Applied Ocean Research 73 (2018) 127–140

Contents lists available at ScienceDirect

Applied Ocean Research

journal homepage: www.elsevier.com/locate/apor

Hogner model of wave interferences for farfield ship waves in shallow

water
Yi Zhu a , Huiyu Wu a , Jiayi He b , Chenliang Zhang b , Wei Li a , Francis Noblesse a,∗
a
State Key Laboratory of Ocean Engineering, School of Naval Architecture, Ocean & Civil Engineering, Collaborative Innovation Center for Advanced Ship
and Deep-Sea Exploration, Shanghai Jiao Tong University, Shanghai, China
b
Shanghai Key Laboratory of Ship Engineering, Marine Design and Research Institute of China, Shanghai, China

a r t i c l e i n f o a b s t r a c t

Article history: The remarkably simple yet realistic Hogner model of farfield ship waves, previously considered to analyze
Received 31 October 2017 the apparent wake angle associated with the highest waves that result from constructive interferences
Received in revised form among the divergent waves created by a fast ship in deep water, is applied to the more general case of
30 December 2017
uniform finite water depth. This theory is used to illustrate notable features of ship waves in finite water
Accepted 29 January 2018
Available online 21 February 2018
depth. In particular, numerical applications to a monohull ship and catamarans illustrate two notable
features of ship waves in shallow water: specifically, the apparent wake angle can be much smaller
than Havelock’s classical asymptote or cusp angles, and the apparent wake angle in finite water depth
Keywords:
Farfield ship waves is nearly identical to the apparent wake angle in deep water if the water depth is greater than the ship
Shallow water length. Moreover, the numerical illustrations demonstrate the paramount importance of interferences
Wave interferences among divergent waves for fast ships. Indeed, constructive interferences among the divergent waves
Divergent waves created by a fast ship largely determine the variation of the amplitude of the waves across the ship wake,
High speed and therefore the apparent wake angle and the wave drag of the ship, a critical element of ship design.
Hogner These numerical illustrations also corroborate main conclusions of an approximate analysis of wave
interferences in shallow water for monohull ships and catamarans modeled as 2-point wavemakers.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction where g denotes the gravitational acceleration. The nondimen-

The farfield waves created by a monohull ship of length L, or
a catamaran with identical twin hulls of length L separated by a
lateral distance S, that travels at a constant speed V in calm water of
uniform finite depth D and large horizontal extent are considered
within the usual framework of linear potential flow theory. This
classical theoretical framework is realistic, and the only practical
option, to analyze farfield ship waves. The highest waves associated
with constructive interferences among the divergent waves created
by the ship hull (predominantly by the bow and the stern of the
ship) at a high Froude number are of primary interest. The Froude
number F is defined as
V
F≡  (1)
gL
∗ Corresponding author.
E-mail address: fnob@sjtu.edu.cn (F. Noblesse).
sional water depth
Dg
d≡ (2)
V2
is also defined. Finite water depth effects are only significant if
d < d∞ ≈ 3 as is well known, and the water depth is then effectively
infinite if 3 < d.
The flow around the ship hull is modeled in accordance with the
classical Hogner flow model [1], i.e. via a distribution of sources and
sinks over the mean wetted hull surface ˙ of the ship. The source
density in Hogner’s model is equal to the component nx of the unit
vector n ≡ (nx , ny , nz ) normal to the ship hull surface ˙, where n
points into the water. This simple flow model explicitly defines the
flow around a ship hull in terms of the Froude number and the ship
hull geometry, and moreover defines farfield ship waves without
computations of the nearfield flow around the ship hull.
Hogner’s flow model has previously been used in [2–5] to ana-
lyze farfield waves created by fast ships in deep water. These studies
of farfield ship waves in deep water show that, despite its remark-
ably simplicity, the Hogner model is realistic and useful. The Hogner
theory of farfield ship waves in deep water is applied here to the
more general case of water of uniform finite depth. Moreover, the

https://doi.org/10.1016/j.apor.2018.01.016
0141-1187/© 2018 Elsevier Ltd. All rights reserved.
128 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140

analysis of the highest waves that result from constructive interfer- by the ship in terms of the water depth, the Froude number, and
ences among the divergent waves created by a fast ship considered the ship hull geometry.
in [2–5] for deep water is extended to shallow water of finite depth. The Green function G in (4) can be formally decomposed into
The analytical and numerical study of wave interference effects a wave component W and a non-oscillatory local flow component,
in shallow water, for monohull ships and catamarans modeled as is shown in, e.g. [14] for deep water. The local flow component
via Hogner’s hull-surface distributions of sources and sinks, that L decays rapidly and is ignored in the analysis of farfield waves
is considered here also extends the analytical studies of interfer- considered here. The wave component W (x̃, x) given in [14] for
ence effects in shallow water given in [6,7] for ships modeled as deep water becomes
2-point wavemakers. Specifically, a monohull ship is modeled as a  k∞
H(x − x̃) Ẽ+ E+ + Ẽ− E−
point source at the ship bow and a point sink at the stern, i.e. as a W (x̃, x) = Im  dk (5)
source-sink pair, and a catamaran is modeled as two point sources F 2 k0 t(k − t)
at the twin bows of the catamaran, i.e. as a pair of point sources,
for the more general case of finite water depth. Here, H(·) is the
in the elementary ship models previously considered in [6,7]. The
usual Heaviside unit-step function, F is the Froude number (1), Im
analysis of interference effects for elementary 2-point ship models
means that the imaginary part is considered, and k0 (d) is defined
considered in [6,7] for uniform water depth is considerably more
as
complex than the similar analysis of 2-point ship models consid-    
ered in [8,11] for deep water, and shows that wave interferences 0 d≤1
k0 (d) ≡ if (6)
are more complicated in shallow water than in deep water. root of k = t 1<d
Thus, the numerical analysis of wave interference effects for
farfield ship waves in shallow water considered here extends where
the analytical and numerical studies of wave interference effects t ≡ tanh(kd) (7)
already given in [2–7]. In particular, main conclusions of the analyt-
± ±
ical studies of basic 2-point ship models given in [6,7] are illustrated Moreover, Ẽ± ≡ Ẽ± (k, x̃) and E ≡ E (k, x) denote the basic wave
and corroborated by the more precise numerical analysis, based functions
on the Hogner hull-surface source distribution flow model, that is cosh[k(z̃ + d)] i(˛x̃±ˇỹ)
Ẽ± ≡ e (8a)
considered here. cosh(kd)

cosh[k(z/F 2 + d)] −i(˛x±ˇy)/F 2

2. Hogner model of ship waves in shallow water E± ≡ e (8b)
cosh(kd)
The waves created by the ship are observed from an orthogo- where
nal frame of reference and related coordinates (X, Y, Z) attached to  
˛≡ kt, ˇ ≡ k(k − t) and k0 ≤ k (9)
the ship. The Z axis is vertical and points upward, and the undis-
turbed free surface is taken as the plane Z = 0. The X axis is chosen The wave functions (8a) represent elementary waves that propa-
along the path of the ship and points toward the ship bow. The gate at angles  ± from the x-axis given by
nondimensional coordinates   
(X̃, Ỹ , Z̃)g X̃g t 1−t
x̃ ≡ (x̃, ỹ, z̃) ≡ ≡ 2 (3a) (cos ± , sin ± ) = ,± (10)
k k
V2 V
(X, Y, Z) X The finite limit of integration k∞ in (5) eliminates unrealistic short
x ≡ (x, y, z) ≡ ≡ (3b)
L L waves, notably waves influenced by surface tension and viscosity,
are used. Hereinafter, X̃ and x̃ denote a flow-field point located that correspond to k∞ < k.
within the flow region outside the mean wetted ship hull surface Expressions (4) and (5) define the potential W ≡ ˚W g/V3 asso-
˙, and X and x denote a point of ˙. The coordinates of the flow-field ciated with the waves aft of the ship as a superposition
point x̃ are nondimensional in terms of the reference length V2 /g
 k∞
1 A+ (k)Ẽ+ (k, x̃) + A− (k)Ẽ− (k, x̃)
associated with the ship speed V, as is appropriate for a farfield W (x̃) = Im  dk (11)
 k0 t(k − t)
analysis, whereas the coordinates of the hull-surface point x are
nondimensional with respect to the ship length L. of elementary wave functions Ẽ± (k, x̃) where the amplitude func-
The flow created by the ship, in water of uniform finite depth, tions A± (k, F) are given by the distributions
is evaluated in accordance with the Hogner flow model [1], as in 
[2–4] for ship waves in deep water. Hogner’s simple approxima- 1
A±
1
≡ 4 nx (x)E± (k, x)da(x) (12a)
tion is shown in [2] to predict wave patterns that can hardly be F ˙
distinguished from the wave patterns obtained via the more precise
of elementary wave functions E± (k, x) over the ship hull surface ˙.
Neumann–Michell theory given in [12,13].
Expression (12a) for a monohull ship becomes
Specifically, the Hogner approximation H to the flow potential

