You are on page 1of 19

Ecole Centrale de Nantes

Hydrodynamics
Final Project Report

PARTICLE TRAJECTORIES IN
DEEP & INTERMEDIATE DEPTHS
MASS TRANSPORT UNDER WAVES

El Mehdi Dakir
El-mehdi.dakir@eleves.ec-nantes.fr

El Mehdi Es-sabar
El-Mehdi.Es-sabar@eleves.ec-nantes.fr

02/05/2022
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

1. Introduction
The aim of this report is the study of particle trajectories for deep and intermediate waters. For
this study, the assumptions of a potential fluid were made in order to better approach the problem,
therefore water is assumed to be as an inviscid fluid since the viscous effects are negligible between
two fluid layers and between the water and the air, also the fluid is considered as incompressible,
in other terms div(V⃗ ) = 0. Moreover, the pressure is continuous through the interface between air
and fluid, also that there is no mass flux through an impermeable boundary. The last assumption
is that a particle position at the free surface will not change with time, which leads to cancelling
all the time dependant factors in our equations.

To have more simplifications for our problem we will consider only small amplitude waves, that
is to say that the wave amplitude is much smaller than the wave length, in addition to that, the
wave height to wavelength ratios λh is taken to be less than 71 , and also by assuming that the
steepness of the wave is so small ϵ ≪ 1, one can linearize the free surface boundary conditions.

Results for each case will be discussed for the different conditions imposed to study the particles
movement (time dependant or not), and these results will be compared to the theoretical values
for both the velocity. The resulting phenomenons will be discussed as well, such as the drifting
motion and the horizontal transport of water mass.

In order to emphasize the effectiveness of the program used to model our problem, a convergence
study will be presented for the different factors affecting the results.

1
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

2. Methods
In this section, the methods followed in order to simulate the particle trajectories along with the
different assumptions used are exposed. Since our methodology bases itself on a mathematical
model, this latter will be detailed hereafter. Following that, its numerical implementation will be
discussed with a list of the tools used in it. This process will enable us a direct linking between
the theory and the numerical model in order to better interpret the results and draw relevant
conclusions.

2.1 Principle of water wave modelling: mathematical model


This first subsection that discusses our methodology will put forward and explain the mathematical
model used in our study, along with the general assumptions related to it. This will pave the way
for our results analysis, as the mathematical model along with its assumptions are crucial to that.

2.1.1 Problem setup


As said in the introduction, to put the problem of free surface gravity wave in a form of an equation,
we will assume the potential theory and ignore all the viscous effects, in addition to that a set
of assumptions presented in the introduction will be applied in order to simplify the real wave
propagation which occurs in 3D, and transform it to a 2D wave problem.

To solve the problem we will set the unknowns first :




Velocity field : V = ∇(ϕ(x, y, z, t))
Free surface elevation : η[x, y, t))
Pressure field : p(x, y, t)

Simplifications of the Bernoulli’s equation gives the following non linear version :

2
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Assuming once a particle is on the free surface, it remains there always. Similarly, the normal
velocity of a particle on the surface follows the normal velocity of the surface itself and that the
bottom is flat, the problem reduces to the following :

The simplifications made led to the linearized version of the problem which is applicable at
z=η.
For further simplifications, we will assume that the fixed boundary η is set to be free surface
where z=0 instead of z=η.

2.1.2 Solution of the linearized problem


The boundary value wave problem consequently become dependent on Laplace’s equation :

∂2ϕ ∂2ϕ
∇2 ϕ(x, z, t) = + 2 (1)
∂x2 ∂z
With the following boundary conditions :
Bottom condition
∂2ϕ
= 0on z=-H (2)
∂z 2

Free Surface Dynamic Boundary Condition


∂ϕ ∂η
= on z=0 (3)
∂z ∂t

Free Surface Kinematic Boundary Condition


∂ϕ
+ gη = 0 (4)
∂t
By the mean of separation of variables we can solve Laplace’s Equation for both η and ϕ.

