International Journal of Offshore and Polar Engineering

Vol. 12, No. 1, March 2002 (ISSN 1053-5381)

Copyright © by The International Society of Offshore and Polar Engineers

A Theoretical Approach to the Slamming Impact Pressure

Acting on the VLFS
Ken Takagi
Department of Naval Architecture and Ocean Engineering
Osaka University, Suita, Japan

ABSTRACT

The matched asymptotic expansion technique is applied to a theoretical analysis of the slamming impact acting on the
VLFS. The theory demonstrates that the memory effect plays an important role in the evaluation of the relative wave
elevation under the bottom exposure. It also shows that the zero-draft approximation can be applicable to the estimation of
the slamming impact. Comparisons between the numerical results of the present theory and the experimental results show
that the theoretical value of impact pressure is in good agreement with the experimental values.

INTRODUCTION ear solution, which is smoothly connected to the ordinary impact

A conceptual design of the VLFS (Very Large Floating Struc-
ture) has been taking place in Japan. According to the initial plan
of this project, the VLFS has a thin plate configuration of very
large horizontal scale. Therefore the motion of the VLFS has a
different nature from that of a ship-like floating structure. Sev-
eral reliable numerical works on the calculation of motion of the
VLFS have been completed recently, and it was found that the
elastic motion of the VLFS seems like a propagation of water
waves beneath a thin elastic-plate. Hence only a limited area of
the bottom near the edge can be exposed, even in high seas. This
fact implies that the matched asymptotic expansion technique is
applicable to the estimation of slamming of the VLFS. In this
technique, the exposed region is considered as an inner region
that will be called the slamming inner region.
The zero-draft approximation is widely applied to the estima-
tion of motions of the VLFS, (e.g., Ohmatsu, 1997). Since this
approximation does not satisfy the body boundary condition at
the edge, the flow around the edge must be investigated when we
apply this approximation to the slamming theory. In the case of 2-
dimensional flow, Bessho and Komatsu (1974, 1976) showed that
the horizontal velocity at the edge has a logarithmic singularity.
This result implies that the flow around the edge is singular, and
an inner problem that will be called the edge inner problem (or
the edge inner region) is considered.
We have 2 inner regions to be considered in this paper, and
it seems to be complicated. When a slamming impact causes a
structural damage, the scale of exposure is much larger than the
theory.
A slamming theory that deals the slamming inner region and the
impact theory is discussed below, as are some numerical results.
Although the edge inner region has no direct relation to the present
slamming theory, it is discussed in Appendix A to justify the zero-
draft approximation.
SLAMMING OF VLFS
The elastic motion of the VLFS in waves seems like waves trav-
eling in the VLFS, whose length is of the same order as the wave-
length of incident waves. Thus, while the slamming occurs only
in the limited region along the edge of the VLFS, the scale of the
exposure area is much larger than the draft of the VLFS. Employ-
ing these facts, a simple slamming theory can be composed.
Relative Wave Elevation in Slamming Inner Region
Suppose a rectangular body floating in the infinitely deep water
as a simple 2-dimensional model of the VLFS. The edge is located
at the x = 0, and the body is floating in the domain x ≤ 0. Fig. 1
shows the coordinate system. It is assumed that the amplitude of
incident waves a is very small and of the same order as the draft
of the VLFS d:
a a
scale of draft. This fact suggests that the scale of the edge inner  1 = O1 (1)
 d
region can be assumed to be much smaller than the scale of the
slamming inner region. Then these 2 inner regions can be treated
separately, and the zero-draft approximation can be applicable to
the slamming inner region. Finally we can construct a simple lin-