 ≡ ˚/(VL) at a flow-field point x̃ is given by ˇs

 A±
2
= 2 cos A±
1
(12b)
2F 2
H (x̃) = G(x̃, x)nx (x)da(x) (4)
˙ for a catamaran with identical twin hulls separated by a distance
s ≡ S/L.
where da(x) ≡ dA/L2 denotes the differential element of area at a
The free-surface elevation e ≡ E g/V2 associated with the waves
point x of the hull surface ˙, nx (x) is the x-component of the unit
is e(x̃, ỹ) = ∂W (x̃, ỹ, 0)/∂x̃. Expressions (11), (8a) and (9) then yield
vector n normal to ˙ at x as was already noted, and G ≡ G(x̃, x)  k∞ + −
represents the velocity potential of the flow created at a flow-field 1 A+ (k)eihϕ + A− (k)eihϕ
point x̃ by a unit source at a point x and is the Green function associ-
e(x̃, ỹ) = Re  dk (13)
 k0 1 − t/k
ated with the classical linearized free-surface boundary condition.
The Hogner approximation (4) explicitly defines the flow created where hϕ± ≡ ˛x̃ ± ˇỹ and Re means that the real part is considered.
 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140 129

The Cartesian coordinates x̃ and ỹ can be expressed as

x̃ = −h cos and ỹ = h sin (14a)

where the horizontal distance h and the angle are given by

 ỹ
h ≡ Hg/V 2 ≡ x̃2 + ỹ2 and tan ≡ (14b)
(−x̃)
The ray angle = 0 corresponds to the track x̃ < 0, ỹ = 0, z̃ = 0 of
the ship. The phase functions ϕ± in (13) are then given by

ϕ± ≡ ±ˇ sin − ˛ cos (15)

as readily follows from (14a).

3. Farfield stationary-phase approximation

In the far field 1  h, the dominant contributions to the wave

integral (13) stem from points where the phases ϕ± of the trigono-
±
metric functions eihϕ are stationary, as is well known. The relations
Fig. 1. Wavenumbers kT ( , d) and kD ( , d) associated with transverse waves (blue
(9) show that the phase functions ϕ± defined by (15) and their dashed lines) or divergent waves (red solid lines) for 0 ≤ ≤ cusp (d) in ‘deep water’
derivatives are given by d =∞ or d = 1.1 (top row), and wavenumber kD ( , d) associated with divergent waves
 for 0 ≤ ≤ asymp (d) in ‘shallow water’ d = 0.9 or d = 0.1 (bottom row). (For interpre-
ϕ± k tation of the references to color in this figure legend, the reader is referred to the
√ =± sin − cos (16a) web version of this article.)
kt t−1

k dϕ± 2k/t − (1 + ad )
2 =±  sin − (1 + ad ) cos (16b) 90
t dk k/t − 1
 2
Cusp and asymptote angles (°) 80
d
k d2 ϕ ± 4ktda (k/t − 1) − (1 − ad ) 70
k =± sin
t dk2 (k/t − 1)
3/2 (16c) 60 ψasymp ψcusp
d 2
+ [4ktda + (1 − ad ) ] cos 50
40
where 0 ≤ ad ≤ 1 is defined as
30
kd
ad ≡ 2 (17) 20
sinh(2kd)
10
Points of stationary phase are determined by the roots of the
equations dϕ± /dk = 0, i.e. the two equations 0
0 0.5 1 1.5 2 2.5 3
 V
k/t − 1 d
tan = for ϕ = ϕ+ (18a)
[2(k/t)/(1 + ad ) − 1]
Fig. 2. Asymptote angle asymp (d) for 0 ≤ d ≤ 1 and cusp angle cusp (d) for 1 < d ≤ 3.

k/t − 1
tan =− for ϕ = ϕ− (18b)
[2(k/t)/(1 + ad ) − 1]
Eq. (18a) for ϕ+ has no real root for −/2 ≤ ≤ 0. Similarly, Eq. (18b) For water depths d ≤ 1 (in ‘shallow’ water), the stationary-phase
for ϕ− has no real root for 0 ≤ ≤ /2. In view of the symmetry equation (18a) has a single real root kD ( , d) that corresponds to
about the plane y = 0 for the (usual) case of a ship hull with port divergent waves, whereas for water depths 1 < d (in ‘deep’ water),
and starboard symmetry considered here, only farfield waves in the this equation has two real roots kT ( , d) and kD ( , d) that are
upper half plane 0 ≤ ≤ /2 and the phase function ϕ+ are analyzed associated with two sets of waves called transverse and divergent
hereinafter. waves, as is well known; e.g. [15,16,6,7]. Moreover, if d ≤ 1, diver-
The stationary-phase relation (18a) yields gent waves exist for 0 ≤ ≤ asymp where the ray = asymp is an
 asymptote of the wave pattern; whereas if 1 < d, transverse and
[ k/t − 1(1 + ad ), 2k/t − (1 + ad )] divergent waves exist for 0 ≤ ≤ cusp where the ray = cusp is a
[sin , cos ]=
  cusp of the wave pattern.
2
k/t 4(k/t − 1) + (1 − ad ) The roots k( , d) of the stationary-phase equation (18) are
depicted in Fig. 1 for four water depths d. For d = 0.1 and d = 0.9,
These two relations and expressions (16a) and (16c) show that the
a single root kD ( , d) exists in 0 ≤ ≤ asymp , whereas two roots
phase function ϕ+ and its second derivative at a point of stationary
kD ( , d) and kT ( , d) exist in 0 ≤ ≤ cusp for d = 1.1 and d =∞.
phase are given by
The angle asymp (d) of the asymptotes of the wave pattern is
−k(1 − ad ) given by
ϕ+ =  (19a)
4(k/t − 1) + (1 − ad )
2

asymp = arcsin( d) where 0 ≤ d ≤ 1 (20)
d 2
d2 ϕ + 8ktda + (1 − ad ) [2 − (1 + ad )/(k/t − 1)]
=  (19b)
dk
2
2 One has 0 ≤ asymp (d) ≤ 90◦ . The asymptote angle asymp (d) is
k 4(k/t − 1) + (1 − ad ) depicted in Fig. 2 for 0 ≤ d ≤ 1.
130 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140

D / L = 0.4
Transverse Divergent
F = 0.50
d = 1.60

Fig. 3. Waterlines y = ±ŷ(x) defined by (28) with b = 0.35, s = 0.25 and b = 0.12.