η(x, t) = acos(kx − ωt + Ψ) (5)



ϕ(x, z, t) = − f (z)sin(kx − ωt + Ψ) (6)
k
u(x, z, t) = aωf (z)cos(kx − ωt + Ψ) (7)
w(x, z, t) = −aωf1 (z)cos(kx − ωt + Ψ) (8)
cosh(k(z + H))
f (z) = (9)
sinh(kH)

3
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

sinh(k(z + H))
f1 (z) = (10)
sinh(kH)
dispersion relation
ω 2 = gktanh(kH) (11)
Where a, ω,Ψ, k are considered as integration constants with the following interpretations : a
is the wave amplitude , k is the wave number, k= 2πλ , ω is the frequency and Ψ is adding a phase
shift.

2.2 Numerical model


The mathematical model discussed beforehand will be implemented numerically. This was done
using the MATLAB software. We first started by defining the different quantities related to our
study such as the wave frequency, the wave number... Part of this section is exposed hereafter:
##-------------------------------------------------------------------------------
%Inputs
a=1.25; % amplitude
lambda = 50; % wave length
h=39; % water Height
rho= 1000; % density
k=2*pi/lambda; % wave number
e=2*a/lambda; % steepness
w=sqrt(9.81*k*tanh(k*h)); % angular frequency
T=lambda*k/w; % period
##-------------------------------------------------------------------------------

We then moved on to defining the quantities related to our code such as the discretization step.
With that being done, we implemented our Euler scheme as follows.

##-------------------------------------------------------------------------------
for i = 1:numel(t)-1

fz1=cosh(k*(z(1,i)+h))/cosh(k*h);
f1z1=sinh(k*(z(1,i)+h))/cosh(k*h);
U = a*w*fz1*cos(k*x(1,i)-w*t(1,i)); %X_component velocity
W = a*w*f1z1*sin(k*x(1,i)-w*t(1,i)); %Y_component velocity

dt=t(i+1)-t(i); %Time step

x(1,i+1)=x(1,i)+(dt*U); %X_component position


z(1,i+1)=z(1,i)+(dt*W); %Y_component position

end
##-------------------------------------------------------------------------------

4
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

3. Results
This section of our work will expose the results of our study. Those latter are first: the generated
displacement field with a mean particle position, i.e. that does not depend on time, and the one
with a time-depending particle position. From those two results, the work about drift velocity will
be initiated. This quantity will be evaluated for different amplitudes of our wave model resulting
in different steepness levels. From that point, we will be able to quantify the mass transport
induced by the wave depending on its amplitude. We will also compare the numerical values of
this last quantity with those of a theoretical model and calculate the corresponding errors. The
errors induced by our code will be studied with respect to the amplitude of the wave, but also to
the number of particles followed. This will allow us to test the code’s convergence and evaluate its
overall performance. All of the results will be exposed for both intermediate and deep water cases.

3.1 Displacement fields


The first set of results generated through our study is the particles trajectories. Two separate
cases were assumed: a mean position, and a time-depending one. While velocities in the first one
are calculated using the mean position values, the ones in the second case are calculated using
current values. In what follows, the trajectories are discussed in the two different cases for both
intermediate and deep waters.

3.1.1 Deep water case


The code runs that were used to generate those results were done with a wavelength λ = 50m. The
deep water assumption is that Hwater
λ > 12 . We accordingly chose a water height of Hwater = 39m.
The amplitude of our wave is 2.5m, i.e. a steepness value of 10%. This configuration was used for
both the mean and time-depending cases.

Mean particle position


The trajectories generated by using velocities of our mean position are illustrated in figure 1.

Figure 1: time independent trajectories in deep waters

This result shows circular particle trajectories with a displacement amplitude that decreases,
in an exponential manner, as we go deeper in water until it fades out at the sea bottom. This
decrease is due to the f (z) term present in amplitudes in our mathematical model. As the mean
position results do not depend on time, the particles remain in their initial position; the trajectory
is then constant for each particle.

5
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Time-depending particle position


Figure 2 illustrates the particle trajectories generated through using the current position’s
velocities in our Euler scheme.

