vessel is transiting in head seas at speed, U =
Deep water.
Density, rho =
Gravity, g =
10 m/s
1000 kg/m^3
9.81 m/s^2
Encounter frequency:
omega_e =
1.3233945 rad/s
Breadth
m
x*Breadth x^2*Brdth
m^2
m^3
Sectional
Area, S
m^2
omega_e^2*R
a33
over g
over rho*S
a33
x*a33
Observed
0
5
10
15
20
20
20
20
20
15
10
5
0
160
0
125
200
225
200
100
0
-100
-200
-225
-200
-125
0
0
0
3125
4000
3375
2000
500
0
500
2000
3375
4000
3125
0
26000
0
9.8
39.3
88.4
157.1
157.1
157.1
157.1
157.1
88.4
39.3
9.8
0
0.000
0.446
0.893
1.339
1.785
1.785
1.785
1.785
1.785
1.339
0.893
0.446
0.000
0.67
0.6
0.64
0.69
0.69
0.69
0.69
0.69
0.64
0.6
0.67
A33 =
B33 =
A35 =
B35 =
A53 =
B53 =
A55 =
B55 =
5m
3576507
1353808
7729979
-35765069
-7729979
35765069
602285755
337862269
b33 over
rho*omega_e*S
b33
x*b33
x^2*b33
0
11044
22347
29233
29103
29103
29103
29103
29103
29233
22347
11044
0
270762
0
0
276088 6902210
446938 8938768
438493 6577400
291030 2910296
145515
727574
0
0
-145515
727574
-291030
2910296
-438493
6577400
-446938
8938768
-276088
6902210
0
0
0 52112497
Observed
0
6578
23562
56549
108385
108385
108385
108385
108385
56549
23562
6578
0
715301
0
164443
471239
848230
1083849
541925
0
-541925
-1083849
-848230
-471239
-164443
0
0
Here , I find the added mass and damping coefficients (3D) for the ship, Aij and Bij for i,j=3,5
The equations to calculate these coefficients were taken from Faltinsen page 56.
Since I have discrete values, I will use the trapezoidal rule as a substitute for integration.
h=
x^2*a33
0
4111069
9424778
12723450
10838495
2709624
0
2709624
10838495
12723450
9424778
4111069
0
79614830
0.85
0.43
0.25
0.14
0.14
0.14
0.14
0.14
0.25
0.43
0.85
Note: The restoring coefficients are not frequency dependent, so these are the same as for omega_o=0.5 rad/s.
1b) Find the restoring coefficients, Cij for i,j=3,5
Area of waterplane = Awp =
z_g =
-7.5
C33 =
7848000
C35 =C53 =
0
C55 =
1456709517
800 m^2
Formulas for restoring coefficients were taken from Faltinsen page 58.
To find C55, I need to find GM_L or KB.
To find KB, I will integrate the underwater volume vertically using waterplanes at 1 m intervals.
Based on geometry, half breadth at each waterline, z, is SQRT(10^2 - z^2)
The area of each waterplane is then 2*HalfBreadth(which is a fn of z)*'Effective Length'
The 'Effective Length' is the length of the waterplane if you cut off the aft triangular portion and
flip it over and put it with the forward triangular portion to make an equivalent rectangular Awp.
z
m
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
Midships "Effective
HalfBr (m) length" (m)
10.00
40
9.95
38
9.80
36
9.54
34
9.17
32
8.66
30
8.00
28
7.14
26
6.00
24
4.36
22
0.00
20
Awp(z)
Trapezoidal
m^2
Multiplier
800.00
0.5
756.19
1
705.45
1
648.68
1
586.57
1
519.62
1
448.00
1
371.35
1
288.00
1
191.79
1
0.00
0.5
Sum =
Here, h =
1m
Volume (m^3) = h*Sum F(M) =
M_volume (m^4) = h^2*Sum F(V) =
KB = M_vol/Vol =
z_B =
3.74 m
-3.74 m
F(M)
400.00
756.19
705.45
648.68
586.57
519.62
448.00
371.35
288.00
191.79
0.00
4915.65
4916 m^3
18375 m^4
Lever
Arm
F(V)
0
1
2
3
4
5
6
7
8
9
10
0.00
756.19
1410.91
1946.04
2346.28
2598.08
2688.00
2599.48
2304.00
1726.12
0.00
18375.09
13.42 Homework #9
Spring 2005
Out: Thursday, April 21, 2005
Due: Thursday, April 28, 2005
Problem 1: From Problem 1 of Homework #8, consider the naval vessel operating in the
given ambient wave spectrum. Then consider the following operational limits:
Point-to-point transit operational limits
~ RMS pitch angle = 2.5 degrees
~ RMS vertical acceleration at the bridge (point B in Figure 1 of HW8) = 0.2g
Helicopter flight operations (helo flight ops)
~ RMS vertical velocity at the flight deck (point A in Figure 1 of HW8) = 1.0 m/s
a) Does the present design satisfy the point-to-point transit operational limits? The
x-coordinate of the bridge is xB = 12 m.
b) Does the present design satisfy the helo flight ops limit? The x-coordinate of the
flight deck is xA = -22 m.
Problem 3: A rigid circular cylinder with diameter, d = 1.8 m, is rigidly mounted at the
seafloor as shown in Figure 2. The water depth in this region is 50 meters and waves can
be considered deep water waves. Monochromatic, linear waves are incident on the
cylinder from one direction. The waves have frequency = 2.0 rad/s and amplitude a =
1.1 m. Assume that there is no current in this location in addition to the waves and we can
neglect effects due to the cylinder piercing the free surface.
d. Calculate the moment acting on the cylinder about the seafloor mount.
e. Find the location of the center of force along the height cylinder (relative
to the seafloor).
Laboratory #1:
Wave Spectra and ship response in waves
Date issued: 31 March 2005
Date Due: 14 April 2005
1.1
INTRODUCTION
Objectives: The object of this lab is to measure the response of a model ship due to
incident waves. Students will analyze the response in heave and pitch motions of a 100:1
scale USS Arleigh Burke class DDG (destroyer, guided missile) model in both
monochromatic and multi-frequency wave trains.
Lab setup: This lab is performed in the Ocean Engineering towing tank. The DDG-51
model is mounted to a free moving linear carriage, and rests in the water at its design
draft. In the first part of the lab waves are generated by wave paddle at one end of the
tank and the model is fixed. In the second part of the lab, the model is towed at a forward
speed in head seas such that the effects of encounter frequency can be observed.
Ship Model: The model is a 100:1 scale USS Arleigh Burke class DDG (destroyer,
guided missile). The primary characteristics of the DDG 51 follow. The model tested
will be without a sonar dome.
LPP
T
B
CP
CX
Cb
Cwp
Sws
466.0 ft
20.69 ft
58.98 ft
0.625
0.830
0.519
0.789
8240 LTsw
29,754 ft2
Length
Draft
Beam
Displacement
Wetted surface area
Data format: There will be four general measurements made in this lab. Two wave
probe outputs in Volts, and the heave and pitch response (also in Volts). The data is
collected by the Dasylab program, through a data acquisition PC card. The format of
each file is in ASCII format. This format is compatible with both Matlab and Excel.
Each run will have to be named appropriately and notes should be taken by each group to
ensure that each run is recorded properly. The TA will transfer the data from Dasylab to
a common directory, which can be accessed by all the students in the class.
Post Processing: Post processing of this lab is made easier by using several Matlab
scripts. The MATLAB scripts are provided in a ZIP archive on the 2.22 Labs page of
MIT OCW.
You will also need the MATLAB Signal Processing Toolbox. The data can be processed
as follows:
1- Calibrate the wave probes using parsedasci.m to find the mean value of each
wave probe height. Then put this data into a matrix and find the slope.
2- Run the matlab script analyze_response.m for each run. It will ask for a
filename corresponding to your datafile (e.g. filename.asc). Type the filename
and press enter. The data will be plotted on the screen. You can zoom in or out
(type help zoom or click on the magnifying glass on the plot window) so that you
can see the responses. When you hit the return key you will then be able to click
with the mouse on amplitudes or wavelengths. INSTRUCTIONS on where to
click are printed as the title of the plot! Click in appropriate places, and then the
program will output information you will need to write down or put in some sort
of array.
3- Run response_plot.m. Response_plot.m requires an array be input at the
beginning of the file the data each student collected from analyze_response.m
should be inserted here.
4- Run bret2.m it will plot the Bret Schneider spectrum from your recorded data,
and the heave and pitch responses to these waves.
5- Running bret_compared.m to see the comparison between the actual heave and
pitch responses and the heave and pitch predictions made with the transfer
function created by looking at the individual monochromatic waves. The heave,
wave, and pitch amplitude arrays will have to be re-entered in the beginning of
the .m file for this function to work properly.
1.3
Treat your project write-up like a formal lab report. Type the final report and embed
all figures into the text as appropriate. Final lab reports will be approximately 10-20
pages on average.
Provide an introduction, an equipment diagram and a narrative of what you actually did
during the laboratory and to process the data afterwards, including the calibration (even
though the calibration data was given to us before the lab it is a good idea to ask how it
was done!). Ensure you address any questions raised in this assignment. Plots should be
well integrated and referenced in your discussion of the data. You may work in groups in
gathering the experimental data for this project but your analysis and discussion must
be your own.
Specific questions to answer in the laboratory write up are listed below. Figures (plots,
tables, sketches, etc) will be very useful in helping to answer each question. Figures
should be embedded with the text and plots should have appropriate axis labels and units
where appropriate.
1- Describe and discuss the heave and pitch frequency response of the ship in
monochromatic waves?
2- Do you observe any cancellation frequencies in heave and pitch? If so at what
frequencies to the motions die off? Does this make sense given the length of the
ship and the characteristics of the waves at that frequency?
3- Do you think that the response curves in heave and pitch made from the single
frequency waves are accurate? Where might sources of error come from?
4- How does the ship response in the Bretschneider wave train compare to the
reponse plot for monochromatic waves?
5- Are there any sea states or wave frequencies (consider only head seas) this ship
should try to avoid operating in? If so please elaborate.
6- Consider the cancellation frequency for the model ship, do you anticipate this will
change for the full scale ship?
7- How does moving the model with a forward speed affect the heave and pitch
response?
8- Throughout your discussions consider the error sources in this lab and how they
might affect your data.
Lab 1
0.4
0.8
1.2
1.6
2
2.4
Lab 2
0.5
0.9
1.3
1.7
2.1
2.5
Lab 3
0.6
1
1.4
1.8
2.2
2.6
Lab 4
0.7
1.1
1.5
1.9
2.3
2.7
Underwater Vehicles
Surface Vessels
Offshore Platform
Basic Recipe
Fundamental Math &v Science
v
Newtons 2nd Law, F = Ma
Conservation of Angular Momentum
Basic Fluid Mechanics
Idealized System
Mathematical Model
M&x&(t ) + Bx& (t ) + Cx (t ) = F (t )
Behavior Prediction
Courtesy of JPL.
Viscous damping
Ship Motions
z
6
1 = surge
2 = sway
3 = heave
4 = roll
5 = pitch
6 = yaw
13.42 Organization
Instructor:
Professor Alexandra Techet
Grading
50% Exams
15% Homework
Homeworks
13.42 Lecture:
Ocean Waves
Spring 2005
Alexandra H. Techet
MIT Ocean Engineering
Ocean Waves
Sea State
Code
0
1
2
3
4
5
6
7
8
9
Description
Calm (glassy)
Calm (rippled)
Smooth (mini-waves)
Slight
Moderate
Rough
Very Rough
High
Very High
Huge
Courtesy of JPL.
Courtesy of JPL.
Courtesy of JPL.
U wind C p = / k = g /
Limiting frequency:
c g / U wind
a(t ) = e
where
= 2 k 2 = 2 4 / g 2
(From Landau and Lifshitz)
Typical Spectrum
Fetch limitations
State of development or decay
Seafloor topography
Local Currents
Effect of distant storms (swells)
Wave Spectra
Many spectra are strictly valid for FULLY
DEVELOPED SEAS.
Developing seas have a broader spectral peak.
Decaying seas have a narrower peak.
Pierson-Moskowitz Spectrum
Developed by offshore industry for fully developed seas in
the North Atlantic generated by local winds. One parameter
spectrum.
Mathematical form of S+() in terms of the significant
wave height, H1/3. (H1/3=)
Spectrum Assumptions
Deep water
North Atlantic data
Unlimited fetch
Uni-directional seas
No swell
Bretschneider Spectrum
Replaced P-M spectrum since need for fully developed seas
is too restrictive. Two parameter spectrum.