Received September 1, 2001; revised manuscript received by the editors

December 12, 2001. The original version (prior to the final revised
manuscript) was presented at the 11th International Offshore and
Polar Engineering Conference (ISOPE-2001), Stavanger, Norway, June
17–22, 2001.
KEY WORDS: VLFS (Very Large Floating Structure), slamming, impact
pressure, matched asymptotic expansion method. Fig. 1 Coordinate system
10 A Theoretical Approach to the Slamming Impact Pressure Acting on the VLFS

where  denotes the wave length of incident waves which is cho- This equation implies that the bottom exposure starts at t̃ = −t˜0
sen as a typical length of scale of the outer solution. The follow- and finishes at t̃ = t˜0 , where t˜0 is given by:
ing nondimensionalizations are convenient for further analysis: 
2
  t̃0 =  (8)
x z t  = T 1 T2 32¯a 72
x̄ z̄ =   t̄ =  1 ḡ = g
  T a 
(2) Coordinates are stretched according to the scale of bottom
T2 2 h
p̄ = p 2¯ =  h̄ =  exposure, as stated previously:
a a a
x z
x̃ z̃ =  (9)
where x is the horizontal measure and the z is the vertical mea- l l
sure, t is the time, T is the period of the incident waves, 1 the
velocity potential, g the acceleration of the gravity, p the water The exact boundary conditions in this region are represented as
pressure, 2 the elevation of the free surface and h the displace- follows:
ment of the VLFS. It is noted that the nondimensionalization for 1 81 1  2
x z and t indicated here is used in the outer problem shown 42 +  9 1  + 4ḡ 2¯ = 0 ¯
on z̃ = 2 (10)
8 t̃ 2
in Appendix A, and another nondimensionalization for these vari-
1 8 2¯ 81 8 2¯ 8 1 
ables that is convenient in the slamming inner region will be 42 + − =0 ¯
on z̃ = 2 (11)
defined in this section. 8 t̃ 8 x̃ 8 x̃ 8 z̃
The relative wave elevation measured from the bottom of the 1 8 h̄ 81 8 h̄ 8 1 
VLFS in the exposure region is assumed to be small in this theory: 42 + − =0 on z̃ = h̄ (12)
8 t̃ 8 x̃ 8 x̃ 8 z̃

32¯a − d¯ = 4  1 (3) It is assumed that asymptotic expansions for the velocity poten-
tial, wave elevation and displacement of the VLFS have the fol-
lowing forms:
where 32¯a denotes the absolute value of the relative wave eleva-
tion 32¯∗ and d¯ = d/a. This assumption coincides with the above- 0 + 4 21 1
=1
1 2 + 4 23 1
1 + 41 4 + · · ·
3 + 41 (13)
mentioned assumption that the exposure length is much smaller
1 1
than the length of incident waves: 2¯ = 2˜0 + 4 2˜1 + 42˜2 + 4 2˜3 + · · ·
2 2 (14)
1 1
h̄ = h̃0 + 4 h̃1 + 4h̃2 + 4 h̃3 + · · ·
2 2 (15)
l = 4 (4)
The free-surface boundary conditions for each order are repre-
where l denotes the typical length of scale of the exposure and sented as follows:
is chosen as the length scale of another inner region, which will
0
81 81n
hereafter be called the slamming inner region. = 0 + ḡ 2˜n−1 = 0 on z̃ = 0 n = 1 2 3 (16)
When the exposure length is smaller than the order of amplitude 8 t̃ 8 t̃
of incident waves, the slamming impact can not be obtained. Thus, 0
81 8 2˜n−1 8 1 n
the following assumption is employed: = 0 − = 0 on z̃ = 0 n = 1 2 3 (17)
8 z̃ 8 t̃ 8 z̃
a  1
= = o 42 (5) The body surface condition for each order is also obtained:
l
0
81 n
8 h̃n−1 8 1
This assumption plays an important role in making the slam- = 0 − =0 on z̃ = 0 n = 1 2 3 (18)
8 z̃ 8 t̃ 8 z̃
ming inner solution simpler, as mentioned below. In order to jus-
tify this assumption, we have to consider the slamming inner solu- There is no forcing term for the 0th, 1st and 2nd order velocity
tion of the pressure up to O4; however, the lowest order of the potential. Hence, the 0th, 1st and 2nd order potential, the 0th
slamming inner solution of the pressure is O4 ln 4. If the 2nd
order of the slamming inner solution is ignored, the free-surface
condition of this region would be a rigid boundary condition, and
no impact pressure could be obtained. This is the case of very
small amplitude of incident waves. Therefore, the slamming inner
and 1st order relative wave elevation, and the 0th and 1st order
displacement are easily obtained, where the inner expansion of
the velocity potential is found in Appendix A:
0 = )1
1 ∗ 0 0* (19)
solution of the pressure up to O4 should be taken into account. 1 = −)1
1 ∗ 0 0*7t̃ (20)
The assumption Eq. 3 claims that the duration of exposure is  
very short compared to the period of incident waves. Thus the 2 = )U ∗ *x̃ + )1
∗ 0 0*72 z̃ 1 2
1 0 − t̃ (21)
measure of time is also stretched: ḡ 2
7 ∗
t 2˜0 = )1 0 0* (22)
t̃ = 4 − 21
(6) ḡ
T 7 ∗
2˜1 = )1 0 0*7t̃ (23)
ḡ
Ignoring higher-order terms, the relative wave elevation in the
slamming inner region can be approximated as follows: p̄∗ −0 7 ∗
h̃0 = − + )1 0 0* (24)
ḡ ḡ
 