Figure 2: time dependent trajectories in deep waters

As opposed to the mean position case, the particle trajectories include a horizontal drifting
motion. The trajectories we get are of an helicoidal shape resulting from adding the mean-position
result and a horizontal movement one (drift). Dispersion is also noticed as the frequency increases
when we go deeper.
The result shown in figure 2 is similar to the first one as we can here also notice that the displace-
ment’s amplitude decreases in an exponential way as we go deeper in water due to the f (z) term
in amplitudes in our model. Particle velocity then fades out at sea bottom.

3.1.2 Intermediate water case


The condition for intermediate water depths is 20 1
< Hwater
λ < 12 . We chose a wavelength of
λ = 50m for the mean position case, and of λ = 5m for the time-depending one. Our code runs
were therefore respectively at water depths of Hwater = 9m and Hwater = 0.75m. We used a
steepness value of 5%, i.e. amplitude values of a = 1.25m and a = 0.125m respectively for the
mean position and the time-depending cases.

Mean particle position


The results generated by using the velocities of the mean particle position in intermediate
depths are shown in figure 3 .

6
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Figure 3: time independent trajectories in intermediate waters

Figure 3 shows elliptic trajectories with an amplitude that still exponentially decreases but
with a steeper slope as we go deeper in water. This decrease trend is close to a linear one. The
normal velocity also decreases as moving to deeper levels and reaches a zero value at sea bottom.
In this case, particles still move at the sea bottom but in a purely horizontal manner, i.e. with
zero normal velocity.

Time-depending particle position


When taking into account the current position’s velocity in our trajectory generation, the results
are as shown in figure 4.

Figure 4: time dependent trajectories in intermediate waters

Similarly to the deep water result, the particle trajectories include a horizontal drifting motion.
Those latter are also of an helicoidal shape resulting from adding the mean-position result and the
horizontal movement (drift). Dispersion is here also noticed as the frequency increases when we
go deeper.
The displacement’s amplitude decreases in an exponential way as we go deeper in water due to the
f (z) term. Particle velocities are however still non-zero at sea bottom (0.75m); the only zero term
is the normal one. Particle trajectories at sea bottom are not illustrated in the graph.

7
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Appearance of drifting motion

Figure 5: drifting motion

At a first sight, one can notice the horizontal movement of water particles for both cases i.e
deep and intermediate waters, especially when the movement of the particle is more realistic, which
means that it is considered as time dependent.

Synthesis and comparison between deep and intermediate depth cases


After elaborating particle trajectories in the mean position and in the time-depending one for
the intermediate and deep water levels, we can say that the main differences between the two
depths are that:
• At the sea floor, particle velocity completely vanishes for the deep level. For the intermediate,
the normal component of the velocity is the only one that vanishes. the particles in this case
still move at sea floor in a purely horizontal manner.
• The trajectories are circular for the deep level and elliptic for the intermediate one.

• The displacement’s amplitude decreases in a steeper way for the deep level. This result is
logical since dissipation of kinetic energy is higher in higher depths.
• The model wave frequency reaches higher values for the deep water case, which is purely
logical since the water depth is higher.

3.2 Drift velocity


As discussed above, the difference between the mean particle position case and the time-depending
one is that a drift motion appears in the second one. The wave simulated by our model indeed
induces a horizontal movement that affects particle trajectories. In this section of our work, the
results of quantifying the drift velocity will be exposed for several amplitudes of our model wave
for both the deep and intermediate water levels. The drift velocity was calculated numerically
using the following piece of code:
VIT(m)=(x(1,numel(x))-x(1,1))/(dt*numel(t)); %Drift velocity

with the variable m referring to a specific vertical position in the water column.

8
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

3.2.1 Deep water case


In the case of deep water, our study was performed with a model wave length of λ = 50m. As
Hwater
λ > 12 , the water height was taken to be Hwater = 39m. The drift velocity was then evaluated
for four different values of the amplitude, i.e. 1.25, 0.25, 0.125, and 0.025m meaning steepness
values of respectively 5, 1, 0.5 and 0.1%. The results are shown in figures 6 to 9.