D / L = 0.4
Transverse Divergent
The angle cusp (d) of the cusps of the wave pattern is determined F = 0.55
by the relation d = 1.32

kcusp /tcusp − 1
tan cusp = (21)
[2(kcusp /tcusp )/(1 + adcusp ) − 1]
D / L = 0.4
Transverse Divergent
Here, adcusp and tcusp denote the values of the functions ad and t F = 0.60
d = 1.11
defined by (17) and (7) at the root k = kcusp (d) of the equation

k(T 2 + 2tT
) = 2t 2 (T + T
) − T 2 (t − T/2) (22a)

where
D / L = 0.4
Transverse Divergent
T
≡ 2kd(kdt − 1)(1 − t 2 )
F = 0.65
T ≡ t + kd(1 − t 2 ) and (22b) d = 0.95

as is shown in [16]. One has K ≈ 19◦ 28

≤ ◦
cusp (d) < 90 for
1 < d ≤∞. The cusp angle cusp (d) is depicted in Fig. 2 for 1 < d ≤ 3.
The values ˛cusp (d) and ˇcusp (d) of the Fourier variables ˛ and ˇ at
D / L = 0.4
the cusps are given by (9) and (7) with k = kcusp (d). The wavenum- Transverse Divergent
F = 0.70
ber kcusp (d) and the corresponding Fourier variables ˛cusp (d) and d = 0.82
ˇcusp (d) vary within the ranges

3
0 < kcusp (d) ≤ for 1 < d ≤ ∞ (23a)
2 -80

3
0 < ˛cusp (d) ≤ ≈ 1.225 (23b) Fig. 4. Wave-amplitude function |a+ ( )| and related cusp, asymptote and apparent
2 wake angles for a monohull ship, at a given water depth D/L = 0.4, for a series of
 Froude numbers F = 0.5, 0.55, 0.6, 0.65, 0.7 and of related water depths d ≡ Dg/V2 = 1.6,
x
3 1.32, 1.11, 0.95, 0.82. The apparent wake angles diverg that correspond to the 2-point
0 < ˇcusp (d) ≤ ≈ 0.866 (23c) ship model considered in [6] are also marked.
2

Divergent and transverse waves correspond to kcusp ≤ k and to

k0 ≤ k ≤ kcusp where k0 is defined by (6). farfield ship waves across the ship wake 0 ≤ is then determined
In the far field 1  h, the wave integral (13) that defines the by the wave amplitude function
free-surface elevation can be evaluated analytically via the classical 
stationary-phase approximation. Specifically, one has |a+ ( )| = Q (k, d) A2r (k, F) + A2i (k, F) (25)
D
ih(ϕ+ +/4) T
Re[aD
+e + H(d − 1)aT+ eih(ϕ+ −/4) ] where k = kD or k = kT , Ar and Ai denote the real and imaginary parts
e(x̃, ỹ) ≈  √ (24a)
of the function A+ (k, F), and Q(k, d) is defined by (24c) in which t
/2 h

and ϕ+ (k, d) are given by (7), (19b) and (17).

where H(·) is the usual Heaviside unit-step function, and ϕ+ D and ϕT
+
+
denote the phase function ϕ (k, d) defined by (19a) for the divergent 4. Simple family of hull forms
or transverse waves k = kD and, if 1 < d, k = kT . The amplitudes aD
+ and
aT+ of the divergent or transverse waves are defined in terms of the The simple case of a hull surface ˙ that has a flat bottom at a
amplitude function A+ (k) in (13) as constant draft and rectangular framelines
a+ (k, d, F) ≡ Q (k, d)A+ (k, F) (24b) y = ±ŷ(x) where − 0.5 ≤ x ≤ 0.5 and − ı ≤ z ≤ 0 (26)
where is considered. Here, ı denotes the draft/length ratio. Expressions
1 (12a) and (8b) then yield
Q (k, d) ≡ 

(24c)
(1 − t/k)|ϕ+ (k, d)|
A+
1
= Ax Az (27a)
and k is taken as the root or k = kD k = kT
of the stationary-phase where Ax and Az are defined as
equation (18a).  0.5
The farfield stationary-phase approximations (24) assume that −2 dŷ 2
kD is smaller than the upper limit k∞ associated with very short
x
A = 2 cos(ˇŷ/F 2 )e−i˛x/F dx (27b)
F −0.5
dx
waves affected by surface tension and viscosity, as is the case in
the numerical illustrations considered further on.
 0
1 cosh[k(z/F 2 + d)]
This equation shows that the phase function ϕ+ (k, d) in (24a), Az = dz

F2 cosh(kd)
its second derivative ϕ+ (k, d) in (24c) and the amplitude function  −ı
(27c)
+
A (k, F) in (24b) are functions of the ray angle with 0 ≤ ≤ asymp t[1 − cosh(kı/F 2 )] + sinh(kı/F 2 )
=
if d ≤ 1 or 0 ≤ ≤ cusp if 1 < d. The variation of the amplitudes of k
 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140 131