¯ t̄ = 32¯a cos 7t̄ = 32¯a 1 − 1 472 t̃ 2 + O42 
32 (7) h̃1 =
7 ∗
)1 0 0*7t̃ (25)
2 ḡ
International Journal of Offshore and Polar Engineering 11

The 3rd order velocity potential and 2nd order wave elevation where the kernel is given as follows:
can be divided into 2 parts. The first part is matched with the
outer solution and the second part vanishes when x̃ approaches 8S z̃2 − x̃ − x̃ 2
= 2 At̃ − >
±infinity: 8 z̃ ?z̃ + x̃ − x̃ 2 @2
   
3 = ;0 + ;
1 (26) + k exp kz̃ cos kx̃ − x̃  cos kḡt̃ − > − 1 dk (39)
0
2˜2 = #0 + # (27)
Substituting Eq. 38 into Eq. 32 or Eq. 33, a Fredholm integral
equation of the 2nd kind is obtained. The solution is obtained by
The matching condition gives the following relations:
solving this integral equation together with the initial condition
1 ∗ 3
# = 0 at t̃ = 0. The numerical procedure to solve this integral
;0 = )1 0 0*73 t̃ equation is described in Appendix B.
6
 
7 ∗ z̃
+ p̄ −0 x̃ ln 4r ˜ − z̃ arctan 7t̃ Impact Phase
ḡ x̃
7 ∗ The partial differentiation of the velocity with respect to the
− )1 0 0*z̃72 t̃ − )U0∗ *x̃7t̃ (28) time represents the pressure of the fluid. But when the free sur-
ḡ
face touches the bottom surface, a strong singularity appears at
7 ∗ 2
#0 = − )1 0 0*72 t̃ the intersection between the two. This singularity makes the inte-
2ḡ
gral equation Eq. 38 difficult to solve. In order to remove this
72 ∗ 1 singularity, an approximation is introduced in which the memory
− p̄ −0x̃ ln 4x̃ + )U0∗ *x̃
ḡ 2  2ḡ
If the local structure of the VLFS were stiff enough, the influ-
ence of the slamming impact on the motion of the VLFS would
be a higher order. This fact is easily proved from the assumption
(29)
effect of the integral equation is neglected.
Suppose, a very short period At̃ which starts at t̃ = t˜1 when the
free surface touches the bottom. Eq. 37 can be approximated as
follows:
of the scale of exposure. Therefore, the 2nd order displacement ;A x̃ z̃ t̃1 + At̃ = ;x̃ z̃ t̃1 + At̃ − ;x̃ z̃ t̃1 
of the VLFS should be fully matched with the outer solution:
1  At̃  0
  =− p̃x̃  > + t̃1 Sx̃ x̃  z̃ −> d x̃ d>
p̄∗ −0 2 2 7 ∗ 2 d h̄∗ 0  1 −
h̃2 = − 7 t̃ − )1 00*72 t̃ + x̃ (30) 0
2ḡ 2ḡ d x̄ z̃ 2
= ;A x̃  0 t̃1 + At̃ 2 d x̃ + OAt̃  (40)