Figure 6: Drift velocity as a function of depth Figure 7: Drift velocity as a function of depth
(5% steepness)-deep waters (1% steepness)-deep waters

Figure 8: Drift velocity as a function of depth ( Figure 9: Drift velocity as a function of depth
0.5 % steepness)-deep waters (0.1% steepness)-deep waters

Figures 6 to 9 show the numerical and theoretical drift velocities versus the water depth for four
different amplitudes. We first notice that the drift velocity decreases exponentially as we increase in
water height. The numerical value decreases for example for an amplitude of 0.125m (0.5% steep-
ness) approximately from 0.00215m/s at water surface to 0 at −22.5m. It is indeed the highest
at the water surface, and zero at the sea bottom. For the four amplitudes from highest to low-
est, maximum numerical drift velocities are approximately 0.14, 0.0078, 0.00215, and 0.000085m/s.

Moreover, we can notice that for an amplitude of 1.25m (5% steepness), the numerical and
theoretical curves diverge from each other. The numerical model is clearly not very accurate at
this amplitude. However, and as we decrease the amplitude, figures 6 to 9 clearly show how
the numerical solution converges towards the theoretical one. Our model indeed becomes more
accurate for decreasing steepness values which is coherent with the linearization hypothesis of our
BVP problem. At an amplitude of 0.025m (0.1% steepness), the numerical model produces a
solution that is very coherent with the theoretical model.

9
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

3.2.2 Intermediate water case


In order to quantify the influence of the water height condition on the drift velocity and on our
numerical model’s accuracy, the intermediate water level study was performed with the same wave
model at the same four values of amplitude, but at a different water height. The model wave
length is therefore here also λ = 50m, the amplitude values are 1.25, 0.25, 0.125, and 0.025m i.e.
respective steepness values of 5, 1, 0.5 and 0.1%. As an intermediate water level is characterized
by the fact that Hwater
λ > 12 , the water height was taken to be Hwater = 17m. The results are
exposed in figures 10 to 13.

Figure 10: Drift velocity as a function of depth Figure 11: Drift velocity as a function of depth
(5% steepness)-intermediate waters (1% steepness)-intermediate waters

Figure 12: Drift velocity as a function of depth Figure 13: Drift velocity as a function of depth
(0.5% steepness)-intermediate waters (0.1% steepness)-intermediate waters

Figures 10 to 13 expose the numerical and theoretical drift velocities with respect to the water
depth for four different amplitudes. Similarly to the deep water case, the drift velocity decreases
exponentially as we increase in water height. For example, the numerical value decreases for an
amplitude of 0.25m (1% steepness) approximately from 0.0075m/s at water surface to 0.0003 at
−17m. Here also, it is the highest at the water surface. It however does not reach zero at the sea
bottom. For the four amplitudes from highest to lowest, the maximum numerical drift velocities
are approximately 0.135, 0.0075, 0.002, and 0.000082m/s.

Furthermore, for an amplitude of 1.25m (5% steepness), the numerical and theoretical curves
diverge from each other as is the case for deep waters. The numerical model is clearly not very
accurate at this amplitude neither. However, and as we decrease the amplitude, figures 10 to 13

10
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

also show the convergence of the numerical solution towards the theoretical one. For intermediate
water levels too, the model becomes more accurate for decreasing steepness values which again is
coherent with the linearization hypothesis of our BVP problem.

Synthesis and comparison between deep and intermediate depth cases


One can infer from the graphs presented that the drifting velocities at the surface (z=0) are
not so different from the case of deep to intermediate waters, ≈ = 0.14 m/s for a steepness of 5%
in both cases. These velocities do not completely vanish at the sea bottom for intermediate water
depths, as the only component who does is the normal one. For the deep water case, velocities
completely vanish before the sea bottom level as expected.