Fig. 5. Wave patterns that correspond to the wave-amplitude function |a+ ( )| depicted in Fig. 4 for Froude numbers F = 0.5, 0.55, 0.6, 0.65. The apparent wake angles max
associated with the first peaks of the function |a+ ( )| are shown as solid ray lines, and the cusps or asymptotes of the Havelock wake are marked as (blue or red) dashed ray
lines. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The local half beam ŷ(x) in (26) and (27b) is chosen as
3(x−0.5)/b
2ŷ(x) = b[1 − e ] for 0 ≤ x ≤ 0.5
−3(x+0.5)/s
2ŷ(x) = b[1 − e ] for − 0.5 ≤ x ≤ 0
where b ≡ B/L denotes the beam/length ratio and b < 0.5 and s
< 0.5
are the effective (nondimensional) lengths of the bow and stern
regions. The relations (28) yield
b
ŷ(x) ≈ for s − 0.5 < x < 0.5 − b
2 marans with identical twin hulls, defined in the previous section.
and the range s − 0.5 < x < 0.5 − b corresponds to a nearly parallel
midship region of length m = 1 − b − s . Fig. 3 depicts the water-
lines (28) with b = 0.12, b = 0.35 and s = 0.25.
The numerical illustrations for a monohull ship and catamarans
considered in the next section assume a draft/length ratio ı = 0.04
and bow and stern regions of lengths b = 0.35 and s = 0.25, for both
the monohull ship and the catamarans. The beam/length ratio b is
chosen as b = 0.12 for the monohull ship or b = 0.08 for the twin
hulls of the catamarans. The separation distance S/L between the
identical twin hulls of the catamarans is taken equal to 0.35, except
in Fig. 14 where narrow and wide catamarans with hull-spacings
S/L = 0.2 or 0.5 are considered.
(28a)
(28b) 5. Illustrative applications
Main features of farfield ship waves, notably the asymptote
angle asymp if d ≤ 1 or the cusp angle cusp if 1 < d, and the appar-
ent wake angle max associated with a peak of the wave-amplitude
(28c) function |a+ ( )|, are now illustrated for a monohull ship and cata-
Wave patterns are also shown for a broad set of Froude numbers F
and water depths D/L.
The wave-amplitude function |a+ ( )|, associated with the
stationary-phase approximation, is determined via the relations
(25) and (24c), and wave patterns are determined via numerical
integration of the wave integral (13). Consistency between these
numerical computations of wave patterns and the apparent wake
angle max determined via the stationary-phase approximation can
be verified in the figures that depict wave patterns, where the
apparent wake angle max and the asymptote angle or the cusp
132 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140
Fig. 6. Wave-amplitude function |a+ ( )| and related cusp, asymptote and apparent
wake angles for a monohull ship in ‘deep water’ d ≡ Dg/V2 = 1.07 (top two rows) and
in ‘shallow water’ d = 0.73 (bottom two rows). The water depth D/L is equal to 0.36
x
(top and 3rd rows) or 0.81 (bottom and 2nd rows). The apparent wake angles diverg
that correspond to the 2-point ship model considered in [6] are also marked.
angle cusp are marked. For the numerical illustrations considered
here, the amplitude function A in (25) and (13) is given by (27) and
(28), as well as (12b) for the catamarans.
Expression (25) for the wave-amplitude function |a+ ( )| is asso-
ciated with the classical stationary-phase approximation, as was
already noted, and is therefore not valid in the vicinity of the cusp
of |a+ ( )| at angles
= cusp , where ϕ+ = 0 and |a+ ( )| is unbounded. This well-known
limitation of the stationary-phase approximation is apparent in the
figures that depict the function |a+ ( )| considered below. How-
ever, these figures show that the region where the approximation
(25) is not valid is extremely narrow and has no appreciable effect
on the peaks of the amplitude function |a+ ( )| that determine the
apparent wake angle max of a ship, except at relatively low Froude
numbers for which the apparent wake angle max is close to the
cusp angle cusp .
5.1. Monohull ship
Fig. 4 considers the influence of the Froude number F, for five
Froude numbers 0.5 ≤ F ≤ 0.7, at a given water depth D/L = 0.4. The
water depths d ≡ Dg/V2 that correspond to the Froude numbers
F = 0.5, 0.55, 0.6 are greater than 1; whereas one has d < 1 for
F = 0.65 and 0.7. Consequently (and in agreement with a classical
basic result of the theory of ship waves in shallow water), Fig. 4
only depicts divergent waves for 0 ≤ ≤ asymp if F = 0.7 and 0.65,
but depicts both divergent waves for 0 ≤ ≤ cusp and transverse
waves for − cusp ≤ ≤ 0 if F = 0.6, 0.55 and 0.5. Divergent waves
are depicted for 0 ≤ and transverse waves are shown for ≤ 0
if 1 < d in Fig. 4, and in all the subsequent figures that depict the
wave-amplitude function |a+ ( )|.
Fig. 4 shows that the first peak (associated with primary inter-
ference) of the wave-amplitude function |a+ ( )| is dominant and
relatively sharp for F = 0.7 and 0.65, but is not as sharp and not dom-
inant at lower Froude numbers F = 0.6 and 0.55, and indeed cannot
be observed at F = 0.5. Moreover, the apparent wake angle associ-
ated with the first peak of the amplitude function |a+ ( )| is much
smaller than the asymptote angle asymp for F = 0.7 and 0.65, and
is also smaller than the cusp angle cusp for F = 0.6 and 0.55, but
cannot be distinguished from the cusp angle for F = 0.5.
These numerical results agree with the analysis of the 2-point
wavemaker model of a monohull ship given in [6]. Indeed, Ref. [6]
shows that the apparent wake angle max associated with the high-
est divergent waves is smaller than the asymptote angle asymp for
every Froude number F in ‘shallow’ water d < 1. However, in ‘deep’
water 1 < d, Ref. [6] shows that one has max < cusp for sufficiently
high Froude numbers, and max = cusp if F is not high enough.
The apparent wake angle divergx predicted by the analysis of the
elementary 2-point wavemaker model of a monohull ship given in
[6] is marked in Fig. 4. The distance x between the point source
and the point sink in the vicinity of the bow and the stern of the
ship in this 2-point ship model is chosen here in accordance with
the analytical relations (27) obtained in [11] for deep water (this
‘deep-water’ relation for the optimal distance between the point
source and the point sink in the 2-point model of a monohull is
a reasonable approximation that, for lack of better knowledge, is
applied here to finite water depth). Fig. 4 shows that the appar-
ent wake angles diverg x and max associated with the elementary
2-point ship model considered in [6] and the more precise Hogner
hull-surface source-distribution model considered here are rela-
tively close, in accordance with the results of the analysis given in
[11] for deep water.
Fig. 4 shows that for the relatively low Froude numbers F = 0.55
and F = 0.6, the second peak of the amplitude function |a+ ( )| is
higher than, or about as high as, the first peak associated with pri-
mary interference. This feature, which can also be observed in Fig. 6
for F = 0.58 (top row), shows that the determination of the appar-
ent wake angle max related to the ray angle = max where the
highest waves are found can be a relatively complicated issue that
involves some degree of uncertainty and arbitrariness, as is noted
in [2,4] for deep water. In particular, Fig. 4 (and the other figures in
the numerical illustrations considered below) show that the peaks
smaller than the first peak are narrow and
closely spaced. Whether these narrow peaks would stand out in
observations of ship wakes, or are less likely to be observed than
the broader first inner peak, is not a-priori obvious.
In any case however, the first peak provides an upper bound
for the apparent wake angle max . Moreover, the wave elevations
depicted in Fig. 5 for F = 0.6 and 0.65 suggest that the dominant
waves in fact appear to be better correlated with the apparent
wake angle max (marked as a solid ray line in Fig. 5) defined as
the first peak of the wave-amplitude function |a+ ( )| than with the
additional narrow closely-spaced peaks of |a+ ( )|.
Fig. 5 also shows that the dominant waves are found in the vicin-
ity of the cusps of the wake F = 0.5, or within the sector bounded by
the cusps and the angles defined by the first peaks of the function
|a+ ( )| for F = 0.55, and to some extent also for F = 0.6, in agreement
with the behavior of |a+ ( )| in Fig. 4.
Furthermore, Figs. 4 and 5 clearly illustrate the close relationship
that exists between the wave pattern and the amplitude function
|a+ ( )|, used here to determine the apparent wake angle associ-
ated with the highest waves that result from interference effects
on farfield ship waves in finite water depth.
In particular, the highest waves in the wave pattern depicted
in Fig. 5 are consistent with the rays, marked in the figure, that
correspond to the apparent wake angles determined from the
 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140 133

Fig. 7. Wave patterns that correspond to the wave-amplitude function |a+ ( )| depicted in Fig. 6. The apparent wake angles max associated with the first peaks of the function
|a+ ( )| are shown as solid ray lines, and the cusps or asymptotes of the Havelock wake are marked as (blue or red) dashed ray lines. (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.)

wave-amplitude function |a+ ( )| considered in Fig. 4. This con-
sistency provides a verification of the stationary-phase analysis
associated with Fig. 4 as well as the numerical evaluation of the
wave integral (13) that is used to determine the wave patterns
depicted in Fig. 5. The wave patterns and the apparent wakes angles
max associated with the wave-amplitude function that are consid-
ered hereafter can all be observed to be consistent.
The wave-amplitude function |a+ ( )| depicted in the top three
rows of Fig. 4, which correspond to water depths 1 < d, is not con-
tinuous across the cusps, where it is singular as was already noted.
This discontinuity and singularity, associated with the method of
stationary phase, do not appear in the wave patterns depicted in
Fig. 5 because wave patterns are determined via the integral rep-
resentation (13) rather than the stationary-phase approximation
(24a).
Fig. 6 considers a ‘deep-water’ case d ≡ Dg/V2 = 1.07 and a ‘shal-
low water’ case d = 0.73. The top two rows of the figure correspond
to the same water depth d = 1.07, and the cusp angle cusp therefore
is the same for these top two rows of the figure. However, the appar-
ent wake angle max associated with the first peak of the amplitude
function |a+ ( )| is smaller in the second row than in the top row
because the Froude number F is higher in the second row than in
the top row. Similarly, the the asymptote angle asymp is the same
in the bottom two rows of Fig. 6 as these two rows correspond to
the same water depth d = 0.73, but the apparent wake angle max is
smaller in the bottom row than in the third row because the Froude
number F is higher in the bottom row than in the third row.
Fig. 6 shows that the apparent wake angle max associated with
the Hogner model is relatively close to the apparent wake angle
x predicted by the analysis of the elementary 2-point wave-
diverg
maker model of a monohull ship given in [6], as was also found in
Fig. 4. The distance x between the point source and the point sink
in the 2-point ship model used to determine the apparent wake
x
angle diverg marked in Fig. 6 is chosen in the manner previously
explained in connection with Fig. 4.
Fig. 6, especially the bottom three rows of the figure, and the
corresponding wave patterns depicted in Fig. 7 provide a vivid
illustration of the facts that the apparent wake angle max can
clearly stand out well inside the asymptote angle asymp or the
134 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140

e e

F=1 F = 0.55
Transverse Divergent Transverse Divergent
D / L=0.9, d=2.98
D / L = 0.2
D/L=d=∞
d = 0.20