− z̃ + x̃ − x̃ 2
It is assumed that the condition of exposure is described by the
following boundary conditions: ;A satisfies the following free-surface boundary conditions which
were initially derived by Wagner (1932):
1
p̄x̃ t̃ + 4 2 p̃x̃ t̃ + ḡ d¯ ≥ 0 (31)
8# 8;
8; ;A = 0 =− A on z̃ = 0 (41)
=0 on z̃ = 0 if # ≥ 0 (32) 8 t̃ 8 z̃
8 z̃
8# 8; On the other hand, the following body-surface condition should
= on z̃ = 0 if # < 0 (33) be satisfied on the wetted surface:
8 t̃ 8 z̃
8;A 8#
where p̃x̃ t̃ is the pressure distribution acting on the bottom of =  on z̃ = 0 (42)
the VLFS due to the slamming. These conditions are applied in 8 z̃ 8 t̃ t̃=t̃1
the region x̃ < 0. On the contrary, in the region of positive x̃, the The solution is obtained by solving the integral equation Eq. 40
following free-surface conditions have to be satisfied: under the boundary condition Eq. 42. But, it is not easy to solve
Eq. 40 because of the singularity at the intersection between the
p̃x̃ t̃ = 0 (34) free surface and the body surface. Hence, a displacement potential
8# 8; C is employed, which was introduced by Korobkin (1982):
= on z̃ = 0 (35)
8 t̃ 8 z̃
 t̃
The flow which satisfies the above boundary conditions is rep- C= ;A d t̃ (43)
t̃1
resented by a pressure distribution on the line z̃ = 0. The velocity
potential of this flow satisfies a linear free-surface condition: The free-surface condition and the body-surface condition are
alternated as follows:
82 ; 8; 8 p̃  
2
+ ḡ =− on z̃ = 0 (36) 8C 8#
8 t̃ 8 z̃ 8 t̃ C + #t̃1  − t̃=t̃1 · t̃ − t̃1  = 0 on z̃ = 0
8 z̃ 8 t̃ (44)
The velocity potential and its derivative are represented by a
convolution integral: C≤0 on z̃ = 0

1  t̃  0 Since the displacement potential has no singularity at the inter-

;x̃ z̃ t̃ = − p̃x̃  >Sx̃ x̃  z̃ t̃ − > d x̃ d> (37) section point, the solution is easily obtained by the successive
 −t˜0 −
 over-relaxation method.
8;  1  t̃  0 8S The pressure distribution is obtained by differentiating the dis-
 = − lim p̃x̃  > x̃ x̃  z̃ t̃ − > d x̃ d> (38)
8 z̃ z̃=0 z̃→0  −t˜0 − 8 z̃ placement potential twice with respect to the time. It is difficult to
12 A Theoretical Approach to the Slamming Impact Pressure Acting on the VLFS