2
Modal frequency
10
Modal Frequency
8
6
H13 = 8 m
4
6m
2
0
4m
2m
0
0.5
1.0
1.5
Frequency (radians/second)
2.0
S+() [metres2/(radian/second)]
20 sec
3
15 sec
2
T0 = 10 sec
1
5 sec
0.5
1.0
1.5
2.0
Frequency (radians/second)
Bretschneider wave energy spectra; characteristic wave height 4 metres.
Figure by MIT OCW. After Faltinsen (1993).
JONSWAP Spectrum
JONSWAP spectrum was developed for the limited fetch
North Sea by the offshore industry and is used extensively.
6
T0 = 10 sec
4
JONSWAP
2
Bretschneider
0.4
0.8
1.2
1.6
2.0
Frequency (radians/second)
JONSWAP & Bretschneider spectra; significant wave height 4 metres.
Amplitude spectrum
Figure by MIT OCW. After Faltinsen (1993).
10
.008
JONSWAP
T0 = 10 sec
.006
Bretschneider
.004
.002
1.0
2.0
3.0
4.0
Frequency (radians/second)
Wave slope spectra: significant wave height 4 meters.
Figure by MIT OCW. After Faltinsen (1993).
Ochi Spectrum
Ochi spectrum is an extension of the BS spectrum,
allowing to make it wider, small, for developing seas, or
narrower, larger, for swell. Three parameter spectrum.
11
determines the width of the spectrum
(x) = the gamma function of x
11
storm
swell
Directionality in waves
In reality, waves are three-dimensional in
nature and different components travel in
different directions.
Measurements of waves are difficult and
thus spectra are made for uni-directional
waves and corrected for threedimensionality.
12
Correction to uni-directionality
/2
/2
13
Storm Statistics
For each storm (i) we use the significant
wave height and average period to construct
a spectrum and then find the short term
statistics.
For structural analysis the failure level is a
large quantity compared to the rms value, so
we use the rate of exceeding some level ao.
14
H1/3
= 1.06 Hv (meters)
T = 1.12 Tv (seconds)
Tz= 0.73 Tv (seconds)
Nordenstrom (1969)
15
13.42 Lecture:
Vortex Induced Vibrations
Prof. A. H. Techet
21 April 2005
Offshore Platform
Fixed Rigs
Spar Platforms
VIV Catastrophe
I n another city, the John Hancock tower wouldn't be anything special -- just another
reflective glass box in the crowd. But because of the way Boston and the rest of New
England has grown up architecturally, this "70's modern" building stands out from the
rest. Instead of being colonial, it breaks new ground. Instead of being quaint, it soars
and imposes itself on the skyline. And Instead of being white like so many buildings in
the region, this one defies the local conventional wisdom and goes for black. For these
reasons and more the people of Boston have fallen in love with the 790-foot monster
looming as the tallest building in New England at the time of its completion. In the
mid-1990's, The Boston Globe polled local architects who rated it the city's third best
architectural structure. Much like Boston's well-loved baseball team, the building has
had a rough past, but still perseveres, coming back stronger to win the hearts of its
fans. The trouble began early on. During construction of the foundation the sides of the
pit collapsed, nearly sucking Trinity Church into the hole. Then in late January, 1973
construction was still underway when a winter storm rolled into town and a 500-pound
window leapt from the tower and smashed itself to bits on the ground below. Another
followed. Then another. Within a few weeks, more than 65 of the building's 10,344
panes of glass committed suicide, their crystalline essence piling up in a roped-off area
surrounding the building. The people of Bean Town have always been willing to kick a
Image removed for copyright reasons.
brother when he's down, and started calling the tower the Plywood Palace because of
the black-painted pieces of wood covering more than an acre of its faade. Some
people thought the building was swaying too much in the wind, and causing the
windows to pop out. Some thought the foundation had shifted and it was putting stress
of the structural geometry. It turns out the culprit was nothing more than the lead
solder running along the window frame. It was too stiff to deal with the kind of
vibrations that happen every day in thousands of office buildings around the world. So
when John Hancock Tower swayed with the wind, or sighed with the temperature, the
windows didn't and eventually cracked and plummeted to Earth. It cost $7,000,000.00
to replace all of those panes of glass. The good news is, you can own a genuine piece
of the skyscraper. According to the Globe, the undamaged sheets were sold off for use
as tabletops, so start combing those garage sales. For any other skyscraper, the
hardship would end there. But the Hancock building continued to suffer indignities.
The last, and most ominous, was revealed by Bruno Thurlimann, a Swiss engineer who
determined that the building's natural sway period was dangerously close to the period
of its torsion. The result was that instead of swaying back-and-forth like a in the wind
like a metronome, it bent in the middle, like a cobra. The solution was putting a pair of
300-ton tuned mass dampeners on the 58-th floor. The same engineer also determined
that while the $3,000,000.00 mass dampeners would keep the building from twisting
itself apart, the force of the wind could still knock it over. So 1,500 tons of steel braces
were used to stiffen the tower and the Hancock building's final architectural indignity
was surmounted.
Courtesy of Artefaqs Corporation. Used with permission.
Source: Glass Steel and Stone, http://www.glasssteelandstone.com.
Reprinted from http://www.glasssteelandstone.com/BuildingDetail/399.php
Potential Flow
U() = 2U sin
P() = 1/2 U()2 = P + 1/2 U2
Cp = {P() - P }/{1/2 U2}= 1 - 4sin2
Base
pressure
(i)
(ii)
i) Potential flow:
-/w < < /2
ii) P ~ PB
/2 3/2
(for LAMINAR flow)
Wake Instability
vortex roll-up
Rd < 5
St=fsd/U
fs is the shedding frequency, d is diameter and U inflow speed
Reynolds Number
UD inertial effects
Re =
v
viscous effects
subcritical (Re<105) (laminar boundary)
Reduced Velocity
U
Vrn =
f
n D
Number
0.47
0.4
Smooth Surface
0.3
0.2
0.1
0
40
Rough
Surface
St = 0.2
102
103
104
105
106
107
Due to unsteady flow, forces, X(t) and Y(t), vary with time.
Force coefficients:
Cx =
D(t)
1/
2
U2
Cy =
L(t)
1/
2
U2 d
DRAG
Cx
Avg. Drag 0
LIFT
Cy
Avg. Lift = 0
structural vibrations
z (t ) = zo cos t
z& (t ) = zo sin t
&&
z (t ) = zo 2 cos t
Lock-in
Shedding
frequency
Natural frequency
of oscillation
v = 2 fv = 2 St (U/d)
n = m +k m
a
to Vortex shedding
z(t)
L(t ) = La &&
z (t ) + Lv z& (t )
(m + La ) &&
z (t ) + (b Lv )z& (t ) + k{
z (t ) = 0
14243 14243
Added mass term
Restoring force
Damping
If Lv > b system is
UNSTABLE
the recorded lift data will give the sinusoidal term which can
LIFT FORCE: L(
t) = Lo cos(t + o )
if <
v
L(t ) =
Lo cos o
Lo sin o
&&
z
t
(
)
z& (t )
+
2
zo
zo
M a ( , a) =
Total lift:
Lo
2 cos o
a
Lo
sin o
a
L(t ) = M a ( , a) &&
z (t ) + Lv ( , a) z& (t)
(a = zo is cylinder heave amplitude)
Total Force:
L(t ) = M a ( , a) &&
z (t ) + Lv ( , a) z& (t)
(
+ (
d
) Cma ( , a) &&z (t )
dU ) CLv ( , a) z& (t )
2
Velocity
Vortex Induced Vibrations are
SELF LIMITED
Gopalkrishnan (1993)
Amplitude Estimation
Blevins (1990)
a/ =~ 1.29/[1+0.43 S ]
3.35
G
d
_
_
^ 2 2m (2) ; f^ = f /f ; m
= m + ma*
SG=2 fn
n
n s
d2
b
=
2 k(m+ma*)
ma* = V Cma; where Cma = 1.0
Drag Amplification
a
= 0.75
d
2
1
0.1
0.2
0.3
fd
Fluctuating Drag:
Mean drag:
Cd = 1.2 + 1.1(a/d)
1.0
a)
One-tenth highest
transverse
oscillation amplitude
ratio
b)
Mean drag
coefficient
c)
Fluctuating drag
coefficient
d)
Ratio of transverse
oscillation frequency
to natural frequency
of cylinder
1.0
Flexible Cylinders
Reading #1
1. Dynamical Systems
Dynamical systems are representations of physical objects or behaviors such that the
output of the system depends on present and past values of the input to the system. For
example:
y(t ) = u 3 (t 1 )dt1
t
t -3
In order to model dynamical systems we need to build a set of tools and guidelines that
can be used to analyze systems such as a ship in waves. This section will introduce tools
for analyzing linear systems.
System:
Modeling:
Inputs:
Outputs:
Reading #1
Replace u (t ) by u (t +t ) (Change of
variables)
Replace y(t ) by y(t +t ) (Replace all
occurrences of t with t + t )
Step 1:
Step 2:
Step 3:
To illustrate this procedure we can use a few simple examples of basic systems with
input, x(t ) , and output, y(t ) .
Example 1: y(t ) = [u( t )]3 / 4 System is clearly time invariant: y ( t + t ) = [u (t +t ) ]
3/ 4
Example 2: y(t ) =
Step (1): Plug in t1 + t for t1 on the RHS and perform a change of variables (let
z = t1 +t ). Note that the limits of integration must also shift with this change of
variables.
u( t1 + t)dt1 =
t+t
u(z )dz
Step (2): Plug in t + t for t on the LHS. Notice that the limits of integration do not
change in the same fashion as in step 1. The original integral on the RHS is bounded
from zero to t , and since we are simply replacing all occurrences of t with t + t we do
not shift the limits of integration as we did in step 1.
y(t + t ) =
t +t
u(t1 )dt1
Step (3): Compare results from steps (1) and (2). They are not equal, therefore this
system is not time invariant.
t+t
u(z )dz
t+t
u(t1 )dt1
Reading #1
t -5
Step (1): Plug in t1 + t for t1 on the RHS and perform a change of variables
(let z = t1 +t ):
t-5
u 4 (t1 +t )dt1 =
t +t
t - 5 +t
u 4 (z )d z
Step (2): Plug in t + t for t on the LHS, and again, note the shift in integration limits:
y(t + t) =
t +t
u 4 (t1 )dt1
t-5 +t
Step (3): Compare steps (1) and (2). They are equivalent, therefore system is time
invariant!
t+t
t-5 +t
u 4 (z )dz =
t+t
t-5 +t
u4 (t1 )dt1
In general, given a linear system L , as shown in figure 1, and some input, x1 (t ) , the
system would result in an output, y1 (t ) , conversely some other input, x2 (t ) , into the
same system would simply yield the output, y2 (t ) , such that the inputs and outputs
obey the following properties:
Reading #1
Linear Superposition:
x1 (t) + x2 (t ) y1 (t) + y2 (t )
Scaling:
a1 x1 (t ) + a2 x2 (t) a1 y1 (t ) + a2 y2 (t )
A system must satisfy both the superposition and the scaling criteria for it to be
considered linear.
Example 1: y(t ) = C
du
dt
Example 2: y(t ) = u (t1 )dt1 . This system is linear. (But it is not time invariant!)
t
Example 3: y(t ) = au3 (t ) . This system is not linear. (But it is time invariant!)
Reading #1
-e/ 2 and
(1)
Figure 3. A function
Reading #1
f (t ) =
(2)
f (t )uo (t -t )dt
(3)
f (t )uo ( t - a)dt = f ( a )
(4)
uo (t )dt = 1
Lets take a closer look at equation (4) from above. Here the value of the constant a is
set to zero and we see that the integral simply equals that function f(t) evaluated at t=0 .
+e /2
-e /2
f ( t) uo (t )dt
= lim
e 0 f (0)
+ e /2
- e/ 2
u o (t ) dt
= f (0)
moment.
For now lets just look at a simple continuous time LTI system with a impulse
input, uo (t ) , shown in figure 4. The output corresponding to the impulse input is the
impulse response, h (t ) .
Reading #1
1.6. Convolution
Given a continuous -time LTI system characterized by an unique impulse response,
h (t ) , the response of this system to some input, x(t ) , at time t = t is simply the input
weighted by the time-shifted impulse response: x(t )h (t - t ) .
Therefore, in order to determine the output of the system, y(t ) , to an input, x(t ) , we can
integrate all possible outputs (responses), x(t )h (t - t ) , in the time interval from minus
infinity to positive infinity:
y(t ) =
(5)
Thus for any continuous time LTI system, the output y(t ) is a weighted integral of the
input, x(t ) , where the weight on x(t ) is h (t -t ) , the time shifted unit impulse response.