F = 0.80
Transverse Divergent
F=1 D / L=0.9, d=1.41
Transverse Divergent
D/L=d=∞
D / L = 0.5
d = 0.50

F = 1.10
Transverse Divergent
D / L=0.9, d=0.74
F=1 D/L=d=∞
Transverse Divergent
D / L = 1.1
d = 1.10

F = 1.50
Transverse Divergent
D / L=0.9, d=0.40
D/L=d=∞
F=1
Transverse Divergent
D/L=∞
d=∞
0
-7

Fig. 9. Wave-amplitude function |a+ ( )| and related cusp, asymptote and apparent
-50 wake angles for a monohull ship, at two given water depths D/L = 0.9 and D/L =∞,
for Froude numbers F = 0.55, 0.8, 1.1, 1.5 and (for D/L = 0.9) related water depths
d ≡ Dg/V2 = 2.98, 1.41, 0.74, 0.4.
Fig. 8. Wave-amplitude function |a+ ( )| and related cusp, asymptote and apparent
wake angles for a monohull ship, at a given Froude number F = 1, for water depths
D/L = 0.2, 0.5, 1.1 and ∞.

cusp angle cusp , as was previously shown in [6] via an analysis

D / L = 0.4
of the farfield waves created by a monohull ship modeled as a 2- F = 0.58
Transverse Divergent

point wavemaker. Indeed, Fig. 7 illustrates how the appearance of d = 1.19

a ship wake can be greatly influenced by interferences among the

divergent waves created by the ship.
Moreover, for F = 0.87, the second row of Fig. 6 and the corre-
D / L = 0.4
sponding wave pattern depicted in the upper right corner of Fig. 7 F = 0.60
Transverse Divergent

illustrate the well-known fact that, in ‘deep’ water 1 < d for which d = 1.11

transverse waves exist, divergent waves can be dominant even at

moderately high Froude numbers. This notable fact is also illus-
trated in several of the figures that are considered subsequently,
notably in the bottom two rows of Fig. 8 at a Froude number F = 1. D / L = 0.4
Transverse Divergent
F = 0.63
Fig. 8 illustrates the influence of the water depth D/L at a given d = 1.01
Froude number F = 1. Specifically, the water depth D/L varies within
the wide range 0.2 ≤ D/L ≤∞ in the figure. The related water depth
d likewise is different in the four rows of Fig. 8. Specifically, one has
d = 0.2 or 0.5 in the top two rows and d = 1.1 or ∞ in the bottom two D / L = 0.4
Transverse Divergent
F = 0.68
rows. Accordingly, the asymptote angle asymp in the top two rows d = 0.87
(for ‘shallow water’) and the cusp angle cusp in the bottom two
rows (for ‘deep water’) are different.
Interestingly however, the apparent wake angle max is very
nearly identical in all four rows. This result illustrates a main con- D / L = 0.4
Transverse Divergent
clusion of the analysis given in [6], namely that the apparent wake F = 0.75
d = 0.71
angle max is not appreciably influenced by the water depth D/L for
Froude numbers Fdeep (D/L) < F where the Froude number Fdeep (D/L)
is defined in [6]. The water depth D/L has no appreciable influence
on the apparent wake angle max in Fig. 8 because the condition
Fdeep (D/L) < F is satisfied for all the water depths D/L considered in
this figure. Fig. 10. Wave-amplitude function |a+ ( )| and related cusp, asymptote and apparent
wake angles for a catamaran with hull-spacing S/L = 0.35, at a given water depth
Another notable conclusion of the analysis given in [6] is illus-
D/L = 0.4, for a series of Froude numbers F = 0.58, 0.6, 0.63, 0.68, 0.75 and of related
trated in Fig. 9, where the wave-amplitude function |a+ ( )| is water depths d ≡ Dg/V2 = 1.19, 1.11, 1.01, 0.87, 0.71.
considered for both the infinite and finite water depths D/L = ∞ or
D/L = 0.9 at four Froude numbers F = 0.55, 0.8, 1.1, 1.5. The top row
 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140 135

Fig. 11. Wave patterns that correspond to the wave-amplitude function |a+ ( )| depicted in Fig. 10 for F = 0.58, 0.6, 0.68, 0.75. The apparent wake angles max associated with
the inner and outer peaks of the function |a+ ( )| are shown as solid black ray lines, and the cusps or asymptotes of the Havelock wake are marked as (blue or red) dashed ray
lines. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