Fig. 2 Definitions of intersection points in slamming inner region

obtain a reliable result by this method, however, as the result con-

tains strong numerical noises. It is also difficult to investigate in
detail the flow near the intersection. Thus, the displacement poten-
tial is used only for obtaining the free-surface elevation. Once
we get the free-surface elevation, it is easy to obtain the velocity
potential by applying the Chebyshev polynomials.
It is defined that the intersection points are located at x̃1 , x̃2 , Fig. 3 Comparison between asymptotic value of pressure and
x̃3 , respectively, as shown in Fig. 2, where x̃0 is the end point numerical one
of the exposure. The velocity potential can be represented as a
distribution of vortex, where the kernel satisfies the impermissible
condition on the bottom surface of the domain x̃ < x˜0 : NUMERICAL RESULTS AND DISCUSSIONS
  x̃  x̃   First of all, we calculate the motion of the VLFS that is car-
8; 1 3 D − x̃0 ried out by a method shown in Yoshimoto et al. (1997). Then we
=− + D  dD (45)
8 z̃ x̃0 x̃2 x̃ − x̃0 x̃ − D use an approximation (Eq. 64) for expressing the outer domain
pressure distribution p̄ that is contained in Eq. 38 as the forcing
It is apparent from Eq. 45 that the potential has singularities at term through the free-surface condition (Eq. 31). It is noted that
the points x̃1 , x̃2 and x̃3 . Thus, an inner solution has to be consid- the approximation (Eq. 64) is practically convenient, although we
ered at these points. The inner solution of complex potential w is can obtain the pressure distribution when the motion of the VLFS
obtained by applying the theory by Armand and Cointe (1986): is computed, because all we have to know is the relative wave
 2  elevation at the edge of the VLFS when we use the approxima-
2 s 1 
wn = − x̃˙n An + ln s + + i  tion (Eq. 64). Thus, the comparison between the magnitude of
 2 2 2
 2  (46) the original pressure distribution and that of the approximation
˙ 2 s 5  (Eq. 64) is shown to demonstrate the accuracy of the approxima-
sgnx̃n x̃ − x̃n  + iz̃ = An − 2is − ln s + − i 
 2 2 2 tion. The original pressure distributions were computed by Yoshi-
moto et al. (1997); they also conducted an experimental work in
where n denotes the number of intersection points, and A is the the same paper. Numerical results of this paper are compared with
thickness of the spray. their experimental results.
The integral equation Eq. 45 is solved by employing the Cheby- Fig. 4 shows the numerical results of the relative wave eleva-
shev polynomials as follows: tion in the slamming inner region. The conditions of the compu-
 tation are adjusted to those of the experiment. It seems that the
x̃1 − x̃0 

cos n1
x̃ = −  1n x̃0 ≤ x̃ ≤ x̃1  motion of the free surface is not simple, because of an influence
 x̃ − x̃0 n=0 sin 1 of the memory effect. This effect is not taken into account in the
(47)
x̃ − x̃0 ordinary slamming theory for a ship or an offshore structure. But
2 − 1 = cos 1  the memory effect on the free-surface elevation plays an impor-
x̃1 − x̃0
tant role for an estimation of the slamming pressure acting on the
1 
cos n2 VLFS, because the slamming pressure acting on the VLFS is very
x̃ = −  x̃2 ≤ x̃ ≤ x̃3 
 n=0 2n sin 2 sensitive to a little change of free-surface shape and the speed of
(48)
x̃ − x̃2 the free-surface elevation.
2 − 1 = cos 2  Fig. 5 shows a numerical result of the speed of the free-surface
x̃3 − x̃2
elevation in the slamming inner region for a variation of the ampli-
where coefficients 1n and 2n are calculated by applying the tude of incident waves. These curves represent the speed of the
orthogonality of the Chebyshev polynomials. The thickness of free-surface elevation just before the free-surface hits the bottom
spray is obtained from the matching conditions: of the VLFS. These curves and the relative wave elevation shown
 2
 x̃1 
 