The integral in equation 5 is the convolution integral, which, through a change of
variables, can also be written as
y(t ) =
(6)
(7)
Reading #1
1.7. Causality
A causal system responds only after being excited (i.e if the input x (t ) is zero before to
therefore the output is also zero before to ). In reality all physical systems are causal.
Thus the response, y (t ) , to the input is zero before time t = 0 and we can rewrite the
convolution integral with integration limited to the interval [ 0 < t <+ ]:
y(t ) =
(8)
Since we are considering dynamical systems that depend only on past and present
inputs, and that cannot see into the future, the response is also bounded by the current
time, t:
(9)
Reading #1
This system has a lumped mass, m , that moves some distance, x ( t ) , as a function of
time. The mass experiences a spring restoring force, f s = -kx ( t ) , proportional to the
spring constant times the distance the mass moves, and a damping term, f d = -bx ( t ) ,
proportional to the damping coefficient times velocity of the mass, and an external
applied body force, f (t ) . A simple free body diagram helps illustrate that the sum of
the spring, damping, and applied forces must, by Newtons second law, equal the
system mass times the acceleration of the object:
body
Reordering
terms
we
arrive
= -bx ( t ) - kx ( t ) + f ( t ) = mx (t )
at
the
classic
differential
(10)
equation
for
mass-spring-dashpot system.
mx (t ) + bx ( t ) + kx (t ) = f ( t )
(11)
In order to evaluate the system appropriately, we can use the following steps:
1. First, we need to identify the initial conditions. In this case, we assume our
system starts at rest, such that the position and velocity of the mass are zero:
x(0) = 0
x (0) = 0
(12)
e/2
-e/2
-e /2
{mx + bx + kx} dt = { f (t )} dt = 1
(13)
Since e/2 is an infinitesima lly small time interval, before time zero we can
write t = -e /2 as t = 0 - . Following the same logic we can also write t = +e/2
as t = 0 + . Considering the integral in equation 13 and the initial conditions on
Reading #1
(14)
The term x (0 - ) is zero since there is no motion before time zero and we are left
with the velocity just after the force is applied:
x (0+ ) =
1
m
(15)
mx + bx + kx = 0
(16)
x(0) = 0
(17)
x (0 + ) =
1
m
(18)
(19)
We can find the constants using the original system equation such that
ms 2 + bs + k = 0
+
-
s1, 2 = - 2bm
let d =
b
2m
and
b2 -4km
2m
w=
k
m
b
- 4m
2
s1, 2 = -d iw .
Let us assume b 2 < 4km ( z =
b
< 1 ), then
2 km
x(0) = C1 + C2 = 0 C1 = -C2
x (0) = C1s1 +C 2 s2 = C1 (s1 - s2 ) =
1
m
s1 - s2 = d + iw d - (d - iwd )
therefore
1
C1, 2 = 2imw
d
10
Reading #1
Thus we can formulate the system response due to the impulsive force input as
x (t ) =
{ (
1
e-d t eiw d t - e-iw d t
2 i mwd
sin (wd t ) =
Since
)}
eiwd t - e -iwd t
,
2i
1
mwd
e-d t sin wd t;
t 0
t <0
0;
3. Useful References
There are several good texts on signals and systems that give a thorough discussion of
Linear Time Invariant systems and their properties. A few suggestions are listed below.
http://www.engin.brown.edu/courses/en4
___________________________
Vibrations.
A.V. Oppenhein, A. S. Willsky, S.H. Nawab (1997) Signals and Systems, 2nd
ed. Prentice Hall Signal Processing Series, New Jersey. (6.003 Course text
book)
Triantafyllou and Chryssostomidis, (1980) "Environment Description, Force
Prediction and Statistics for Design Applications in Ocean Engineering"
11
Reading #2
1. Fourier Series
Fourier series are very useful in analyzing complex systems with periodic inputs as
they can be used to represent a periodic signal as a summation of scaled sines and
cosines :
(1)
n =1
where wo = 2p/T is considered the fundamental frequency and the coefficients are
written as
1 T
f (t ) dt
T 0
(2)
An =
2 T
f (t) cos(nwo t )dt
T 0
(3)
Bn =
2 T
f ()sin
t
( nw o t )dt
T 0
(4)
Ao =
Reading #2
The Fourier series can be written more compactly using complex notation
( eiwt = cos wt + i sinwt ).
f (t ) =
Ce
inwo t
(5)
n =-
C n =
1 T
f (t )e-inwo t dt
0
T
(6)
We can use Fourier series to represent a periodic, absolutely integrable function, f(t).
| f (t ) | dt <
(7)
2. Fourier Transform
The Fourier transform (FT) converts a function of time into a function of frequency.
The inverse Fourier transform (IFT) reverts the function in the frequency domain back
to the time domain. We will assume that f (t ) is absolutely integrable.
The Fourier Transform of f (t ) is f (w ) such that
+
f (w ) = f (t )e -iwt dt
-
(8)
f (t ) =
1
2p
f (w )ei wt dw
(9)
Reading #2
Example 1: Let
1;
x (t ) =
0;
t T1
t > T1
(10)
x (w ) =
x (t )e -iw t dt =
x (w ) =
T1
(1)e -iw t dt =
-T1
2sin wT1
w
2sin wT1
, - < w <+
w
(11)
(12)
x(t ) =
x(t ) =
1
2p
1
2p
w )eiwt dw
x(
(13)
2sin wT1 iw t
e dw
w
(14)
t T1
t > T1
(15)
w W1
w > W1
(16)
1
2p
x(
w )e iwt dw =
x(t ) =
1
2p
+W1
-W1
(1)e iwt dw =
sin W1t
, - < t <+
pt
sin W1t
pt
(17)
(18)
Reading #2
Notice the similarity between the two functions in examples 1 and 2 specifically
equations (10) and (16), and also equations (12) and (18).
Parsevals theorem
explains that there exists a dual pair of functions with time and frequency
interchanged i.e. a symmetric pair of functions.
a.
b.
c.
d.
Figure 2. Symmetric Functions: (a.) Function of time: x(t); (b.) Fourier transform of x(t) in the
frequency domain. (c.) Function of frequency:
time domain.
Reading #2
y(t) =
x( t )h (t -t )dt ,
(19)
y (w ) = FT
x(t ) h (t -t ) dt
-iwt
(20)
dt
(21)
We can then rewrite the exponential e- iwt as e -iw (t+t -t ) = e -iw ( t-t ) e iwt without changing
our equation. Then let t1 = t -t and dt1 = dt , such that equation(21 ) becomes
w) =
y(
(22)
Reordering the terms within the integrals we see that we have two separable integrals
such that
+
h (w )
(23)
x (w )
Note the form of the two separate integrals on the RHS of equation (42) they are the
Fourier transform of the input and the impulse response. Now, the Fourier transform
out the system output is simply the multiplication of the Fourier transform of the input
and impulse response:
Reading #2
y (w ) = h (w ) x( w ) .
(24)
Where h(w ) is the Fourier transform of the impulse response and is referred to as the
TRANSFER FUNCTION, commonly written as H(w ) .
w ) = H (w ) x( w)
y(
(25)
Convolution: y (t ) = h (t ) * x (t )
Multiplication: y (w ) = H (w ) x( w)
1.
2.
3.
For a given a harmonic input u (t ) and transfer function H (w ) we can easily write the
output of our LTI system in terms of the amplitude and freque ncy of the input and the
amplitude and argument (phase) of the transfer function.
x (t ) = xo cos(w ot + y )
(26)
y (t ) = yo cos(w ot +y + y 1 )
(27)
where yo = xo | H (wo ) | is the amplitude of the response and y 1 = arg{H (wo )} is the
phase shift of the response from the input.
Reading #2
H (w ) =| H (w ) | eiH (w )
(28)
(29)
y (t )
= Re
Since we are only interested in the real part of y(t ) (the input was cosine) we have
(30)
This process can be extended in the same fashion for an input of sine. However
instead of the real component we look at the imaginary component of
Reading #2
6. Useful References
There are several good texts on signals and systems that give a thorough discussion of
Linear Time Invariant systems and their properties. A few suggestions are listed
below.
A.V. Oppenhein, A. S. Willsky, S.H. Nawab (1997) Signals and Systems, 2nd ed.
Prentice Hall Signal Processing Series, New Jersey. (6.003 Course text book)
Triantafyllou and Chryssostomidis, (1980) "Environment Description, Force
Prediction and Statistics for Design Applications in Ocean Engineering" Course
Supplement.
Reading #
Simple events are those which do not share any common area within an event space, i.e.
they are non-overlapping, whereas composite events overlap (see figure 1). The
probability that an event will be in the event space is one: p( S ) = 1.
A2
A1
A3
A4
A1
A3
Simple Events Ai
1.
A2
A4
Composite Events Ai
Reading #
intersection; A
union; A
A'
not A; A'
2. Union, Intersection, and Complement.
Reading #
EXAMPLE: One store carries six kinds of cookies. Three kinds are made by Nabisco and
three by Keebler. The cookies made by Nabisco are not made by Keebler. Observe the next
person who comes into the store to buy cookies. They choose one bag. It can only be made
by either Nabisco OR Keebler thus the probability that they choose one made by either
company is zero. These events are mutually exclusive.
p( A B) = 0 Mutually Exclusive
(1)
(1) p ( A) 0
(2) p( S ) = 1 (all events)
(3) If A1 , A2 , A3 ,L, An are a collection of
mutually exclusive events then:
p ( A1 A2 A3 L An ) = i=1 p ( Ai )
n
(2)
(3)
This holds since the probability of the event space, S , is exactly one.
Reading #
p( A B) = p( A) + p( B) p( A B)
(4)
p( A B) = p ( A) + p( B A)
(5)
becomes
since B is simply the union of the part of B in A with the part of B not in A:
B = (B A) (B A).
(6)
These two parts are Mutually Exclusive thus we can sum their probabilities to get the
probability of B. So
p( B) = p ( B A) + p( B A)
(7)
p( A B) = p( A) + p ( B) p( A B).
(8)
Example 1:
Toss a fair coin. Event A = heads and event B = tails . p( A) = 0.5; p( B) = 0.5
Example 2:
If A, B, and C are the only three events in S and are mutually exclusive events, where
Reading #
Example 3:
2. Conditional Probability
Conditional probability is defined as the probability that a certain event will occur given
that a composite event has also occurred. We write this conditional probability as p( A | B)
and say "probability of A given B".
Given that a composite event, M (see figure 3), has happened what is the probability that
event Ai also happened?
By stating that event M has happened we then have excluded all events that do not
overlap with M as possible outcomes. The implication is that now the event space has
shrunk from S to M . Therefore we must redefine the probabilities of the events such
that p( M ) = 1 and all other events have p ( Ai ) = 0 if M Ai = 0 but if M Ai 0 (i.e. if
Ai has some overlap with M ) then 0 p ( Ai ) 1 . The greater the overlap, the higher the
probability of the event.
Reading #
A2
A1
A3
A4
M (composite event)
3. Composite Event
M.
Thus for any two events A and B with p ( B) > 0 the CONDITIONAL PROBABILITY of
A given B has occurred is defined as:
p( A | B) =
p( A B)
p( B)
(9)
p( A B) = p( A | B) p( B)
(10)
and is commonly referred to as the Multiplication Rule and is often an easier form of
equation 9.
Example 1:
A gas station is trying to determine what the average customer needs from their station.
The have determined the probability that a customer will check only his/her oil level or
only his/her tire pressure and also the probability they will check both.
Reading #
Their next step is to determine the probability that a person checks their oil given they also
checked their tire pressure.
(1) Choose a random customer and find the probability that a customer has checked his
tires given he/she checked the oil:
p(T | L) =
p(T L) 0.01
=
= 0.1
p( L)
0.1
(11)
(2) Choose a random customer and find the probability that a customer has checked his oil
given he/she checked the tires:
p( L | T ) =
p ( L T ) 0.01
=
= 0.05
p(T )
0.2
(12)
Example 2: What is the probability that the outcome of a roll of a dice is 2 ( A2 ) given that
Let the complex event M be the occurrence of all possible even numbers. Since a die is six
sided with three possible even numbers, 2,4, and 6, the probability that an even number
will be rolled is 0.5:
M = A2 A4 A6
p ( M ) = 1/ 2
(13)
Reading #
since there are three possible even numbers and a 50% chance of rolling an even number.