for F = 0.55 corresponds to a water depth d = 2.98 that is nearly equal
to 3 (effectively infinite), and the function |a+ ( )| and the related
cusp angles cusp for D/L =∞ and D/L = 0.9 therefore cannot be dis-
tinguished in this case. One has d = 1.41, 0.74, 0.4 in the second,
third and fourth rows, and the cusp angle cusp for d = 1.4, as well
as the asymptote angles asymp for d = 0.74 and d = 0.4, accordingly
are significantly different for D/L = 0.9 and for D/L =∞.
Interestingly however, Fig. 9 shows that the wave-amplitude
function |a+ ( )| and the related apparent wake angle max are
very nearly identical inside the wake that corresponds to D/L = 0.9.
This result agrees with the analysis given in [6], which shows that
the apparent wake angle max is not appreciably influenced by
the water depth D/L if D/L is greater than approximately 0.9. This
conclusion is of practical interest for experimental observations of
the farfield Kelvin wakes of fast ships in deep water, as is further
explained in the conclusion.
Figs. 4–9 illustrate the classical result that transverse waves do
not exist in ‘shallow’ water d ≤ 1, and also show that divergent
waves dominate transverse waves in ‘deep’ water even at moder-
ately high Froude numbers, as has already been noted. Moreover,
Figs. 4–9 show that the highest waves associated with construc-
tive interferences among divergent waves created (predominantly)
by the bow and the stern of a fast ship can be strongly dominant.
This fact is clearly apparent from the dominance of the first peak of
the amplitude function |a+ ( )| in Fig. 8 (especially the second and
third rows of the figure), the bottom row of Fig. 6 and the bottom
two rows of Fig. 9. These numerical results (and related wave pat-
terns) show that the analysis of interference effects on farfield ship
waves is also of practical relevance for the design of high-speed
ships, as is noted in [3] for deep water and is further explained in
the conclusion.
5.2. Catamarans
The analytical and numerical studies of wave-interference
effects on the farfield waves created by a fast catamaran in deep
136 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140
e r r interferences, which are associated with the inner peak, increase
Transverse Divergent
d = 1.03
D / L = 0.20
F = 0.44
d = 1.03
Transverse Divergent
D / L = 0.45
F = 0.66
d = 0.56
Transverse Divergent
D / L = 0.20
F = 0.60
d = 0.56
Transverse Divergent
D / L = 0.45
F = 0.90
Fig. 12. Wave-amplitude function |a+ ( )| and related cusp, asymptote and appar-
ent wake angles for a catamaran with hull-spacing S/L = 0.35 in ‘deep water’
d ≡ Dg/V2 = 1.03 (top two rows) and in ‘shallow water’ d = 0.56 (bottom two rows).
The water depth D/L is equal to 0.2 (top and 3rd rows) or 0.45 (bottom and 2nd
rows).
water given in [3,4,11] show that interferences of divergent waves
are more complicated for catamarans than for monohull ships,
notably because apparent wake angles associated with two peaks
– called inner and outer peaks – of the wave-amplitude function
|a+ ( )| can be defined for catamarans. Moreover, these previous
studies of wave interferences for fast catamarans in deep water
show that the inner peak is largely due to lateral interferences
between the twin hulls of the catamaran, whereas the outer peak
is influenced by both lateral interferences between the twin hulls
and longitudinal interferences between the bow and stern waves.
Refs. [3,4,11] also show that lateral interferences among the
divergent waves created by the twin hulls of the catamaran mostly
dominate longitudinal interferences between the bow and stern
waves for large hull spacings S/L and/or high Froude numbers F. As
a result, the inner peak is dominant for ‘wide and/or fast’ catama-
rans. However, the situation is different and more complicated for
‘narrow and slow’ catamarans, i.e. for small hull spacings S/L and
(relatively) low Froude numbers F, for which longitudinal interfer-
ences are important.
Fig. 10 illustrates the influence of the Froude number F, for five
Froude numbers 0.58 ≤ F ≤ 0.75, at a given water depth D/L = 0.4.
The water depths d that correspond to the Froude numbers F = 0.58,
0.6, 0.63 are greater than 1, whereas one has d < 1 for F = 0.68
and 0.75. Accordingly, Fig. 10 only depicts divergent waves for
0 ≤ ≤ asymp if F = 0.75 or 0.68, but depicts both divergent waves
for 0 ≤ ≤ cusp and transverse waves for − cusp ≤ ≤ 0 if F = 0.63,
0.6, 0.58. Fig. 10 shows that the outer peak of the wave-amplitude
function |a+ ( )| is much broader than the inner peak, and can-
not be identified for F = 0.58. The figure also shows that the inner
peak becomes higher and more important as the Froude number
increases, and indeed is higher than the outer peak for F = 0.68 and
F = 0.75. These numerical results agree with the analysis given in
[4,11] for deep water and illustrate the general finding that lateral
as F increases.
Fig. 11 depicts the wave patterns that correspond to the wave-
amplitude function |a+ ( )| considered in Fig. 10. Fig. 11 shows that
the dominant waves are close to the cusps of the Havelock wake for
F = 0.58, and are approximately aligned with the outer peak of the
function |a+ ( )| for F = 0.6, in accordance with Fig. 10. For F = 0.68
and especially for F = 0.75, strong waves can be observed in Fig. 11
along the rays that correspond to the inner peak of |a+ ( )|, also in
agreement with Fig. 10.
Fig. 12 considers a ‘deep-water’ case d ≡ Dg/V2 = 1.03 and a ‘shal-
low water’ case d = 0.56. The top two rows of the figure correspond
to the same water depth d = 1.03, and the cusp angle cusp there-
fore is the same in these two rows. However, the inner wake angle
associated with the inner peak of the amplitude function |a+ ( )| is
smaller in the second row than in the top row because the Froude
number F is higher in the second row than in the top row. Moreover,
the outer wake angle, marked in the second row, cannot be iden-
tified in the top row. Similarly, the asymptote angle asymp is the
same for the bottom two rows of Fig. 12 because these two rows cor-
respond to the same water depth d = 0.56, but the inner and outer
wake angles are (much) smaller in the bottom row than in the third
row because the Froude number F is higher in the bottom row than
in the third row.
The wave patterns that correspond to the wave-amplitude func-
tion |a+ ( )| considered in Fig. 12 are depicted in Fig. 13. The upper
left corner of that figure shows large dominant waves in the vicinity
of the cusps of the Havelock wake and short waves along the rays
that correspond to the inner peak of the function |a+ ( )|. The upper
right and lower left corners of the figure shows strong waves along
the rays that correspond to both the outer and the inner peaks of
|a+ ( )|. Finally, the lower right corner of the figure shows dominant
waves along the rays associated with the inner peak of |a+ ( )|. The
wave patterns depicted in Fig. 13 are consistent with the wave-
amplitude function |a+ ( )| considered in Fig. 12 as expected.
Figs. 12 and 13 also show that the inner and outer wake angles
can both be much smaller than the asymptote angle asymp or the
cusp angle cusp , and that the inner peak and the related waves can
be dominant for high Froude numbers, as was already noted.
Fig. 14 depicts the wave-amplitude function |a+ ( )| for a ‘nar-
row’ catamaran and a ‘wide’ catamaran with hull-spacings S/L = 0.2
or S/L = 0.5 to illustrate the influence of the water depth D/L on lat-
eral interferences between the twin hulls of a catamaran at a high
Froude F = 1. Specifically, the water depth D/L varies within the wide
range 0.1 ≤ D/L ≤∞ in Fig. 14. The related water depths d are differ-
ent in the five rows of the figure. Specifically, one has d = 0.1, 0.15 or
0.25 (which corresponds to ‘shallow water’) in the top three rows
and d = 1.1 or ∞ (which corresponds to ‘deep water’) in the bottom
two rows. Accordingly, the asymptote angle asymp in the top three
rows and the cusp angle cusp in the bottom two rows are different.
Fig. 14 shows that, for both the narrow and wide catamarans, the
inner peak of the function |a+ ( )| is nearly identical to the apparent
y
wake angle diverg associated with lateral interferences between
two point sources in the 2-point wavemaker model of a catamaran
analyzed in [7]. This result illustrates the fact that the inner peak
of the amplitude function |a+ ( )| is predominantly associated with
lateral interferences between the twin hulls of the catamaran for
the relatively high Froude number F = 1 considered in Fig. 14.
The figure also shows that, for the narrow catamaran, the inner
peak is found at the same ray angle for all five water depths
0.1 ≤ D/L ≤∞. For the wide catamaran, the inner peak is also found
at the same ray angle for the three water depths D/L = 0.25, 1.1
and ∞, but is found at slightly different ray angles for the water
depths D/L = 0.1, 0.15 and 0.25. These numerical results illustrate
the conclusion, given in [7], that water-depth effects on lateral
 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140 137

Fig. 13. Wave patterns that correspond to the wave-amplitude function |a+ ( )| depicted in Fig. 12. The apparent wake angles max associated with the inner and outer peaks
of the function |a+ ( )| are shown as solid black ray lines, and the cusps or asymptotes of the Havelock wake are marked as (blue or red) dashed ray lines. (For interpretation
of the references to color in this figure legend, the reader is referred to the web version of this article.)

interferences between the twin hulls of a catamaran are negligi-
ble if 0.47 < D/S but may not be ignored if D/S < 0.47. Indeed, one
has 0.5 ≤ D/S for the narrow catamaran in all the cases considered
in Fig. 14. For the wide catamaran, one also has 0.5 ≤ D/S in the cases
considered in the bottom three rows but one has D/S ≤ 0.5 in the
top three rows.
Moreover, Fig. 14 shows that the first inner peak dominates the
outer peak, and is located well inside the asymptote or cusp angle
at a high Froude number F = 1. This result is especially clear for the
wide catamaran considered in the right column of the figure, and
can also be observed for the narrow catamaran considered on the
left column. Finally, Fig. 14 illustrates the influence of the hull-
spacing S/L on interference effects, as is now explained. The five
rows correspond to water depths d ≡ Dg/V2 = 0.1, 0.15, 0.25, 1.1, ∞,
and the water depths are the same for the narrow and wide cata-
marans. Accordingly, the asymptote or cusp angles are the same for
the narrow and wide catamarans in every water depth d. However,
the inner and outer peaks are different because they depend on the
hull spacing S/L.
Fig. 15 compares the wave-amplitude functions |a+ ( )| that cor-
respond to a catamaran with hull-spacing S/L = 0.35 in deep water
D/L =∞ or in finite water depth D/L = 0.9 for four Froude numbers
F = 0.55, 0.8, 1.1 and 1.5. The top row for F = 0.55 corresponds to
a water depth d = 2.98 that is almost equal to 3 (effectively infi-
nite). The function |a+ ( )| and the related cusp angles cusp for
D/L =∞ and D/L = 0.9 therefore are indistinguishable in this row.
One has d = 1.41, 0.74, 0.4 in the second, third and fourth rows, and
the cusp angle cusp for d = 1.4, as well as the asymptotes angles
asymp for d = 0.74 and d = 0.4, accordingly are significantly different
for D/L = 0.9 and D/L =∞. However, the wave-amplitude functions
|a+ ( )| for D/L = 0.9 and D/L =∞ are almost identical inside the wake
that corresponds to D/L = 0.9. This result, also illustrated for a mono-
hull ship in Fig. 9, agrees with the conclusion given in [7] that the
138 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140