n
n−m
A1 = 1n −1 42 (49)
4 x̃˙ 2 n=1 m=1
1
 2
 x̃3 − x̃2 

A2 = 2n −1n 42 (50)
16 x̃˙ 2 n=1
2
 2
 x̃3 − x̃2 

n
A3 = 3n −1 42 (51)
16 x̃˙ 2 n=1
3

The composite solution is obtained to compose the inner solu-

tions and the outer solution as shown by Arman and Cointe
(1986), but it is not represented here, as the explicit form of the Fig. 4 Curves of relative wave elevation in slamming inner region
composite solution is too lengthy. for time interval of 3t̃ = 01
International Journal of Offshore and Polar Engineering
Fig. 5 Speed of free-surface elevation in slamming inner region
for variation of amplitude of incident waves
in Fig. 4 are used as inputs of the computation of the impact
phase.
Fig. 6 shows the time history of the pressure acting at a point
x̄ = −00256. The curve called modified represents the average of
the original pressure in a circle of diameter 12 mm which corre-
sponds to the diameter of pressure gauges used in the experiment.
The maximum value of the modified curve is smaller than that of
the original curve, since the spatial distribution of the pressure at
t̄ = 0012 is very peaky. This technique was initiated by Takemoto
and Tanizawa (1986), and they demonstrated an improvement of
the agreement between a theory and an experiment.
Fig. 7 shows time histories of the pressure at different positions.
It is apparent that a slight change of position makes for a big
variation in the maximum value of the pressure. This tendency
can be seen in the experimental result.
The maximum pressures at x̄ = −00256 for the variation of the
amplitude of incident waves are compared with those of exper-
imental results in Fig. 8. While the theoretical estimation is in
good agreement with the experimental results, it is very difficult to
get a good estimation because the maximum value of the impact
pressure is very sensitive to a small change of conditions.
CONCLUSIONS
The matched asymptotic expansion technique is applied to a
theoretical analysis of the slamming impact acting on the VLFS.
The theory suggests that the memory effect should be taken into
account for the evaluation of the relative wave elevation in the
slamming inner region, and the numerical result demonstrated that
the memory effect plays an important role in the calculation of the
slamming impact. It also showed that the zero-draft approximation
can be applicable to the outer problem as well as to the slamming
inner problem. Thus, the theory can be composed in a framework
Fig. 6 Time history of pressure acting at point x̄ = −00256 (a =
007 m, T = 10 s)
13
Fig. 7 Time history of pressure at different points: p − 1: x̄ =
−00256, p − 2: x̄ = −00449, p − 3: x̄ = −00641 (a = 007 m,
T = 10 s)
Fig. 8 Maximum pressures for validation of amplitude of incident
waves (T = 10 s)
of the linear theory and can be connected to the ordinary water-
impact theory to achieve a reliable numerical result.
The numerical results show that a slight change of the mea-
suring point makes for a big variation of the maximum value of
the pressure. This suggests that it is very difficult to get a good
estimation of the maximum impact pressure. However, compar-
isons between the numerical results of the present theory and the
experimental results show that the theoretical value of the impact
pressure is in good agreement with the experimental values.
REFERENCES
Armand, JL, and Cointe, R (1986). “Hydrodynamic Impact Anal-
ysis of a Cylinder,” Proc 5th Int Symp Offshore and Arctic Eng,
Tokyo, Vol 1, pp 250–256.
Bessho, M, and Komats, M (1974). “On Hydrodynamic Forces
Acting on a Flat Plate Oscillating on Water Surface,” J Kansai
Soc Naval Arch, Japan, Vol 154, pp 69–76 (in Japanese).
Bessho, M, and Komats, M (1976). “On Hydrodynamic Forces
Acting on a Flat Plate Oscillating on Water Surface (2nd
Rept),” J Kansai Soc Naval Arch, Japan, Vol 163, pp 67–74 (in
Japanese).
Korobkin, A (1982). “Formulation of Penetration Problem as a
Variational Inequality,” Din Sploshnoi Sredy, Vol 58, pp 73–79.
Ohmatsu, S (1997). “Numerical Calculation of Hydroelastic
Responses of Pontoon Type VLFS,” J Soc Naval Arch, Japan,
Vol 182, pp 329–340 (in Japanese).
Takemoto, H (1984). “Some Considerations on Water Impact
Pressure,” J Soc Naval Arch, Japan, Vol 156, pp 314–322 (in
Japanese).
14 A Theoretical Approach to the Slamming Impact Pressure Acting on the VLFS

Wagner, H (1932). “Über Stoss und Gleitvergänge an der Ober- velocity potential is obtained:
fläche von Flüssigkeiten,” Zeitschr. für Angewandte Mathematik  
2
und Mechanik, Vol 12, Heft 4, pp 193–215. ∗ 0 0 1 + 7 z̄
∗ x̄ z̄ = 1
1
Yoshimoto, M, Hoshino, K, Ohmatsu, S, and Ikebuchi, T (1997). ḡ
 