Thus the probability that a person will roll a two given that an even number is rolled is
p( A2 | M ) =
p( M A2 1/ 6
=
= 1/ 3 = 33%
1/ 2
p( M )
(14)
(15)
or
n
p( B) = p( B | Ai ) p ( Ai ).
(16)
i=1
This is the Law of Total Probability. We can prove this simply by looking at the
conditional probability of each of the events.
Proof: Since the events, Ai , are mutually exclusive and exhaustive, in order for event B to
occur it must exist in conjunction with exactly one of the events Ai .
B = ( A1 and B ) or ( A2 and B) or L or ( An and B) = ( A1 B) L ( An B)
p ( B ) = i=1 p ( Ai B ) = i=1 p ( B | Ai ) p ( Ai )
n
Reading #
p( Ak | B) =
p( Ak B)
p( B)
(17)
using the multiplication rule and the Law to total Probability we get
BAYES THEOREM
p( Ak | B) =
p ( B| Ak ) p ( Ak )
i=1
n
p ( B| Ai ) p ( Ai )
; for k = 1, 2, 3,..., n
4. Examples
Lets look at a few complex examples.
Youve been hired by a salvage company to determine in which of two regions the
company should look to find a ship that sank in a hurricane with treasure worth billions.
This is very exciting, so you pull out your probability book and get to work.
You know that group X is using a side scan sonar that has a success rate of finding objects
equal to 70% (probability that it finds an object is 0.7) and that group Y has equipment that
has a success rate of 60%. You also know that meteorologists predict that there is an 80%
probability that sunken ship and her treasure lie in Region I, at the edge of the continental
shelf, and 20% probability that it is in Region II, beyond the shelf.
Reading #
The two groups have been searching the regions for some time now. What is the
probability that the treasure is in region I, call this event AI , given that person Y did not
find it in region I, call this event B ?
p( AI | B) = p(treasure is in I | Person
Y did not find it in I)
p (Y didnt find it in I | it is in I ) p ( it is in I )
p ( B| AI ) p ( AI )+ p ( B| AI ) p ( AI )
=
=
(0.6)
PY
(0.8)
PI
PII
(0.2)
(1PY ) PI
(1PY ) PI + PII
0.40.8
0.40.8 + 0.2
= 0.62 = 62%
Y finds in I
=PYPI
=0.48
is in I
1-PY
(0.4)
is in II
=PII= 0.2
10
Reading #
The next day you discover that in addition to Y, X did not find the treasure in area I. So
what is the probability that X didnt find it after Y didnt find it and the probability that it is
in area I even though both parties failed to locate it? See the probability tree in figure 5
(0.6)
PY
(0.8)
PI
(0.7)
PX
1-PY
(0.4) 1-P
X
(0.3)
PII
(0.2)
Total probability that X does not find it in Area I given
that Y did not find it there either:
(1-PX)(1-PY)PI
PII+(1-PX)(1-PY)PI
= 0.324
5. Probability tree for the sunken treasure problemgiven that Y did not find it in area I.
11
Reading #
5. Recap of Probability
For events A & B in the
p( A B) = p( A) + p( B) p( A B)
space S :
p( A | B) =
p ( AB )
p(B)
p( A | B) = p( A)
If A is wholly contained in B
then
p( A B) = p ( A) + p( B)
p ( A B) = 0
p( A | B) = 0
n!
k!( n k )!
12
Reading #
(18)
Thus the single event containing (n k) failures and k successes has the probability
n
p k q n k and there are possible combinations.
k
7. Useful References
There are many probability text books each slightly different. Additional textbooks are
listed below.
Bertsakis and Tsikilis: Text book used for 6.431
Devore, J "Probability and Statistics for Engineers and Scientists"
Triantafyllou and Chryssostomidis, (1980) "Environment Description, Force
13
Reading #
1. Random Variables
A random variable is a variable, x , whose value is assigned through a rule and a random
experiment, , that assigns a priori a value to the outcome of each experiment,
A1 , A2 , A3 ,... This rule states that
x( A1 ) = x1
x( A2 ) = x2
L
x( An ) = xn
One example of a random variable is a Bernoulli random variable which assigns either a
1 or 0 to the outcome. For example, toss a fair coin. If it lands heads up you get one
dollar, if it land tails up you loose a dollar. The amount won or lost in this case is the
random variable.
Symbolically, x( ) denotes the random variable which is a function of the random
event = {A1 , A2 ,, An } which has associated probabilities: p( A1 ) = p1 , p( A2 ) = p2 , etc.
A1 x1 , p1
A2 x2 , p2
M
An xn , pn
Reading #
The variables xi are the values of the random variable, Ai , the possible events in the
event space, and pi is the probability of event Ai .
EXPECTED VALUE of the random variable can be thought of as follows: after many
( M ) repetitions of a random experiment in which event A1 occurred d1 times, A2
occurred d 2 times, and so on to An occurred d n times, the total number of experiments is
simply
M = d1 + d 2 + d3 +L + d n .
(19)
If a weight, or cost, xi , is assigned to each event, Ai , then the total cost of all of the
events is
xT = d1 x1 + d 2 x2 +L + d n xn .
(20)
Then given pi , the probability of event Ai , the expected value of the event is
N
x = E { X ( )} = pi xi .
(21)
i=1
x=
As m :
di
M
xT
M
(22)
by the total number of events, M , is equal to the probability of the event, pi , and the
expected value of x is the average income as defined in 22 as M
Reading #
x = x1 p1 + x2 p2 + L + xn pn .
(23)
E{x + y} = E{x} + E{ y}
E{C} = C ; C is a constant
E { g ( x( ))} = i=1 g ( xi ) pi
n
Properties of Variance
Vx
= E x( ) x
= E x 2 ( ) 2 x( ) x + x 2
= E{x( ) 2 } 2xE{x( )} + x 2
= E { x 2 ( )} x 2
2. Probability Distribution
Discrete Random Variable: possible values are a finite set of numbers or a countable set
of numbers.
Continuous Random Variable: possible values are the entire set or an interval of the
real numbers. The probability of an outcome being any specific point is zero (improbable
but not impossible).
EXAMPLE: On the first day of school we observe students at the campus bookstore buying computers.
The random variable x is zero if a desktop is bought or one if the laptop is bought. If
20% of all buyers purchase laptops then the pmf of x is
2004, 2005 A. H. Techet
Reading #
p( x) = p( X = x) = 0 for x 0 or 1.
(24)
F ( x xk ) = p ( xk )
(25)
j =1
xo
f x ( y )dy
(26)
Reading #
which is the area under the probability density curve to the left of value xo . Note that
F ( xo ) is a probability in contrast to the PDF. Also
F ( xo ) = p ( x xo ) =
xo
f x (x)dx
(27)
and
f x ( xo ) =
dF ( xo )
.
dx
(28)
Let x be a continuous random variable with a pdf f x ( x) and cdf F ( x) then for any value,
a,
p ( x > a) = 1 F (a);
(29)
p(a x b) = F (b) F (a )
(30)
Expected Value: The expected value of a continuous random variable, x, with pdf f x ( x)
is
x = E ( x) = x f x ( x)dx
(31)
E[h( x)] = h ( x ) = h( x) f x ( x) dx
(32)
Reading #
Conditional Expectations: The expected value of the random variable given that the
(33)
F ( X xo ) = F ( y ( x) g ( xo ))
(34)
The probability that the random variable, X, is less than some value, xo , is the same as
the probability that the function y ( x) is less than the at function evaluated at xo .
EXAMPLE: Given y = x + b and the pdf f x (x) for all > 0 , then x + b < yo for
x
yo b
and
F ( y yo ) =
yo b
f x (x)dx
(35)
(36)
| dy |
| dx |
(37)
Reading #
If y x1 , x2 ,L
, xn
then
fy =
f x ( x1 )
dy ( x1
)
dx
+L+
f x (xn )
(38)
dy ( xn )
dx
x=
1
n
xi
n i =1
(39
and
2 ( x) =
1 n 2
n i =1
(40)
1. Random Processes
A random variable, x( ) , can be defined from a Random event, , by assigning values xi
to each possible outcome, Ai , of the event. Next define a Random Process, x( , t) , a
function of both the event and time, by assigning to each outcome of a random event, , a
function in time, x1 (t ) , chosen from a set of functions, xi (t) .
A1
A2
M
An
p1
x1 (t )
p2 x2 (t )
M
M
pn xn (t )
(6)
This menu of functions, xi (t) , is called the ensemble (set) of the random process and
may contain infinitely many xi (t) , which can be functions of many independent variables.
EXAMPLE: Roll the dice: Outcome is Ai , where i = 1 : 6 is the number on the face of the
dice and choose some function
xi (t ) = t i
to be the random process.
(7)
1 T
xi (t ) dt = x t
0
T
(8)
1 T
[xi (t) M { xi (t)}]2 dt
T 0
(9)
M { xi (t )} =Tlim
Rit ( ) =Tlim
1 T
[ xi (t) M t { xi (t)}][ xi (t + ) M t { xi (t + )}]dt
0
T
(10)
If Rit is large (i.e. Rit ( ) 1) then xi (t) and xi (t + ) are similar. For example, a
sinusoidal function is similar to itself delayed by one or more periods.
EXPECTED VALUE:
x t1 = E{x(t1 )} = x f ( x, t1 )dx
(11)
STATISTICAL VARIANCE:
(12)
AUTO-CORRELATION:
Rx xt1 , t2 = E{x(t1 , )x(t2 , )} = E {[ x(t1 , ) E{x(t1 , )}][ x(t2 , ) E{x(t2 , )}]} (13)
Example: Roll the dice: k = 1 : 6 Assign to the event Ak (t) a random process function:
xk (t ) = a cos ko t
(14)
MEAN:
VARIANCE:
M t {xk (t)} =
lim 1
T T
V t {xk (t )} =
CORRELATION: R t {xk (t )} =
=
lim 1
T T
lim 1
T T
a 2 cos 2 ko tdt =
a cos ko tdt = 0
a2
2
cosko
Looking at the correlation function then we see that if ko t = /2 then the correlation is
zero for this example it would be the same as taking the correlation of a sine with cosine,
since cosine is simply the sine function phase-shifted by /2 , and cosine and sine are not
correlated.
Now if we look at the STATISTICS of the random process, for some time t = to ,
xk ( , to ) = a cos(koto ) = yk ( )
(15)
EXPECTED
VALUE:
VARIANCE:
V { y ( )} = k =1 16 a 2 cos 2 (ko to )
CORRELATION:
In general the Expected Value does not match the Time Averaged Value of a function i.e.
the statistics are time dependent whereas the time averages are time independent.
x (t1 ) = E{x(t1 , )} f (t )
V (t ) = x (t1 ) = E [x(t1 ) x (t1 )]2 = x
2
The statistics, or expectations, of a stationary random process are NOT necessarily equal to
the time averages. However for a stationary random process whose statistics ARE equal to
the time averages is said to be ERGODIC.
(16)
yi (t ) = a cos(ot + i )
(17)
where ( ) is a random variable which lies within the interval 0 to 2 , with a constant,
uniform PDF such that
1/ 2 ; for(0 2 )
f ( ) =
else
0;
(18)
1
a cos(oto + )d = 0
2
(19)
V (to ) = R( = 0) =
a2
2
(20)
1 2
a cos(oto + ) cos(o [to + ] + )d
2
1 2
a cos o
2
(21)
Lets next look at the temporal averages for this random process:
=Tlim
1 T
a cos(o t + i )dt
T 0
1 a
[sin(oT + i )] = 0
T o
(22)
TIME VARIANCE:
V t = R t (0) =
a2
2
(23)
CORRELATION:
R t ( ) =Tlim
1 T 2
a cos(o t + i ) cos(o [t + ] + i )dt
T 0
1
= a 2 cos o
2
(24)
1 T
yi (t ) yi (t + )dt
T 0
(25)
y( , t) = an cos(n t + n ( ))
(26)
n=1
where n ( ) are all independent random variables in [0, 2 ] with a uniform pdf. This
random process is stationary and ergodic with an expected value of zero.
The autocorrelation R( ) is
R( ) =
n=1
an2
cos(n )
2
(27)
Reading #
y(z , t) = x(t, z ) - xo .
Again we note that for the stationary ergodic random process the time statistics and event
statistics are equal. We write the autocorrelation R(t ) :
R(t ) = E{ y (t ,z ) y (t +t ,z )} = Rt (t ) = lim
1 T
yi (t ) yi (t + t )dt
T 0
(1)
CORRELATION PROPERTIES
1. R(0) = variance = s 2 = (RMS) 2 0
2. R(t ) = R( -t
)
3. R(0) | R (t ) |
(2)
n=1
where y n (z ) are all independent random variables in [0, 2p ] with a uniform pdf. This
Reading #
an2
cos(wnt )
2
R(t ) =
n=1
(3)
2. SPECTRUM
Given a random process that is stationary and ergodic, with an expected value of zero and
autocorrelation R(t ) , the power spectral density, or spectrum, of the random process is
defined as the Fourier transform of the autocorrelation.