Cusp Asymptote Inner Outer ψydiverg S / L = 0.2 S / L = 0.5

0.06 F=1 F=1

Transverse Divergent Transverse Divergent
D / L = 0.10, D / S = 0.50 D / L = 0.10, D / S = 0.2
0.04 d = 0.10 d = 0.10
|a+|

0.02

0
0.08
F=1 F=1
Transverse Divergent Transverse Divergent
0.06 D / L = 0.15, D / S = 0.75 D / L = 0.15, D / S = 0.3
d = 0.15 d = 0.15
+
|a | 0.04

0.02

0.08 F=1 F=1

Transverse Divergent Transverse Divergent
D / L = 0.25, D / S = 1.25 D / L = 0.25, D / S = 0.5
0.06
d = 0.25 d = 0.25
+
|a | 0.04

0.02

0
F=1 F=1
0.08 Transverse Divergent Transverse Divergent
D / L = 1.10, D / S = 5.50 D / L = 1.10, D / S = 2.2
0.06 d = 1.10 d = 1.10
+
|a |
0.04

0.02

0.08 F=1 F=1

Transverse Divergent Transverse Divergent
D/L=D/S=∞ D/L=D/S=∞
0.06
d=∞ d=∞
|a+| 0.04

0.02

0
-50 -40 -30 -20 -10 0 10 20 30 40 50 -50 -40 -30 -20 -10 0 10 20 30 40 50
ψ ψ

Fig. 14. Wave-amplitude function |a+ ( )| and related cusp, asymptote and apparent wake angles for narrow and wide catamarans with hull-spacings S/L = 0.2 (left side) or
y
0.5 (right side), at a given Froude number F = 1, for water depths D/L = 0.1, 0.15, 0.25, 1.1 and ∞. The apparent wake angles diverg that correspond to the 2-point model of a
catamaran considered in [7] are also marked.