“Slamming Load Acting on a Very Large Floating Structure i7 ∗ z̄
with Shallow Draft,” J Marine Sci and Tech, pp 37–46. − p̄ −0 x̄ ln r¯ − z̄ arctan
ḡ x̄
+ U0∗ x̄ + or
¯ (61)
APPENDIX A
where r¯2 = x̄2 + z̄2 .
The 1st order velocity potential satisfies Laplace’s equation in
Substituting the asymptotic expansion of the velocity potential
the outer domain:
into the boundary conditions, the asymptotic expansion of wave
elevation, displacement of the VLFS and the pressure under the
2 1
9 =0 (52) VLFS are obtained, respectively:

The velocity potential satisfies the following boundary condi- i7 ∗ 72

2¯∗ x̄ = − 1 0 0 − 2 p̄∗ −0x̄ ln x̄
tions as well: ḡ ḡ 
i7 ∗
82 1  
81 − U0 x̄ + ox̄ (62)
+ ḡ =0 for x̄ > 0 z̄ = 0 (53) ḡ
8 t̄ 2 8 z̄
p̄∗ −0 i7 ∗ d h̄∗ 0
8 h̄ 8 1  h̄∗ x̄ = − − 1 0 0 + x̄ + ox̄ (63)
− =0 for x̄ ≤ 0 z̄ = 0 (54) ḡ ḡ d x̄
8 t̄ 8 z̄  
72
p̄∗ x̄ = p̄∗ −0 1 − x̄ ln x̄ + ox̄ (64)
It can be assumed that the 1st order solution of the outer prob- ḡ
lem is harmonic in time with a nondimensional angular frequency
7. Thus, we use the complex function f ∗ x̄ z̄ in the discussion It is apparent in the above expressions that the horizontal veloc-
of this problem instead of the real function f x̄ z̄ t̄: ity logarithmically approaches infinity as r approaches zero, as
mentioned previously. Therefore an inner expansion in which the
  draft of the VLFS can be recognized is considered:
f x̄ z̄ t̄ =  f ∗ x̄ z̄ exp i7t̄  (55)
 
x z
It is well known that the 1st order solution of the velocity x̂ ẑ =  (65)
d d
potential is represented by a distribution of pressure on z̄ = 0:
The coordinates are nondimensionalized by the draft of the
∗ i7  0 ∗  ∗ VLFS. The following asymptotic forms are assumed:
1 x̄ z̄ = − p¯ x S x̄ − x  z̄ dx  (56)
ḡ 
 = 10 + A ln A11 + · · · 
1 (66)
where:
2¯ = 2ˆ0 + A ln A2ˆ1 + · · ·  (67)
1   exp kz cos kx
S ∗ x z = lim dk (57) h̄ = ĥ0 + Aĥ1 + · · ·  (68)
→+0  0 k − K + i
where A = d/.
The following relation holds between the velocity potential and
Substituting these equations into the exact boundary conditions
the pressure under the VLFS:
and neglecting terms of oA ln A, boundary conditions for the 1st
order velocity potential 11 are obtained:
∗
81 i7
∗ +
−K 1 = − p¯∗ on z̄ = 0 (58)
8 z̄ ḡ 8 11
=0 on ẑ = 2ˆ0 (69)
8 ẑ
This implies that the pressure at x̄ z̄ = 0 0 is not continu-
ous, and the relative wave elevation of the water surface 32¯∗ at 8 11
=0 on the body surface. (70)
x̄ = 0 is obtained from the pressure at the edge of the VLFS: 8 n̂

p̄∗ −0 The flow satisfying these boundary conditions is a simple

32¯∗ = h̄∗ 0 − 2¯∗ 0 = − (59) step-wise flow which has no singularity. The matching condition
ḡ
implies that 2ˆ0 and ĥ0 are equal to the wave elevation and the dis-
placement of the VLFS at x̄ = 0, respectively. The 1st order wave
Because, Bessho and Komatsu (1976) derived that the hori-
elevation is calculated from the 1st order velocity potential:
zontal velocity at the edge is logarithmically singular, it can be
assumed that the pressure under the VLFS has the following form:
1 8 11
2ˆ1 = − on ẑ = 2ˆ0  (71)
ḡ 8 t̄
p̄∗ x̄ = p̄∗ −0 + p̄1∗ x̄ ln x̄ + Ox̄ (60)
When the assumption Eq. 1 is applied, the 1st order inner solu-
Substituting this pressure into the integral equation and neglect- tion is a higher-order term. Therefore, the relative wave height
ing higher-order terms, the following asymptotic expansion of the obtained from the outer solution is valid.
International Journal of Offshore and Polar Engineering 15