S (w ) = R(t )e-iwt dt
(4)
1
2p
R(t ) =
S (w )eiwt dw
(5)
1
2p
S (w)d w
Reading #
1
2p
S (w ) = lim
-T
y (t,z k )e -iwt dt
(Beyond the scope of this course see Papoulis p. 343 for more info)
A spectrum covers the range of frequencies from minus infinity to positive infinity
( - < w < + ).
A one-sided spectrum, S + (w ) , is a representation of the entire spectrum only in the
positive frequency domain. This one-sided spectrum is convenient and used traditionally,
but is not strictly correct.
The one sided spectrum is a representation of the entire spectrum only in the positive
frequency domain. We fold" the energy over w = 0 and introduce the
1
2p
factor we get:
S (w ) w 0
S (w ) = 2p
0
else
+
(9)
This representation for the one-sided spectrum comes from the variance, R(0) :
R(0) = s 2 =
1
2p
S(w )d w =
2
2p
S(w )dw
(10)
Reading #
s 2 = S + (w )d w
(11)
where
S + (w ) =
2
S (w); for w 0
2p
(12)
We can use the spectrum to help us analyze linear time invariant systems. Since the LTI
system is characterized by its impulse response, h(t ) , given an input, u (t ) , the output can
be found from the convolution of the impulse response and the input:
y ( t ) = u (t ) * h( t )
(13)
Reading #
or the Fourier transform of the output is equal to the Transfer function, H(w), times the FT
of the input:
y (w ) = H (w ) u (w )
(14)
For such a LTI system, if u (t ) is a stationary and ergodic random process then y ( t ) is also
stationary and ergodic. Defining the spectrum of u (t ) as Su (w ) and the spectrum of y ( t )
as S y (w ) we can show that the following holds true:
S y (w ) =| H (w ) |2 Su (w )
where H (w ) is square of the magnitude the transfer function of the LTI system. This is
2
known as the Wiener-Khinchine Relation. We would like to use this relationship and
properties of the spectrum to gain insight about the system output, essentially the statistics
of the output, knowing only the input and the system transfer function.
Reading #
Often we need to know how often is a certain level is exceeded by the process, in this case
the wave height. In order to determine this, we can look at the occurrences of
UPCROSSINGS only in a certain time period of interest. (There is further information on
this subject in section 3 of the supplemental notes by Traintafyllou and Chryssotomidis.)
Zeroth Moment:
M o = S + (w ) dw = s 2 = VARIANCE
0
Second Moment:
(15)
Reading #
M 2 = S + (w ) w 2 dw
(16)
Fourth Moment:
M 4 = S + (w ) w 4 dw
(17)
Lets define h( A) as the average frequency of upcrossings past a certain level A (crossings
above A per time) and h(0) as the average frequency of all upcrossings (past a zero level),
such that
h( A) =
1
2p
M 2 - A2/ 2M o
e
Mo
(18)
and
h(0) =
1
2p
M2
.
Mo
(19)
We now have an equation for the upcrossing frequency which can be easily determined if
we know the average period, T . The average period can be thought of here as the expected
value of all periods, Ti , such that
1
1
T 2p
M2
Mo
(20)
and
h( A) =
1 - A2/ 2M o
e
T
7
(21)
Reading #
Su (w ) =| H (w ) |2 S y (w ) = 1 S y (w )
(22)
We want the frequency of upcrossings above the deck height, h , to be one every ten
minutes:
h(h) =
10
min
1
= 1/ 600(times/sec ).
*60 sec/min
(23)
We also have the equation for the number of upcrossings above height h as
h(h) =
1 - h2/ 2Mo
e
T
(24)
1
T
e - h / 2Mo = 1/ 600
2
T
600
= e - h /2 Mo
2
8
2M o ln ( 600
) = h2
h = 5. 877 meters
Reading #
This number of times above a level is sufficient for deck wetting frequencies. However for
structural design calculations where peak stresses are needed to check for structural safety,
we need to consider the probability of a maximum level, xm , reaching a certain design
level, A . For this it is ideal to define here the bandwidth of the spectrum in terms of the
moments of the Spectrum.
Bandwidth is defined as
e 2 = 1-
M 22
MoM4
(25)
The value of e is between 0 and 1. In the ocean, a bandwidth between 0.6 and 0.8 is
common.
Reading #
f ( xm ) =
2/ M o e
1 A2
EXP
- 2
1+ 1 - e 2 2p
e 2M o
1 -e 2
A2
A2
EXP + 1- e 2
1f
Mo
e
2M o
A2
Mo
(26)
f (x ) =
1
2p
e - u /2 du.
2
(27)
f ( xm )
A2
2 1- e 2 A
EXP
1 + 1 -e 2 M o
2M o
(28)
Note that this approximation is not a true pdf as it does not integrate to one.
IF e = 0
f (xm = A) =
A - A2/ 2M o
e
; 0< A<
Mo
(29)
f ( xm = A) =
1 - A2/ 2M o
; 0< A<
e
2p
(30)
IF e = 1
10
Reading #
For most studies of structures and ships we are most interested in the probability that a
wave height exceeds a certain value. It can be shown that for large values of the wave
maxima equation 26 reduces to the approximate pdf
f ( X m = A)
2 1- e 2
1 + 1 -e 2
A - 2AM2o
e
Mo
(31)
This is not a true pdf since it will not integrate to 1. However the approximation holds
over a range bandwidths. It is especially good for values of e below 0.6 with A/ M o
greater than 1.4. For larger e we need larger values of A/ M o . Since e = 0.6 is typical in
the ocean we can justify this approximation.
Lets define some height h = A/ M o , where h is large, and rewrite the approximate pdf
as
f (h m =h )
2 1- e 2
1+ 1- e
(32)
Since we know that a pdf must (technically) integrate to one and that the probability that a
random variable is below a value A is the integral of the pdf from zero to the height A, then
we can readily calculate the probability that the wave height is above the level A using the
pdf given above. The probability that a wave height is above a level a is
P ( xm A) =1 - f ( X m = a)da = f (X m = a)da
0
(33)
For the most part this integral cannot be easily evaluated without using integral tables or
11
Reading #
numerical integration.
Using the non-dimensional amplitude, ho which is always greater than 1. Recall:
h o =
A
> 1
Mo
(34)
To find the probability that the wave maxima is greater than some amplitude, lets look at
the probability that the non-dimensional wave amplitude is greater than ho :
2 1- e 2
ho
1+ 1- e 2
P(h h o ) = f (h = u )du
e-ho / 2
2
(35)
5. Useful References
Devore, J "Probability and Statistics for Engineering and Sciences"
Triantafyllou and Chryssostomidis, (1980) "Environment Description, Force
Prediction and Statistics for Design Applications in Ocean Engineering" Course
Supplement.
12
1. Gaussian Distribution
Distributions of random variables are often gaussian in shape, or can be approximated as
such. The gaussian density function is described by the probability density function
f ( x) =
2 2
( x x2)
(1)
xx
1+ erf
F ( x) =
2
(2)
1
2
e y / 2 dy
(3)
2. Poisson distribution
Discrete events occur randomly in time with the following probability as t 0 :
to have 1 occurence in time interval t
t
1 t
0
to have more than one occurence in time t
(4)
Thus the probability to have k occurrences within a finite time t can be shown to be
P(k in t) = e t
( t ) k
k!
(5)
This can be useful when designing platforms that require less than k occurrences of an
event in a certain time t (e.g. less than ten times water on deck in one day).
Reading #
1. Wave energy spectra. Red text indicates wave generation mechanisms and blue text indicates
damping/restoring forces.
The majority of ocean waves are wind generated. Other wave generating mechanisms
include earthquakes and planetary forces. Planetary forces drive tides and cause long
Reading #
period waves on the order of 12 to 24 hours. Earthquakes are the major cause of tsunamis
which, while rare, can be catastrophic if the earthquake occurs near or on the coast.
Waves also encounter forces that tend to restore them to a flat surface. For small
wavelength (high frequency) waves surface tension plays a large role in damping out these
waves. The majority of waves are restored by gravity and longer period waves are damped
by the Coriolis force.
As wind begins to blow (between 0.5 - 2 knots) on a calm surface small ripples, capillary
waves or cat-paws, tend to form. These small waves are on the order of less than 2 cm.
As the wind becomes stronger wave amplitude increases and the waves become longer in
order to satisfy the dispersion relationship. This growth is driven by the Bernoulli effect,
frictional drag, and separation drag on the wave crests.
Wind must blow over long periods of time and large distances to reach a fully developed
sea state. When the phase speed of the wave crest matches the wind speed non-linear
interactions stop (except friction) and the phase speed is maximized. The limiting
frequency of the waves can be determined by the equation for phase speed and the
dispersion relationship:
C p U w = w/k = g /w
wc
(1)
g
Uw
(2)
where U w is the wind speed and wc is the limting frequency. Once wind stops viscosity
erodes the waves slowly. The smallest wavelengths decay the fastest. Sample spectrum
shapes are shown in figure 2.
Reading #
For a storm with wind speed, U w , the effects of the storm can be felt at a distance from the
storm, R . The number of wave cycles between the storm and the observation location is
N = R/l . The amplitude of the waves decays as e -g t where g = 2n k 2 = 2nw 4 /g 2 (from
The development of storms can be tabulated. Fetch is the length over which the wind must
blow to have fully developed seas (given in standard miles), and the storm duration, given
in hours, is the time the storm must last to result in a fully developed sea.
Wind
warnings
small craft
gale
hurricane
Beaufort
scale
Wind
speed (mph)
Fetch
(miles)
Storm
duration (hr)
3-4
12
15
5-6
25
100
12
35
400
28
50
1050
50
70+
Reading #
S + (w ) =
A - B/w 4
e
w5
(3)
The frequency peak is called the modal frequency. The area under the spectrum is the
zeroth moment, M o , which may be defined in terms of the significant wave height. For a
narrow-banded spectrum the significant wave height is approximately four times the
square root of the zeroth moment. Since the significant wave height depends on the wind
speed, the spectrum could be formulated in terms of the wind speed instead of the
significant wave height. While certain spectra can have more than one peak, it is assumed
that a single storm produces a single-peaked spectrum and any second peak is due to a
distant storm that sends waves to the considered location.
Several more specific theoretical representations of wave spectra have been developed
using data collected by observation platforms and satellite data in various regions. These
spectra are discussed below. When considering which spectrum formulation it is important
to take into account the specific criteria that were used in developing the spectrum.
Reading #
Presence of swells: Swells are waves that result from distant storms that travel a
significant distance and arrive often at an angle that differs from the wind direction.
If we use a spreading function to correct a unidirectional spectrum it will not
account for the presence of swell. It is also important when measuring waves that
the component that results from swell be accounted for separately.
The Pierson-Moskowitz spectrum (equation 11) was developed for fully developed seas in
the Northern Atlantic Ocean generated by local winds.
2 2
8.1 g 2 -0032(
S (w ) = 3 5 e . g/zw )
10 w
(4)
z H 1/ 3 = 4 M o ,
(5)
wm = 0.4 g /z .
(6)
This spectrum is developed under the following conditions: unidirectional seas, North
Atlantic Ocean, fully developed local wind generation with unlimited fetch. The most
critical of these assumptions is the fully developed assumption. For it is possible to achieve
a larger heave response for a platform from a developing sea, even though the significant
wave height may be smaller that that of a fully developed sea, since the modal frequency is
higher and heave motions tend to have higher natural frequencies. In the case of a rolling
ship the decaying sea might excite a larger roll motion since the natural frequency of roll
tends to be relatively low.
In order to overcome the limitation of fully developed seas, a two parameter spectrum was
developed. This spectrum is the Bretschneider spectrum (equation 14). The B-S spectrum
Reading #
S + (w ) =
1.25 wm4
. (wm /w )4
z e-125
5
4 w
(7)
z H 1/ 3 = 4 M o ,
(8)
If wm satisfies equation 13 then equation 14 reduces to equation 11. By allowing the user
to specify the modal frequency and significant wave height, this spectrum can be used for
sea states of varying severity from developing to decaying.