e e e 6. Conclusion
F = 0.55
Transverse Divergent
D / L=0.9, d=2.98
D/L= d= ∞
The remarkably simple yet realistic Hogner model of farfield
ship waves, previously considered in [2–5] to analyze the appar-
ent wake angle associated with the highest waves that result
from constructive interferences among the divergent waves cre-
ated (predominantly) by the bow and the stern of a fast ship in
F = 0.80
Transverse Divergent
D / L=0.9, d=1.41 deep water, has been applied here to the more general and more
D/L= d= ∞ complicated case of uniform finite water depth.
The Hogner model provides a particularly simple way of evalu-
ating the farfield waves created by arbitrary monohull ships and
catamarans at any given Froude number F in water of uniform
F = 1.10
Transverse Divergent finite depth. Specifically, the farfield waves are determined explic-
D / L=0.9, d=0.74
itly, without the need for nearfield flow computations, in terms of
D/L= d= ∞
the water depth D, the speed V and the length L of the ship, and
the shape of the ship hull, via the Fourier–Kochin representation
defined by (13), (12), (8b), (9), (7), (6), (15) and (14b). This model
has been used here to determine the wave patterns depicted in
F = 1.50
Transverse Divergent Figs. 5, 7, 11 and 13.
D / L=0.9, d=0.40
D/L= d= ∞ The apparent wake angle max associated with the highest
waves due to interferences among the divergent waves created a
fast ship can be determined from the Hogner model in an even
0 simpler manner than the wave pattern via the relations (25), (24c),
-70
(19b), (17), (12), (8b), (9), (7) and (6). This simple method, based
on numerical identification of the peaks of the wave-amplitude
Fig. 15. Wave-amplitude function |a+ ( )| and related cusp, asymptote and apparent
function |a+ ( )| related to the stationary-phase approximation, has
wake angles for a catamaran with hull-spacing S/L = 0.35, at two given water depths
D/L = 0.9 and D/L =∞, for Froude numbers F = 0.55, 0.8, 1.1, 1.5 and (for D/L = 0.9) been used here in Figs. 4, 6, 8, 9, 10, 12 10, 14 and 15. The same
related water depths d = 2.98, 1.41, 0.74, 0.4. method was previously used, for the simpler case of deep water, to
analyze the variation of the apparent wake angle with respect to the
Froude number F and (for catamarans) the hull spacing S/L, as well
apparent wake angles in finite water depth and in deep water are
as the hull shape, via parametric studies [2,4]. The numerical anal-
nearly identical for water depths D/L greater than about 0.9.
 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140
ysis considered in [2,4] could in principle be extended to the more
general case of water of finite depth D, although this additional
parameter significantly increases the dimension of the parametric
analysis.
The rays that correspond to the apparent wake angle max (as
well as the asymptote angle or the cusp angle) are marked in the
wave patterns depicted in Figs. 5, 7, 11 and 13. The rays associ-
ated with the apparent wake angle max and the highest waves in
the wave patterns depicted in Figs. 5, 7, 11 and 13 are consistent.
This consistency provides a verification of both the method used
to determine max (via identification of the peaks of the wave-
amplitude function |a+ ( )| associated with the stationary-phase
analysis) and the numerical integration of the wave integral (13)
used to compute the wave patterns.
The numerical examples considered in the study illustrate and
corroborate the analysis of wave interferences in shallow water
given in [6] and [7] for monohull ships and catamarans modeled as
2-point wavemakers.
In particular, Figs. 4 and 6 show that the apparent wake angle
x predicted by the analysis of the elementary 2-point wave-
diverg
maker model of a monohull ship (a point source near the ship bow
and a point sink near the ship stern) given in [6] and the apparent
wake angle max determined from the more precise Hogner hull-
surface source-distribution model are relatively close. This finding
is consistent with the fact that lateral interferences are ignored
and longitudinal interferences are approximately represented in
the 2-point model of a monohull ship considered in [6,11].
y by a fast ship largely determine the variation of the amplitude of the
Similarly, Fig. 14 shows that the apparent wake angle diverg
predicted by the analysis of the elementary 2-point wavemaker
model of a catamaran (a pair of point sources at the twin bows
of the ship) considered in [7] and the apparent wake angle max
determined from the Hogner hull-surface model are quite close.
This finding is consistent with the fact that lateral interferences
between the twin hulls of a catamaran are dominant, except for
slow and narrow catamarans [4,11].
The consistency that can be observed in Figs. 4 and 6 between
the apparent wake angles max and divergx , and the close agreement
y peak and can therefore be reduced by reducing the magnitude of the
between max and displayed in Fig. 14, provide a verification
diverg
of the analysis given here. The variations of the wake angles
y
and diverg
with respect to the Froude number and the water depth
are depicted in Fig. 9 of [6] and Fig. 3 of [7].
The Hogner theory of ship waves is primarily used here to
illustrate important and interesting features of the farfield waves
created by monohull ships and catamarans for several Froude num-
bers F and water depths d ≡ Dg/V2 and D/L. These features of a ship
wake include the cusp angle cusp for water depths 1 < d or the
asymptote angle asymp for d ≤ 1, and the apparent wake angle max
related to peaks of the wave-amplitude function |a+ ( )|.
Several notable conclusions can be drawn from the examples
considered in the study. In particular, these examples show that the
apparent wake angle associated with constructive wave interfer-
ences can clearly stand out and be much smaller than the Havelock
asymptote or cusp angles [15] based on the 1-point ship model,
commonly used since Kelvin’s classical analysis of ship waves in
deep water [17].
Another interesting conclusion is that, for a monohull ship, the
apparent wake angle max is not appreciably influenced by the
water depth D/L for Froude numbers Fdeep (D/L) < F where the Froude
number Fdeep (D/L) is defined in [6].
For catamarans, inner and outer peaks of the wave-amplitude
function (WAF) can be defined. The inner peak predominantly
stems from lateral interferences between the twin hulls of the
catamaran, whereas the outer peak is influenced both by lateral
interferences between the twin hulls and longitudinal interfer-
139
ences between the bow and stern waves. Moreover, the inner peak
of the WAF becomes more important, and can dominate the outer
peak of the WAF, as the Froude number increases and/or as the spac-
ing S/L between the twin hulls of the catamaran increases, i.e. for
fast and/or wide catamarans. Indeed, the 2-point wavemaker model
of a catamaran considered in [7], which can only account for lat-
eral interferences between the divergent waves created by the twin
bows (or sterns) of the catamaran, is accurate for fast and/or wide
catamarans, but a 4-point model is required for catamarans that are
narrow and (relatively) slow as is shown in [11]. Yet another conclu-
sion for catamarans is that at high Froude numbers, the inner peak
of the WAF occurs at a ray angle that is not appreciably influenced
by the water depth D/S if 0.47 < D/S.
An important conclusion is that the apparent wake angle max in
finite water depth is nearly identical to the apparent wake angle in
deep water if the water depth D is greater than the ship length L, or
more precisely if 0.9L < D. This conclusion is useful for experimen-
tal observations of the farfield Kelvin wakes of fast ships in deep
water. Indeed, although observations of narrow ship wakes have
been widely reported in the literature [18–23], reliable systematic
observations are missing and much needed. The analysis given in
[6,7] and the more precise Hogner model considered here show
that such experimental observations can be performed by towing
ship models of length L in a water tank of depth D if 0.9L < D.
Moreover, the examples considered here clearly illustrate the
importance of interferences among divergent waves for fast ships.
Indeed, constructive interferences among divergent waves created
waves across the ship wake, and consequently the apparent wake
angle as well as the wave drag of the ship. Thus, the analysis of
constructive interference effects on farfield ship waves considered
in [2–11] and here is of practical relevance for the design of high-
speed ships.
Indeed, the fact that constructive interferences among divergent
waves can result in a strongly dominant peak of the wave-
amplitude function, as in Figs. 4–9 for a monohull ship, means that
the wave drag of a fast ship is largely determined by this dominant
peak, i.e. by reducing the intensity of the constructive interferences
x
diverg that occur between the bow wave and the stern wave.
As is shown in [11], the amplitudes of the bow and stern waves
have no influence on the occurrence of constructive interferences
but do affect the intensity of interferences, which are strong if the
amplitudes of the bow and stern waves are commensurate or weak
otherwise. The height of the peak of the wave-amplitude function
and the wave drag therefore are minimized if the magnitudes of
the bow and stern waves are quite different, as for common fast
ships with transom sterns. Thus, the analysis of wave interferences
considered here is consistent with the common design practice for
fast ships, and indeed provides a simple theoretical justification for
this practice.
References
[1] E. Hogner, Hydromech., Probl. d. Schiffsantriebs, Herausgeg v. Kempf u. E.
Foerster, Hamburg, 1932, pp. 99–114.
[2] C. Zhang, J. He, Y. Zhu, C.-J. Yang, W. Li, Y. Zhu, M. Lin, F. Noblesse, Interference
effects on the Kelvin wake of a monohull ship represented via a continuous
distribution of sources, Eur. J. Mech. B Fluids 51 (2015) 27–38.
[3] Y. Zhu, H. Wu, C. Ma, J. He, W. Li, D. Wan, F. Noblesse, Michell and Hogner
models of farfield ship waves, Appl. Ocean Res. 68 (2017) 194–203.
[4] J. He, C. Zhang, Y. Zhu, L. Zou, W. Li, F. Noblesse, Interference effects on the
Kelvin wake of a catamaran represented via a hull-surface distribution of
sources, Eur. J. Mech. B Fluids 56 (2016) 1–12.
[5] F. Noblesse, C. Zhang, J. He, Y. Zhu, C.J. Yang, W. Li, Observations and
computations of narrow Kelvin ship wakes, J. Ocean Eng. Sci. 1 (2016) 52–65.
[6] Y. Zhu, J. He, C. Zhang, H. Wu, D. Wan, R. Zhu, F. Noblesse, Farfield waves
created by a monohull ship in shallow water, Eur. J. Mech. B Fluids 49 (2015)
226–234.
140 Y. Zhu et al. / Applied Ocean Research 73 (2018) 127–140
[7] Y. Zhu, C. Ma, H. Wu, J. He, C. Zhang, W. LI, F. Noblesse, Farfield waves created
by a catamaran in shallow water, Eur. J. Mech. B Fluids 59 (2016) 197–204.
[8] F. Noblesse, J. He, Y. Zhu, L. Hong, C. Zhang, R. Zhu, C. Yang, Why can ship
wakes appear narrower than Kelvin’s angle? Eur. J. Mech. B Fluids 46 (2014)
164–171.
[9] J. He, C. Zhang, Y. Zhu, H. Wu, C.J. Yang, F. Noblesse, X. Gu, W. Li, Comparison
of three simple models of Kelvin’s ship wake, Eur. J Mech. B Fluids 49 (2015)
12–19.
[10] C. Ma, Y. Zhu, J. He, C. Zhang, H. Wu, W. Li, F. Noblesse, Wavelengths of the
highest waves created by fast monohull ships or catamarans, Ocean Eng. 113
(2016) 208–214.
[11] Y. Zhu, J. He, H. Wu, W. Li, F. Noblesse, G. Delhommeau, Elementary ship
models and farfield waves, Eur. J Mech. B Fluids 67 (2018) 231–241.
[12] F. Noblesse, F. Huang, C. Yang, The Neumann–Michell theory of ship waves, J.
Eng. Math. 79 (1) (2013) 51–71.
[13] F. Huang, C. Yang, F. Noblesse, Numerical implementation and validation of
the Neumann–Michell theory of ship waves, Eur. J. Mech. B Fluids 42 (2013)
47–68.
[14] F. Noblesse, Alternative integral representations for the Green function of the
theory of ship wave resistance, J. Eng. Math. 15 (1981) 241–265.
[15] T.H. Havelock, The Propagation of Groups of Waves in Dispersive Media, With
Applications to Waves on Water Produced by a Travelling Disturbance, Proc.
Royal Society, London, 1908, pp. 89.
[16] F. Noblesse, C. Yang, Elementary water waves, J. Eng. Math. 59 (2007)
277–299.
[17] W. Thomson, On ship waves, Proc. Inst. Mech. Eng. 38 (1887) 409–434.
[18] D.W. Taylor, Resistance of Ships and Screw Propulsion, Macmillan, 1910.
[19] G.S. Baker, Ship Form, Resistance and Screw Propulsion, D. Van Nostrand
Company, 1915.
[20] W.H. Munk, Ships from space, Proc. R. Soc. Lond. A 412 (1987) 231–254.
[21] E.D. Brown, S.B. Buchsbaum, R.E. Hall, J.P. Penhune, K.F. Schmitt, Observations
of a nonlinear solitary wave packet in the Kelvin wake of a ship, J. Fluid Mech.
204 (1989) 263–293.
[22] A.M. Reed, J.H. Milgram, Ship wakes and their radar images, Annu. Rev. Fluid
Mech. 34 (2002) 469–502.
[23] M. Rabaud, F. Moisy, Ship wakes: Kelvin or Mach angle? Phys. Rev. Lett. 110
(2013) 214503.