Moreover, numerical results converge towards theoretical ones faster for the deep water case.
The gradient between the theoretical and numerical results for the velocity is indeed varying from
one case to the other for the same steepness values. Figure 13 shows that for a steepness of 0.1%
the numerical and theoretical results are overlapping. On the contrary, as exposed in figure 10,
we have a velocity gradient of a magnitude of 10−4 for the intermediate water case. Therefore, we
can say the accuracy of the numerical approximation is higher in the case of deep waters.

The other interesting remark is that the slope of the drift velocity as a function of the water
depth is slightly different from one case to another. That is to say that in the case of deep waters,
the slope is sharper at the low depths. For the 1% steepness cases, d(Drif t velocity)
d(water depth) ≈ 1.3 × 10
−4
for
deep waters. This value is different in the case of intermediate waters where d(Drif t velocity)
d(water depth) ≈ 10
−3

for the same steepness value 1%. One can infer from this comparison that for the deep water case,
the drift velocity decreases in a faster way as we go deeper in water as compared to the intermediate
water depth one, which is logical, since the kinetic energy dissipation is less important for lower
water depth levels.

3.3 Error analysis


Drift velocities generated by the model wave induce a transport of mass within the medium. This
mass transport quantity can be evaluated as the integral of the drift velocity on the water depth
from the free surface to the sea bottom, multiplied by the water density. This defines a starting
point, from which we can quantify the error generated by our code and how it converges. Therefore,
the percent difference between the numerical mass transport value and the theoretical one will be
calculated in order to define an error quantity related to our code. In what follows is the detail of
what was discussed. Numerically, the mass transport was evaluated as follows:

M_t_anlt=rho*trapz(VIT,z_m); %mass transport


with zm being a discretized variable vertically travelling through the water column, and VIT
being the drift velocity distribution through the same quantity. A trapezoidal rule is then used to
integrate the drift velocity on the water column.

3.3.1 Influence of amplitude


The steepness of the wave ϵ is a characteristic that is linearly related to the amplitude λ where :
2a
ϵ= λ

The following figure allows concretely reading the results presented afterwards, as it shows the
steepness ϵ as a function of the amplitude for deep and intermediate waters.

11
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Figure 14: Linearity of steepness as function of amplitude

The slope of this function is constant for both cases .Yet, and as discussed on the section 3.2.3
the steepness has a direct effect on the results for the different cases of water depth. In fact, it is
noticed on the previous section that the lower the steepness value is, the smaller the error will be
when comparing numerical results to theoretical ones.This effect is amplified when the depth of
the water increases till overlapping the numerical and the theoretical results. That is to say that
the amplitude has a direct influence on the convergence of the numerical results. Therefore we
can infer that, the smaller the wave amplitude is, the more accurate results one would have, and
of course, this statement is dependant on the water depth and it is more obvious for deep waters.
Figures 15 and 16 expose and detail the analysis of numerical error with respect to the amplitude,
but also to the number of followed particles for intermediate and deep waters.

Figure 15: error% as function of amplitude for deep waters

Figure 15 illustrates a set of calculated values of the percent difference between numerical and
theoretical mass transport values for amplitudes ranging from 0 to 1.25m, i.e. and according to
figure 14 steepness values that range from 0 to 5%. This code run was performed with a wavelength
λ = 50m at a water height of Hwater = 39m, with a time step equal to 1000 T
with T being our
model wave period. The total time duration is 7 periods, where 50 particles were followed and 50
different values of the error recorded. From Figure 15, we can see that for an amplitude a → 0, we
get a 1% error for the deep water case, that is to confirm the results presented for the theoretical
and numerical values of the drifting velocity as a function of water depth. Moreover, the error %

12
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

is quasi-linearly increasing with respect to the amplitude, as it approximately reaches 27% for an
amplitude of 1.2m.

Moving on to the intermediate depth case, the code was run with the same configuration
detailed above for the deep water case, but with a water height Hwater = 9m. Figure 16 illustrates
the results.

Figure 16: error% as function of amplitude for intermediate waters

From Figure 16, one can infer that for an amplitude a → 0 , we get around 36% error in the case
of intermediate waters, these results are in accordance with the previous statements and figures,
and it is explaining the gradient between the numerical and theoretical results for the velocity
in the intermediate water depths. When increasing the value of the amplitude we also notice a
quasi-linear increase of the error%.