The Ochi Spectrum (equation 16) is a three parameter spectrum that allows the user to
specify the significant wave height, the modal frequency, and the steepness of the spectrum
peak.
l
4l +1
4
4l +1 wm 4
1 4 wm z 2
S (w ) =
exp
-
4l +1
4 G(l ) w
4 w
+
(9)
where G ( l ) is the gamma function, and l is the parameter that controls the spectrum
steepness. For l = 1 , equation 16 reduces to equation 14. The Ochi spectrum is limited in
that it also considers only unidirectional seas and unlimited fetch, but the designed can now
specify the spectrums severity (z ), the state of development (peak frequency wm ) and
isolate the important frequency range by dictating the spectrum width ( l ). The ability to
dictate l allows the designer to account for swell from a distant storm.
The JONSWAP spectrum (equation 17) was developed by the Joint North Sea Wave
Project for the limited fetch North Sea and is used extensively by the offshore industry.
This spectrum is significant because it was developed taking into consideration the growth
of waves over a limited fetch and wave attenuation in shallow water. Over 2,000 spectra
were measured and a least squares method was used to obtain the spectral formulation
Reading #
ag 2 - 54 ( wwm ) d
g
e
w5
(10)
(w - wm )2
2s 2wm2
(11)
S + (w ) =
where
d =-
a = 0.076 x (-0 .2 2 )
(12)
gx
U2
(13)
0.07; w wm (21)
0.09; w > wm
(14)
x=
s =
The wind speed in knots is U , x is the fetch in nautical miles, and the modal frequency
can be found as
.
wm = 2p * 3.5 * ( g /U ) x -033
.
(15)
(16)
We can account for the effects of two separate storms by adding the respective spectrums:
S + (w ) = S1+ (w ) + S 2+ (w )
(17)
We can also correct for directionality multiplying the spectrum by a spreading function,
Reading #
M (m) ,
+
(w ) M (m )
S + (w, m ) = S BS
(18)
where M ( m ) spreads the energy over a certain angle contained within the interval [ -p , p ]
from the wind direction. The integral of M ( m ) over this interval is one.
-p
M (m) d m = 1
(19)
M (m) =
2
cos2 m
p
(20)
on the interval
m
m
<m<
2
2
(21)
3. Bretschneider Spectrum
To recap, the 15th International Towing Tank Conference (ITTC) in 1978 recommended
using a form of the Bretschneider spectrum for average sea conditions when a more
specific appropriate form of the wave spectrum is well defined. The general form of this
spectrum is equation refeq:specgen.
S + (w ) =
A - B/w 4
e
w5
(22)
The two parameters A and B are dependent on the modal frequency, wm , and the variance
of the spectrum, M o = (rms) 2 = s 2 .
Reading #
wm4 =
4
B ; B = 5wm4 /4
5
(23)
Variance = s 2 = A/(4 B) ; A = 4s 2 B
(24)
(25)
S (w ) =
. ( wwm )
1.25 wm4 2 -125
z
e
4 w5
(26)
The moments of the spectrum can be calculated numerically. For simplification the
following relationships have been given (see Principles of Naval Architecture vol. III for
further discussion). The fourth moment diverges slowly as w thus the approximation
helps analysis when calculation of this moment is necessary.
M o = VARIANCE = (RMS) 2
Reading #
duration of time which can consist of multiple periods), then within this interval of time
there are a number of maxima, a1, a2, a3
, a 4, , an , where a1/N is the value exceeded by 1/N
of the maxima. (Note: a1/N is not a to the 1/N power)
3.
EXAMPLE: Take N = 3 ; From a sequence of twelve measured wave heights find the
5 3 4 7
11
8 9 5 4 2 5
There are 12 recorded wave maxima so there are four wave maxima above the 1/3rd
10
Reading #
highest wave height. For this sequence a1/ 3 = 6 since it is the next highest wave maxima
below 7. Had this been a infinitely long series of observed wave heights and 7 was the
lowest value of the 1/3rd highest wave maxima, then 6.99 might been a better estimation
for this value. However in this short sequence the answer is 6.
[1 8 3 4 11
11
11
11 5 4 2 5]
then a1/ 3 = 8 , since the four highest wave heights are all equal to 11.
The probability of wave heights occurring above the 1/Nth highest wave is given by
P(h h 1/N ) =
1
2 1 - e 2 -(h1/N )2 / 2
e
N 1 + 1- e 2
(27)
where
h 1/N =
2 1- e 2
a1/N
N
= 2ln
1+ 1- e 2
Mo
(28)
The average value of ALL maxima above a1/N is called the 1/N (Nth) highest average
amplitude. This can be found using the formula for expected value of a variable:
(29)
This is the expected value given am is greater than a1/N as can be represented as
11
Reading #
(30)
p am = a | am > a
1/N
(31)
Keeping the amplitude in non-dimensional form we can calculate the Nth highest average
wave height using the approximate pdf given in the last reading.
h 1/N
2 N 1 -e 2
1+ 1- e 2
1 /N
h o2 e -ho / 2dh o
2
(32)
a1/N 2s = 2 M o for e < 0 .5 . The significant wave height is defined as twice the
significant wave amplitude,
H 1/ 3 = 2 a1/ 3 .
(33)
This value is very close to that which a casual observer would estimate as the wave height
when watching the sea. This makes the significant wave height a very useful statistic. Maps
of the significant wave height over the entire earth can be seen on satellite images. There
are several links on the course webpage that illustrate this quantity over the earth surface.
From such a satellite composite image we can see that the southern ocean is the most
tumultuous ocean and has the highest significant wave height.
12
Reading #
Ti = T Pi
(34)
From short term statistics we have the frequency of waves, upcrossings, exceeding a
certain amplitude, ao .
n(ao ) =
1
2p
M 2 -ao2 / ( 2 M o ) 1 - ao2/ ( 2 M o ) 1
=
e
=
e
m
Mo
T
T
(35)
Where T is the average period of waves in a storm. So it is easy to find the total number
of times the level ao is exceeded in a storm i , given by Ni as
N i = n(ao ) Ti =
Ti
mi
Ti
(36)
Where Ti is the life of the storm i , Ti is average period of the waves within that storm.
The total number of times level ao is exceeded during the life of the structure then
becomes:
N ao = N i =
all i
T
- a2/ ( 2M )
Pi e o o i
T
(37)
where M o i is the zeroth moment of the spectrum of storm i . It is necessary to make the
assumption that the spectrum is narrow banded so that we can write the zeroth moment in
13
Reading #
terms of the significant wave height. This simplifies the problem since the significant wave
height is the data most often available that describe sea conditions over a certain time
period.
Next we can adjust the equation for total number of upcrossings to reflect a number of
upcrossings exceeding the height, ho which is simply defined as twice the amplitude, ao .
N ho =
i
2h2
- 2o
T
Pi e z i
T
(38)
Equation 5 represents the total number of upcrossings past a level ho over a time
interval T . The total number of upcrossings past the mean water level, ho = 0 , is given by
equation 6.
No =
i
T
Pi
T
(39)
Having determined the equation for the total number of waves exceeding a level ho and
also the total number of upcrossings during the life of the structure, we can find the
probability of a wave height exceeding a design height ho in equation (40)
P(h > ho ) =
Nho
No
Pi
e -2ho /z i
2
T
Ti
i
Pi
(40)
T
Ti
This equation represents the fraction of total waves above level ho . Realizing the life span
of the structure, T , is constant we can cancel it out of equation (40) and rewrite it in the
form
P (h > ho ) =
N ho
No
1
Ti
( )P
i
14
e -2 ho /z i Pi
1
Ti
(41)
Reading #
The numerator and denominator of equation (41) each have the form of the expected value
formula for a random process with random variables T and z . So we can rewrite the
probability as
P(h > ho ) =
E T1 e -2ho /z
2
{ }
(42)
1
T
Note the subscript i is dropped since T is the random variable and Ti is a single random
event. Thus the expected value of the random variable is the sum over all the ith random
events. This can also be extended to a random process.
We now have and equation for the probability of waves above a level that is not implicitly
dependent on the life span of the structure. In practical application the average wave period
can be neglected (cancelled) and the probability approximated by
P (h > ho ) E e-2 ho /z .
(43)
One valuable calculation is related to the hundred year wave. This is the wave that exceeds
some value, h1 00 , on the average only once in every one hundred years.
(44)
Note: This equation takes into consideration 100 years with the appropriate number of leap
years. So now we have an equation for the probability of the hundred year wave.
E e -2h100 /z
T
3.16e + 9
(45)
Where T is the average period of waves over all storms. In order to solve for h100 we must
have statistics for very rough storms with high significant wave heights. Smaller, more
15
Reading #
frequent, values of wave height do not play a role in this calculation and it is only the most
ferocious storms that contribute to this value. Thus you can imagine in order to collect this
data observations during major storms, hurricanes, typhoons, etc., must be performed over
time. Under these conditions it is too risky for human observation and often equipment is
lost at sea making obtaining this challenging.
(
P ( x > xo ) = e
-
xo -x1
x 2 -x1
(46)
where x1 , x2 , and g are estimated from historical data and x1 is a threshold level below
which the statistics will not effect the extremes.
6. Encounter Frequency
The principle motions of a ship or structure in seas are defined as surge, sway, heave, pitch,
roll and yaw. A diagram of the coordinate system is shown in figure 1.
16
Reading #
4. Motion of a ship defined with a right handed coordinate system with forward motion as the positive
x-direction and positive heave upwards.
Waves incident on the structure or ship can be described as head seas, following seas, beam
seas, or quartering seas depending on the incident direction. Figure 2 illustrates the first
three cases and quartering seas are defined as those that approach the ship from the either
the port (left) or starboard (right) stern quarter of the ship (between 90o and 180 o or 180 o
and 270o ).
5. Incident sea description. Figure at left shows head, beam and following seas. The incident angle, m ,
is measured from the bow (x-axis is
m = 0o ) counterclockwise.
The motion of a ship, forward or otherwise, affects the way incident waves are viewed by
someone aboard the vessel. For example if the ship is making way in head seas with a
constant velocity, U , then the waves will appear to meet the ship at a faster rate than the
actual frequency of the waves. This new, or observed, frequency is termed the encounter
frequency, we . If the waves are incident on the ship at some angle, m , then the component
of the speed of the ship in the direction of wave propagation is U a = U cos m . The wave
crests move at the phase speed, C p = w/k and the relative speed between the ship and the
17
Reading #
waves is
U r = Ua + Cp = U cos m + w/k
(47)
Thus it appears that the waves have a phase speed U r such that
Ur =
we
w
= U cos m +
k
k
(48)
Using the dispersion relationship for waves in deep water we can rewrite the equation for
encounter frequency as
we =
Uw 2
cos m + w
g
(49)
for w > 0 . In practice we usually have the encounter frequency since it is observed and we
would like to calculate the actual wave frequency, so taking equation (49) we can solve for
frequency, w .
g
U cos m
w=
-1 1 + 4we
2U cos m
g
(50)
Looking at the equation 17 we see that there are several possible solutions for w . These
are dependent on the incident angle m . For 4we
U cos m
g
18
Reading #
| U cos m |> C p ,
we < 0 ,
| U cos m |= C p ,
we = 0 ,
| U cos m |< C p ,
we > 0 ,
The final case, where the waves move faster than the ship, can tend to cause problems in
controlling a vessel, especially when the seas are not directly from behind the ship. This
effect is greatest in roll and yaw. When a ship is moving with the waves but overtaking
them it can also appear as if the seas are approaching from the bow, thus the actual incident
angle is ambiguous and could be m or m + p .
The effect of encounter frequency also changes the observed spectrum of the seas. Energy
must be preserved under the spectrum thus we can look at the following
S (we )| dwe |= S (w )| dw |
S (we ) =
(51)
S (w )
(52)
dwe
dw
We can find the derivative of encounter frequency by the actual frequency by using
equation (49).
dwe
2w
= 1+
U cos m
dw
g
(53)
S (w e ) =
S (w )
1+ 2wg U cos m
19
(54)
Reading #
-g
.
2U cos m
w=
(55)
we =
g
.
4U | cos m |
(56)
we = w +
w |w |
U cos m
g
(57)
6. Encounter frequency versus actual frequency for following seas ( cos m < 0 ) and w > 0 .
20
Reading #
7. Encounter frequency versus actual frequency for the three conditions: head, beam and following
seas.