APPENDIX B time. The coefficients Hkl are given by:

A numerical procedure for solving the integral equation is Hkl = 4I)k3x l3t* − 2I)k + 13x l3t*
described in detail. The relative wave elevation is calculated by
performing the following integral. This integral is not easy since − 2I)k3x l + 13t* + I)k + 13x l + 13t*
the integration function does not converge when the time t̃ goes + I)k − 13x l + 13t* + I)k + 13x l − 13t*
to infinity. Therefore the integration function is partially evaluated
by an analytical integration: + I)k − 13x l − 13t* − 2I)k3x l − 13t*
− 2I)k − 13x l3t*/3x3t (77)

1  t̃  0 The function I)x t* is defined by the following integral and

#x̃ t̃ = − p̃x̃  >
 0 − can be evaluated analytically:

 k   1
 
× cos kx̃ − x̃  sin kḡt̃ − > dk d x̃ d> (72)
0 ḡ I)x t* = − sin kx cos kḡt dk
0 kḡ
" 3  2  2
2 4x 2 gt t2  gt
=− √ q0 + √ ḡxq1
This integral is discretized by the Galerkin method:  ḡ 3  ḡ 4x 3  4x
√   √ #
ḡ 1 ḡ
+ txq2 √ t + ḡt 3 q3 √ t
 t̃  0 2 x 6 2 x
#kl = #x̃  >G x̃  kG > l d x̃ d> (73)
" 3 #  2
0 − 1 1 1  √ 4 1 x2 ḡt
+ √ −i 2 ḡt 2 x + i 2 exp i
ḡ  3 3 ḡ 4x
   
where the shape function used in this paper G is a simple step 1 i 1 ḡ
+ − ḡt 3 + tx erf −i 2 √ t (78)
function: ḡ 6 2 x

where erf is the error function.

!
1 n − 13x ≤ x̃ ≤ n3x Functions q0 , q1 , q2 and q3 are represented by special functions:
G x̃ n = (74)
0 n − 13x ≥ x̃ or x̃ ≥ n3x
 
√ 3 3
q0 x = −  exp ix − i   ix  (79)
4 2
√
 
The relative wave elevation at x̃ t̃ = k3x l3t is represented   1
q1 x =  exp ix − i   ix  (80)
by the summation of coefficients Hkl : 2 4 2
$  
 1 1 5 x4
q2 x = x 1 F2 M  M−
2 4 2 4 4

N 
l
$  
#kl = − pnm Hk−nl−m (75) x 3
 3 3 7 x4
n=1 m=1 + x 1 F2 M  M−
3 2 4 2 4 4
 4

1 3 5 x
The vertical velocity at the free surface is computed by the for- − x4 2 F3  1M  M −  (81)
2 4 4 4
ward difference scheme in time to guarantee the numerical stabil- $  
  1 1 5 x4
ity: q3 x = − x 1 F2 M  M−
2 2 4 2 4 4
3
$  
x  3 3 7 x4
8; N  l
Hk−nl+1−m − Hk−nl−m + x F M  M−
=− pnm (76) 3 2 1 2 4 2 4 4
8 z̃ 3t 4
 4

n=1 m=1 x 5 7 x
− 2 F3 1 1M   2M − + ln 2x (82)
3 4 4 4
2
The finite difference scheme has a numerical error of O3t̃ . where  is the Gamma function and n Fm is the Hypergeometric
This error is of the same order as an error due to discretization in Function.