Synthesis and comparison between deep and intermediate depth cases


As a synthesis of this section, we can say that the convergence of the numerical results depends
on the water depth and on the value of the steepness. Moreover, in both deep and intermediate
water levels, the error exhibits a quasi-linear trend with respect to the amplitude. Also, for the
steepness values used, 0% < ϵ < 5%, the code showed better convergence rates for deep waters
than for intermediate ones. In fact, for very low amplitudes and therefore steepness ones, the
model produced accurate results in the deep water case which goes inline with the linearization
hypothesis of our mathematical model. To sum up, we can say that our mathematical approach
and the code generated to model the wave problem are efficient at low steepness values for deep
water levels. They are however less efficient for intermediate waters, and for high steepness values.

A further analysis of our code has proven that another variable of our numerical model has a
significant influence on the obtained results; the number of followed particles. This relationship
will be studied in details in the next section.

3.3.2 Influence of number of particles


In the following, we will discuss the effect of the number of points used to discretize the water
depth, or number of followed particles, in order to study the variation of the error % since that it
was noticed that this also would have an effect on the accuracy of the obtained results for the case
of deep and intermediate waters as well.

In figure 17, we present the case of deep waters. The code was run with a wavelength λ = 50m,
an amplitude of 0.25m (1% steepness), a water height of Hwater = 39m, with a time step equal to

13
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

T
1000 with T being our model wave period. The total time duration is 7 periods. Error values were
recorded for 99 different numbers of followed particles, from 2 to 100.

Figure 17: error% as function of the number of the follower particles-deep waters

Figure 17 shows that the percent error drastically decreases as we increase the number of
followed particles. Very high errors are induced when following less than 5 particles; the error
is nearly 60% for N = 4. The error is then relatively small for values close to 8 particles. The
code then converges and error values stagnate after approximately 20 particles. The curve also
shows an inflection point at values near 10, where the error reaches its minimum value then slightly
re-increases. This is shown in a closer view in figure 18.

Figure 18: zoom on the error inflection point-deep waters

As discussed previously, the percent error reaches its minimum, a very close to zero value,
at N = 12 particles where we have an inflection point. The value then slightly increases, in a
logarithmic trend, as we start to see the beginning of stagnation that happens at N = 27 particles.
This inflection point can be due to code characteristics with respect to our mathematical model.
Its exact origin is however not completely known. Moreover, figure 19 shows a closer view of the
error curve after stagnation.

14
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Figure 19: zoom on the error stagnation line-deep waters

After N = 27 particles, the error of our code converges to a value of 7% and stagnates there as
shown in figure 19 for a high number of particles.

We then move on to the intermediate water case. An identical configuration, detailed above in
the deep water part, was used as to allow us a further comparison basis between the two water
levels. The water height used is Hwater = 21m. The results are shown in figure 19 below.

Figure 20: error% as function of the number of the follower particles-intermediate waters

Similarly to the deep water case, the error curve exhibits an initial very steep decrease for an
increasing number of followed particles. For N = 2, the percent error is more than 130% while it
drops for N = 3 to approximately 37%. The error curve then reaches it minimum value at around
7 particles where we have an inflection point. It slightly increases afterwards to reach relatively
low values, and finally stagnates at N = 25. Figure 21 shows a closer view of the inflection point
of our curve.

15
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Figure 21: zoom on the error inflection point-intermediate waters

As shown in figure 21 for the intermediate water level case, the percent error drops down to
its minimum, approximately 2%, at 7 particles and re-increases again in a logarithmic manner to
stagnate after N = 25. The stagnated error curve is shown in a closer view in figure 22 below.

Figure 22: zoom on the error stagnation line-deep waters

Similarly to the deep water case, and after N = 25, the percent error curve stagnates at a value
of 9% as shown in figure 22 for values greater than 65.