7. Useful References
Read section four of the supplemental notes: Triantafyllou and Chryssostomidis, (1980)
"Environment Description, Force Prediction and Statistics for Design Applications in
Ocean Engineering"
21
(1)
(2)
Given a body in the presence of the wave field we much consider the relevant boundary
conditions on the free surface, the seafloor and the body. Boundary conditions are, on the
bottom,
n =
= 0,
n
(3)
n =
= VB n = VBn .
n
(4)
At the free surface the kinematic and dynamic boundary conditions must be satisfied. The
version 1.0
updated 3/29/2005
-1-
2004, aht
linearized free surface boundary conditions are both taken about z = 0 as we are
accustomed to, given that a / << 1 ,
2
+g
=0
2
t
z
(5)
1
.
g t
(6)
In the case of a free floating body we must also take into consideration the wave field
generated by the body motion alone. At some distance far away from the body, the
potential function must take into consideration the waves radiating from the body.
Figure 1. Boundary conditions for the total potential must be met at three places: sea
floor (3), free surface (2), structure surface (4). The continuity equation must be
satisfied within the fluid (1).
Due to the nature of potential flow and linear waves it is possible to sum multiple potential
functions to obtain the total potential representative of the complete flow field. Each
component of the total potential must also satisfy the appropriate boundary conditions. For
linear waves incident on a floating body the total potential is a sum of the undisturbed
incident waved potential, I ( x, y, z, t ) , the diffraction potential, D ( x, y, z, t ) , due to the
version 1.0
updated 3/29/2005
-2-
2004, aht
( x, y, z, t ) = I ( x, y, z, t ) + D ( x, y, z, t ) + R ( x, y, z, t )
(7)
It is good here to note the important conditions on each component of the total potential.
The incident potential is formulated from that of a free wave without consideration for the
presence of the body. Therefore I ( x, y, z, t ) satisfies only the free surface boundary
conditions and the bottom boundary condition, in addition to the Laplace equation.
The diffraction potential, D ( x, y, z, t ) , must also satisfy the Laplace equation, the free
surface and the bottom boundary conditions. In order to compensate for the disturbance of
the incident wave around the body by the an additional condition at the body boundary
such that the normal gradient of the diffraction potential is equal but opposite in sign to the
normal gradient of the incident potential:
D
= I .
n
n
(8)
The radiation potential satisfies the same conditions as the incident potential as well as an
additional condition on the moving body boundary in the absence of incoming waves. On
the body boundary the normal gradient of the velocity potential must equal the normal
velocity of the body.
R
= VB n .
n
version 1.0
updated 3/29/2005
-3-
(9)
2004, aht
It is often easiest, for the purpose of systems analysis, to write the wave potential in its
complex form:
it + ,
(x, y, z, t) = Re
aI + aD e
R
(10)
where a is the wave amplitude and is the incoming wave frequency. The complex
m
incident potential is I = a I eit . The amplitude of the incident potential is the wave
m
amplitude times I , which is a function of depth and position in space. The diffraction
potential takes the same form in order to satisfy the body boundary condition.
The radiation potential is not necessarily directly related to the wave amplitude. Since R
results from the motion of a floating body in the absence of waves, we must consider the
body motion in all six degrees-of-freedom. The vessel motions are prescribed by, x j ,
where j = 1, 2, 3, 4, 5, 6 (surge, sway, heave, roll, pitch, and yaw). It is customary to write
the complex radiation potential in the following form:
6
R = x j j
(11)
j =1
where x j is the velocity of the body in the j th direction and j is the velocity potential due
to a unit motion in the j th direction.
version 1.0
2I
t 2
updated 3/29/2005
-4-
2004, aht
3.
I
z
4. Incident potential does not have any knowledge of body in the flow.
I
z
I
z
= 0;
= 0;
4. Incident potential does not have any knowledge of body in the flow.
1.2.2. Diffraction Potential Boundary Condtions
The diffraction potential, D ( x, y, z, t ) , results from the presence of the structure in the
flow field. Without the structure, there would be no wave diffraction. This potential
accounts for the alteration of the incident wave train by the structure and we must now
include a boundary condition on the body (condition 4).
1. 2D = 0 ; Continuity equation.
2.
2D
3.
D
z
4.
D
n
t 2
4.
version 1.0
D
=0;
D
I
n
D
z
= 0;
updated 3/29/2005
-5-
2004, aht
2.
2R
3.
R
z
4.
R
n
t 2
(12)
(13)
version 1.0
R
z
=0;
R
= 0;
R
updated 3/29/2005
-6-
2004, aht
The total wave potential boundary conditions for unidirectional waves traveling in the
positive x -direction are a combination of the above cases. For a two-dimensional potential
and a body in heave-only motion:
( x, z, t ) = I ( x, z, t ) + D ( x, z, t ) + R ( x, z, t )
= Re a ( x, z ) + ( x, z ) eit + ( x, z ) eit .
x3 R
I
D
I ( x, z ) =
i cosh(k[ z + H ]) ikx
e ,
k
sinh(kH )
(14)
i kz ikx
e e .
k
(15)
I ( x, z ) =
3.
t 2
+ g z
= 0 ; Combined free surface condition.
The body boundary condition for the total potential (in heave motion) is
dx
= n = nz 3
n
dt
(16)
(17)
In order for equation (17) to hold true it the following must hold true:
version 1.0
updated 3/29/2005
n I + D = 0
(18)
R n = nz i
(19)
-7-
2004, aht
p(t ) +
1
2
+ + gz = C (t )
t 2
(20)
in order to use this with our linearized wave equations we must linearize the pressure.
Ignoring the hydrostatic term and the second order terms the unsteady pressure effects due
to the waves is the dynamic pressure. Since C(t) is arbitrary, we set it to zero. The force can
now be determined as follows
Fz (t ) = p(t ) n ds =
nz ds
S
S
t
(21)
Fz (t ) = Re
i eit nz a I + D + x 3 R ds ,
(22)
where x3 (t ) = x 3 eit is the heave motion. Each component of the total potential causes a
l it } , where F
l is the
force on the body. Given a general form of force F (t ) = Re{Fe
amplitude of the forcing function. The total force, F (t ) , is written as
lI + F
l D ) eit + m
l Reit + Fhydrostatic ,
F (t ) = Re a ( F
x3 F
(23)
where the hydrostatic force in heave is Fh = C33 x3 (t ) . The force amplitudes for the
incident, diffraction, and radiation forces are
l I = i nz I dS ,
F
s
version 1.0
updated 3/29/2005
-8-
(24)
2004, aht
l D = i nz D dS ,
F
s
(25)
l R = i nz R dS .
F
s
(26)
The amplitude of the radiation force comes from the case of the body heaving in a still fluid
(no waves) using from Newtons second law. The radiation force amplitude is simply the
sum of the added mass and damping components in heave
l R = 2 A33 i B33
F
(27)
where A33 and B33 are the added mass and damping coefficients in the heave direction due
to heave motion.
For the complete problem of a body heaving in an incident wave field, we must consider
the total force on the body. By Newtons law we can write an equation of motion by
equating the inertial force, the body mass times the heave acceleration, to the total force
due to the total wave potential: m
x 3 = F (t ) .
lI + F
l D ) eit + x 3 F
l Reit + Fh
2 m x 3 eit = Re a ( F
(28)
lI + F
l D)
2 (m + A33 ) + i B33 + C33 x 3 = a ( F
(29)
Using the above equation of motion we can now find the amplitude of heave motion in
terms of wave amplitude, inertial and diffraction forces, system natural frequency, system
mass, added mass, damping and hydrostatic coefficients.
lI + F
lD
x 3 =
F
2
a (m + A33 ) + i B33 + C33
(30)
where A33 , B33 , and C33 , are the added mass, damping and hydrostatic coefficients in
lI + F
l D ) is the excitation force amplitude, a F
l I is the
heave found using strip theory, a( F
l D is the diffraction force amplitude. All of these
Froude-Krylov force amplitude, and a F
depend on the frequency, , however the Froude-Krylov force amplitude, diffraction
force amplitude, added mass, damping, and hydrostatic coefficients are independent of
heave and wave amplitude.
version 1.0
updated 3/29/2005
-9-
2004, aht
= (I + D + R ) eit .
(1)
The radiation potential is comprised of six components due to the motions in the six directions,
= i n j ; ( j = 1, 2, 3)
(2)
= i (r n) j 3 ; ( j = 4, 5, 6)
(3)
n
j
n
r = (x, y, z )
(4)
n = n j ( j = 1, 2, 3) = (nx , ny , nz )
(5)
In order to meet all the boundary conditions we must have waves that radiate away from the body.
Thus j e ikx as x .
For the diffraction problem we know that the derivative of the total potential (here the incident
potential plus the diffraction potential without consideration of the radiation potential) normal to
the body surface is zero on the body:
version 1.0
updated 3/29/2005
T
n
= 0 on S B , where T = I + D .
-1-
2005, aht
= D ; on S B
n
n
(6)
We have so far talked primarily about the incident potential. The formulation of the incident
potential is straight forward from the boundary value problem (BVP) setup in lecture 15. There
exist several viable forms of this potential function each are essentially a phase shifted version of
another. The diffraction potential can also be found in the same fashion using the BVP for the
diffraction potential with the appropriate boundary condition on the body. This potential can be
approximated for a long wave condition. This long wave approximation assumes that the incident
wavelength is very long compared to the body diameter and thus the induced velocity field from
the incident waves on the structure can be assumed constant over the body and approximated by
the following equation:
i I
1 + I 2 + I 3
y
z
x
(7)
F FK (t ) =
version 1.0
updated 3/29/2005
-2-
I
n dS
t
(8)
2005, aht
x
T
B
Deep water incident wave potential is:
I =
a kz
e Re i ei (t kx)
k
(9)
The force in the vertical direction is found from the incident potential using eq. 8 along the bottom
of the vessel. Here the normal in the z-direction, nz , is negative: nz = 1, so the force per unit
length in the z-direction is
ia kT i (t kx )
B/ 2
i
FzFK = Re
e e
dx
B/ 2
k
o 2
k2
= Re
(10)
(11)
= Re 2 2 a e kT eit sin(kB/2)
k
eiz e iz
2i
(12)
Using the vertical velocity we can rewrite the force in terms of the velocity.
version 1.0
updated 3/29/2005
w(t ) = Re a ekz i ei (t kx )
(13)
w(t ) = Re a 2 e kz ei (t kx)
(14)
w( x = 0, z = 0,t) = Re a 2 eit
(15)
-3-
2005, aht
Now we can write the force in the vertical direction as a function of the vertical (heave)
acceleration,
2
(16)
Lets look at the case where 0 the wavenumber, k = 2 /g 0 , also goes to zero and the
following simplifications can be made:
e kt
1 kT
sin(kB/2)
(17)
kB/2
(18)
k2
Re 2
FzFK
a (1 kT ) (kB/ 2) eit
(19)
2 it
Re g aB 1
Te
(20)
If we look at the case where 0 and consider the heave restoring coefficient, C33 = g B ,
and the free surface elevation, ( x, t ) = Re a ei (t kx) we can rewrite this force as
Re {C33 (x = 0, t)}
FzFK
(21)
SB
= Re i
i a
k
I
nx dS
t
(22)
e kz dz ei (t kB/ 2) ei (t +kB/ 2)
(23)
a 2
kT i t
= Re i
2 sin( kB/ 2)
1 e
e
version 1.0
updated 3/29/2005
-4-
(24)
2005, aht
As frequency approaches zero similar simplifications can be made like above for the vertical
force:
a 2
Re i
( KT ) eit 2 kB/ 2
k
Fx (t )
(25)
u (t ) = Re a ekz ei (t kx )
(26)
u (t ) = Re i a 2 e kz ei (t kx )
(27)
Fx (t )
Re { TB u ( x = 0, z = 0, t )}
(28)
Where TB = , and is the vessel volume such that we are left with the surge force
(29)
C33 + w
(30)
Fx
Fz
-B/2
B/2
T
FxFK
bT u ( x = B/ 2, z = 0, t ) + bT u ( x = B/ 2, z = 0, t )
(31)
( x, t ) = a cos(t kx)
(32)
u ( x, z, t ) = a 2 e kz sin(t kx)
(33)
(34)
version 1.0
updated 3/29/2005
-5-
2005, aht
FxFK
(35)
2 b T (a 2 ) cos(kB/ 2) sin(t )
(36)
-B/2
B/2
c
b
Use the same assumptions from above to find the x-force adjusted for the additional pontoon
between the two hulls.
FxFK
2 b T (a 2 ) cos(kB/2) sin(t)
+c p ( x = B / 2 + b / 2, z = 0, t)
c p ( x = B / 2 b / 2, z = 0, t)
The last two terms are the adjustment to the force for the addition of the pontoon, FxFK (t ) .