Synthesis and comparison between deep and intermediate depth cases


After fully establishing the error evolution with respect to the number of particles for the in-
termediate and deep water cases, we can say that the two sets of data exhibit the same trend with
a steep decrease to an inflection point followed by a logarithmic increase towards convergence.
However, our code shows a slightly faster convergence in the intermediate depth case. As an ex-
ample, in the deep water case, the error approximately drops from 345% to 125% from N = 2 to
N = 3 representing a 63.8% decrease. For the intermediate water case, the same values happen
are 135% to 35% from N = 2 to N = 3, i.e. 74% decrease which is higher than for the deep water.
The error curve also reaches its absolute minimum, i.e. the inflection point, at a lower number of
particles than for the deep water case, with the two coordinates being 12 for the deep and 7 for
the intermediate cases.

16
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

Moreover, the error curve stagnates for the deep water case at a lower value than for the in-
termediate one, with the two being respectively 7 and 9%. This confirms the previous result we
exposed in section 3.2 in the numerical versus theoretical analysis of the drift velocity for the two
depths. The mathematical model we used along with the code are indeed more accurate for deep
water levels.

17
Dakir, Es-sabar Final Project - Introduction to Hydrodynamics

4. Conclusion
The starting point of this work was the wave theory mathematical model that bases itself on a
fundamental assumption that is the linearization of the boundary value problem. The parameter
driving non linear effects, i.e. the steepness, was assumed to be much lower than 1. This assump-
tion allowed us to derive the final model on which we based our study.

This mathematical model was translated into a code that simulates particle trajectories in vari-
ous linearized wave configurations for two separate cases: a mean position and a time-depending
one. Comparing trajectories between the two cases underlined the fact that when using the current
position’s velocity in our Euler scheme, a drifting motion of the particles appears, which itself in-
duces a mass transport through the water medium. Calculating this latter quantity and comparing
it with a theoretical model allowed us to derive error values for our code, which will serve as an
evaluation basis for the assumptions made and for how the numerical results cohere with them.

When analyzing the evolution trend of the error with respect to the amplitude, we were able to
affirm that it is a quasi-linear relationship, but also that indeed; our study is only viable for small
values of the model wave steepness which is in complete coherence with our mathematical model’s
main assumption.

Moreover, model waves were studied in two different depth setups in order to test the viability
of the mathematical model with respect to the water depth, to test our entire study and its best
domain of use, but also to evaluate the sea floor influence on the flow. Results were in compliance
with theory:

• For deep waters, particle velocities completely vanished. However, for intermediate ones, the
only component that vanished is the normal one, i.e. the particles still moved in this case at
the sea floor but only purely horizontally.
• The dissipation of kinetic energy throughout the water column was observed, where the drift
velocity decreased faster when going deeper in water for the deep case than in the intermediate
one. This dissipation can also be connected to the sea floor which more realistically induces
a part of it.
• Drift velocities at the sea surface were found to be nearly the same for both depth situations,
which is logical since assuming the same model wave.

• The dissipation effect was also noticed in particle trajectories, where the wave frequency
increased for particles moving in deeper levels. The frequency logically reached higher values
in the deep water case than in the intermediate one.
• Trajectories shapes were in full accordance with the mathematical theory. They were closed
for a mean position situation, circular for deep waters and elliptic for intermediate ones.
From the results, the decrease in displacement’s amplitude was also found to be exponential
for deep waters, but more linear in intermediate ones.
When testing our numerical model for the two different depth setups, we were able to conclude
that it was more accurate in the case of deep waters. This main conclusion was drawn from evol-
utions of the drift velocity, but also from the mass transport error analysis. Our code however
exhibited a slightly higher numerical convergence for intermediate water depth.

In overall, our code demonstrated a satisfactory convergence trend along with a very acceptable
accuracy when inline with the mathematical model’s assumption, even if this accuracy was slightly
affected by the wave amplitude ’a’. The code is very versatile and flexible as it can generate results
for a multitude of configurations in terms of water depth and wave characteristics, and for different
orders of magnitude. It is still however a bit demanding in terms of resources when highly accurate
results are needed.

18

You might also like