(37)
version 1.0
updated 3/29/2005
-6-
(38)
2005, aht
Once we have set up the simple equation of motion for a vessel in heave it is natural to extend this
discussion to the coupled heave-pitch equations of motion. Considering a ship floating on the free
surface in waves. This ship will naturally heave and pitch due to the incident waves. It is not
guaranteed that these two motions will be independent, however. Thus it becomes necessary to
consider the motions together.
The body boundary condition must be properly specified, thus we need to know both the linear and
angular velocities of the vessel:
dx3 G dx5
G
k + r
VB=
dt
dt
j = (z x , x x x )
5
3
5
(1)
(2)
G
where the position vector, r , crossed with the unit outward normal, n , is
version 1.0
updated 3/29/2005
n4
G
n5 = r n
n6
-1-
(3)
2005, aht
In general the potential functions do not change significantly from our previous cases. However it
is necessary to account for the pitching motion in the radiation component of the total potential.
(4)
(5)
where the motions in the heave (3) and pitch (5) directions are
x3 (t ) = Re x 3eit
(6)
x5 (t ) = Re x 5eit .
(7)
The boundary conditions for the potentials, I , D , and 3 are the same as in the pure heave case.
We must now, however, consider also the boundary conditions for the radiation potential due to
pitching motions. Following the same formula from last lecture the equation of motion and
boundary conditions at the free surface, sea floor and on the body are:
1.
2 5 = 0
2.
2 5 + g
3.
4.
5
z
=0
5
= 0 on z = H
5
= n5 ; on the body
In addition to the force in the vertical direction, we also need to formulate the pitching moment
(about the y-axis) acting around the center of gravity. First we can re-write the vertical force
dependent on the two motions as:
F3 (t ) = nz
S
= Re
version 1.0
updated 3/29/2005
T
ds
t
(8)
i eit nz a I + D + x 3 3 + x 5 5 ds
-2-
(9)
2005, aht
l I3 + F
l D 3 + x 3eit ( F
l 33 ) + x 5eit ( F
l 35 )}
= Re {a eit F
(10)
l I3 + F
l D3 + x 3eit 2 A33 i B33 + x 5eit 2 A35 i B35
= Re a eit F
(11)
Next, we can find F5 , the moment on the body due to pitching and heaving motion.
M 2 = F5 =
S
G
(r n)dS
t
(12)
T
n5 dS
t
(13)
where
F5 (t ) =
S
l I5 + F
l D 5 + x 3eit ( F
l 53 ) + x 5eit ( F
l 55 )}
= Re {a eit F
(14)
l I5 + F
l D5 + x 3eit 2 A53 i B53 + x 5eit 2 A55 i B55
= Re a eit F
(15)
(16)
(17)
where the coefficients, Cij , are based on the vessel geometry. Taking the pitch motion about the
center of gravity (usually the c.g. is near midships), where xcg = 0 , we can write the equations of
d 2 x3
= F3 + F3h
dt 2
(18)
d 2 x5
= F5 + F5h
dt 2
(19)
The system of equations can now be written in matrix form assuming a harmonic input and an LTI
system:
version 1.0
m + A33
A53
A35
x 3 B33
+
I + A55
x 5 B53
updated 3/29/2005
l I3 + F
l D 3 it
B35 x 3 C33 C35 x3
F
+
= a
e
l I5 + F
l D 5
B55 x 5 C53 C55 x5
F
-3-
2005, aht
(20)
Simply expressed in matrix notation, the system of equations for the coupled pitch and heave
motions for a freely floating body can be rewritten as
[ M + A] x + [ B ] x + [C ] x = a Fl eit
(21)
where the vector motion is comprised of the linear heave motion, x3 , and the angular pitch motion,
x5 .
x 3 it
x = Re e
x 5
(22)
Writing the equation in matrix form allows us to better determine the amplitude response function
(transfer function) between the forcing and the ship motions:
x =
version 1.0
updated 3/29/2005
a
l
F
{ [ M + A] + i [ B ] + [C ]}
(23)
-4-
2005, aht
13.42 04/01/04:
Morrisons Equation
SPRING 2004
A. H. TECHET
1
Fx (t ) = D(t ) = CmU + Cd AU U
2
(1.1)
In order to obtain rough estimates of the magnitude of the force of a body, it is advantageous to
use Morrisons equation with constant coefficients. Supposing we want to find the estimates of
the wave forces on a fixed structure, then the procedure would be as follows:
1.) Select an appropriate wave theory (linear waves, or other higher order if necessary).
2.) Select the appropriate CM and CD based on Reynolds number, and other factors (see
table below).
3.) Apply Morrisons Equation
Wave Theory
Cd
Cm
Comments
Reference
Linear Theory
1.0
0.95
Wiegel, et al
on 13-24in cylinders
(1957)
Agerschou and
Edens (1965)
1.0-
2.0
1.4
data
Stokes 3rd order
Stokes 5th order
1.34
0.81.0
1.46
2.0
Keulegan and
Carpenter (1958)
Agerschou and
We can see from the above table that for linear waves the recommended values for drag
and mass coefficients are 1.0-1.4 and 2.0, respectively. The range of drag coefficients allows us
to account for roughness and Reynolds number effects. These values are for rough estimates. In
reality these coefficients vary widely with the various flow parameters and with time.
Bretschneider showed that the values of CD and CM can even vary over one wave cycle. Even if
we ignore the time dependence of these coefficients we must account for the influence of other
parameters.
Reynolds number and roughness effects: For smooth cylinders at Reynolds numbers
around 105, laminar flow transitions to turbulent flow, and there is a dip in CD as a function of
Re. For larger Reynolds numbers the separation point remains essentially constant and thus so
does the drag coefficient. In this range CD is Reynolds number independent. Roughness causes
the change from laminar to turbulent flow at a lower Reynolds number and increases the friction
and causes a larger CD. The mass coefficient is influenced by the changes in the boundary layer
and is thus also affected Reynolds number and roughness.
u(z,t)
dF ( z , t ) = CM
d 2 u ( z , t )dz + CD d
F (t ) =
z=L
z=0
1
u ( z, t ) u ( z, t ) dz
2
dF ( z , t )
(1.2)
(1.3)
The moment on the structure around the origin (point 0) is found by integrating the height z
times dF.
L
M (t ) = z dF ( z , t )
0
(1.4)
There are several limitations to these integrals. First we are limited to assume that each section
does not influence the adjacent sections flow. This assumption becomes questionable in the case
of a cross flow that forces direct interference between the flow of neighboring sections. The
second assumption is that the cylinder is not piercing the free surface because in this case the
water splashing must be taken into account. In the absence of experimental data we use
Morrisons equation as a first estimate in this class, but in reality we should attempt a correction
for the surface piercing phenomena.
2. Morrisons Equation when both the body and fluid are moving
Assume that a vertical cylinder is moving with velocity, u(t), within a fluid with velocity, v(t),
both velocities in the horizontal direction and uniform in space, then we can write Morrisons
equation as follows:
F (t ) = CM
d 2l v(t ) ( CM 1)
d 2 h u (t ) + CD
1
d l [ v(t ) u (t )] v(t ) u (t )
2
(2.1)
where d is the cylinder diameter and l the cylinder height. It is good to note that this equation
does not account for the inertial force due to the mass of the cylinder as required by Newtons
law. For example if the cylinder was subject to an inflow we could set F (t) = m a (t ) , where F(t)
is found using equation 3.1, m is the mass of the cylinder, and a(t) is the acceleration.
3. Forces on an Inclined Cylinder
Suppose that a cylinder of diameter, d, and large length, l, is at an angle within an
unsteady inflow, u(t), and we would like to use Morrisons equation. It has been suggested that
in such cases of slender objects (large l/d) that we can use the following approach: first
decompose the inflow velocity into two components Un and Ut, where Un is the velocity normal
to the cylinder and Ut is the component tangential to the cylinder; then we can use the following
expressions:
Fn = CM
1
d 2 U n + CD d U n U n
4
2
(3.1)
and
Ft = C f
1
d Ut Ut
2
(3.2)
to determine the normal and tangential forces per unit length. The mass and drag coefficients are
found using the diameter of the cylinder and the normal velocity as in the general (non-inclined)
case. The frictional coefficient, Cf, is used in the tangential case instead of the drag coefficient
since the drag results from the flow along (tangential) to the cylinder.
In the three-dimensional case where we have a vertical cylinder subject to a velocity
vector (u, v, w) then we must decompose this velocity into components normal and tangential to
the cylinder. The convention is to take the z-axis parallel with the cylinder axis (centerline) so
that the w component is the tangential component of velocity:
U n = u 2 + v 2 and U t = w
(3.3)
The velocity components (u, v, w) used in eq. (3.3) are taken in the coordinate system with zaxis parallel to cylinder axis. Equations (3.1) and (3.2) can then be used to determine the normal
and tangential force components. Finally the resultant force in a global coordinate system (say
with z-axis perpendicular to the free surface) can be determined using simple trigonometry.
These equations have some limitations on as incline angle, as experiments have shown that these
expressions are valid up to incline angles of about 60.
h
cos ( kx t ) we can determine the
2
relative magnitude of the integrated inertial and drag force terms in Morrisons equation.
Understanding the maximum forces and also when inertial effects dominate over viscous drag, or
vice versa, is important in designing offshore structures. The horizontal velocity and
acceleration under the wave at the centerline (x=0) of the cylinder are
u ( x = 0, z , t ) =
h cosh k
( z + H ) cos t
( )
2
sinh kH
(3.4)
cosh k
u
( z + H ) sin t
(0, z, t) = a 2
( )
t
sinh kH
(3.5)
Consider the force caused on a vertical cylinder in water depth, H, by a linear wave with
wavelength, , and height, h, given in the form
F (t ) = FI (t ) + FD (t )
(3.6)
where FI(t) is the inertial force term and FD(t) is the viscous drag force term. The inertial force
FI is dependent on the mass coefficient and can be determined by integrating the force acting
over the height of the structure. For a structure fixed to the seafloor the inertial force (in the xdirection) due to waves impinging on a circular cylinder is
FI (t ) =
cosh k ( z + H )
h
d 2 2
sin (t ) dz .
4
2
sinh ( kH )
(3.7)
FI (t ) = CM
d2
h 2
sin (t ) .
2k
(3.8)
sinh ( 2kH ) kH
+
4
2
1
h 2 2
FD (t ) = CD d
cos (t ) cos (t )
k sinh 2 ( kH )
2
4
(3.9)
(3.10)
The maximum values of these two force components are useful when considering design loads
on a structure. One way to determine the maximum is to plot equations (3.8) and (3.10), or
simply to look generally at the equations. The maximum inertial force using the result in eq.
(3.8) is
FI max =
CM d 2 2
h
,
2k
(3.11)
FDmax
sinh ( 2kH ) kH
+
2
4
2
1
h
2
= Cd d
.
k sinh 2 ( kH )
2
4
(3.12)
To compare the effects of inertial versus drag components we can take the ratio FI to FD.
FDmax
FI max
CD h sinh ( 2kH ) kH
1
+
2
CM d
4
2 sinh ( kH )
(3.13)
Substituting the typical values for drag and mass coefficients, 1.0 and 2.0, respectively, allows us
to determine how the maximum drag force compares to the maximum inertial force, i.e.
If
(3.14)
e kH
h
e
4 4 H
d
e( ) + 4
If
(3.15)
is small (very shallow water) then sinh(kH ) kH and the shallow water asymptote for h/d,
2
H
= ( 2 )
2 H
(3.16)
We could also consider the maximum drag and inertial force (force/length) terms at a specific
depth, z, below the free surface:
2
h 2 2 cosh k
1
( z + H )
FDmax ( z ) = CD d
2
4
sinh 2 ( kH )
(3.17)
cosh k
h
( z + H )
d 2 2
4 2
sinh ( kH )
(3.18)
FI max ( z ) = CM
CD h cosh k
( z + H ) .
sinh ( kH )
CM d
( 3.19)
It is evident from equation (3.17) that the drag force is most significant nearest to the surface and
it decreases with increasing depth. Take as an example, the case of deep water waves with
typical values for Cd and Cm (1.0 and 2.0 respectively), the ratio of drag and inertial forces as a
function of depth are
FDmax ( z )
FI max ( z )
1 h kz
e
2 d
(3.20)
The maximum total force integrated over the height of the cylinder can be determined using
equations (3.8) and (3.10). Thus total force as a function of time is
h 2
sin (t )
4 2k
sinh ( 2kH ) kH
+
4
2
1
h 2 2
+ CD d
cos (t ) cos (t )
2
4
k sinh 2 ( kH )
F (t ) = CM
d2
(3.21)
(3.22)