Water-entry and -exit of a horizontal circular

cylinder

MARTIN GREENHOW

Institute o f Marine Hydrodynamics, University of Trondheim, N-7034 Trondheinl, Norway.

and
Tile Norwegian Marine Technology Research btsititute A[S, Otto Nielsens vei 10, N-7017
Trondhehn, Norway.

The strongly non-linear free-surface deformations and inviscid fluid loadings which occur
during the constant velocity water entry and exit of a horizontal circular cylinder are here
calculated by the numerical method of Vinje and Brevig ~. Comparison with the loading
predicted by conventional slamming theory (here called added-mass theory) shows fair agree-
ment in some cases. However, a wide and fascinating diversity of free-surface flows is
calculated for various starting positions and values of Froude number and this precludes the
use o f added-mass theory in general. Interesting effects calculated by the numerical scheme are
supported by photographs o f experiments. For entry these effects include inflow over the top
of the cylinder, cavity formation, jet formation and combinations of these. For exit the calcula-
tions predict the lifting of the water above the cylinder and subsequent formation o f thin layers;
the draw-down and rush-up o f the free surface beneath the cylinder, the rush-up terminating
in localised breaking clearly seen in the experiments; the fluid motion after complete cylinder
exit; and, perhaps most interestingly, the formation of large regions o f strongly negative
pressure on the cylinder surface during late stages o f exit which explains the occurrence of the
"spontaneous" breaking o f the free surface seen in experiments.

I. I N T R O D U C T I O N numerical calculations with fully non-linear boundary

conditions. The latter yields a diverse and fascinating
Some members of offshore structures and pipelines are
variety o f fluid flows depending on Fr and initial posi-
horizontal circular cylinders which are often near to, or
tion, which are also present in photographs and a video
penetrating, the fluid boundaries i.e. the free surface, of experiments.
sea bed or other solid boundary. The hydrodynamic For translational or low-frequency motions, a simple
loading and motions of such cylinders are therefore im- analytic model may be built by ignoring wave damping
portant in a wide variety of applications e.g. pipelines,
(viscous damping may be included empirically) and con-
wave slam on cross-members o f oil rigs, and the motiva-
sidering the added masses o f the submerged portion of
tion for the present study which is accurate calculation
the body. With simplified conditions on the free surface
of fluid loading during installation and retrieval o f ob-
(z = 0), these added masses are known analytically, see
jects lowered through the free surface and to the sea
Greenhow and Yanbao 2 and Figure 1. For horizontal
bed, where the crane operator must keep control o f the
motions the free surface may not deflect very much, im-
body and avoid snatching on the cables. Clearly near the plying that O(~]Oz= 0 and z = 0 is an appropriate bound-
free surface, wave motion is likely to be important and
ary condition. Figure 1 shows a comparison, taken from
it is hoped to extend the methods described here to cover
Greenhow and Yanbao between the theoretical a d d e d
this in the near future. However, as a useful initial
mass and that inferred from a simple horizontal oscilla-
study, and because of its intrinsic interest, we here con-
tion experiment during which free surface deflections
sider the simplified problem of impulsively-started
were indeed small. The variation with dimensionless
constant-velocity V entry and exit o f a horizontal
distance from the fluid boundary, 6, is large in both
cylinder in initially undisturbed water at Froude theory and experiment.
numbers Fr ( = V2/ag where a = cylinder radius and
For vertical motions, except at very small Fr, the free
g = gravity) in the range relevant to these crane opera-
surface does deflect considerably during entry and exit.
tions. This paper compares the predictions of a simple
Nevertheless a crude approximation often employed in
analytical model outlined below with the results of
slamming (high-Froude-number entry) is to apply 4~= 0
on z = 0 giving the added masses shown dotted in Figure
1 (closed-form expressions are given in the appendix).
Accepted February 1988. Discussion closes January 1989. The force on the body is then given by the rate o f change

9 1988 Computational Mechanics Publications Appfied Ocean Research, 1988, Vol. 10, No. 4 191
tVater-entry attd -exit o f a horizontal circular cyfinder: Martin Greenhow

Table I Froude number (F,) attd bfitial positions zo]a chosen for the
calculations. The table indicates run mtmbers.
b 9 t r gt ", zo
tl
~ t~"! A
V2

,0 /2 Fr=--
ag

ENTRY.082
2 1.5 1~, 0.5

ENI
0

EN2
-0.5

EN3
ENTRY.51 EN4 EN5 EN6 EN7 EN8
,. / _ - - - -~:.7, . . . . . . EXIT .082 EXI EX2 EX3
EXIT .51 EX4 EX5 EX6 EX7 EX8

the dynamics o f the intersection points of free and body

Figure 1 Variation o f added mass with depth o f surfaces are described in Lin, Newman and Yue 4,
submergence f o r free surface condition on z = 0 given Greenhow and Lin 5 and Yim 6. Later Greenhow 7
by Orb/Oz= 0 or 4 = O. 9 denotes simple oscillation described the formation o f jets of fluid during high-
experiments. The inset compares the slamming coef- speed wedge entry and the occurrence of negative
ficient C~ variation o f Campbell and IVeynberg 3 with pressure on the upper part o f the wetted wedge surface
that o f theory, see Greenhow and Yanbao 2. if the jet is retained on the wedge. This implies that the
jet should fall off the wedge surface and some success
in calculating this was achieved for the entry o f wedges
of fluid momentum with deadrise angle o f more than 45 ~ Similar effects
also occur for cylinder entry described below. The
d dA (z) ~2
F = - d t (1 AZ) = - 1 ~ - 1 A~" (1) restriction to large deadrise angles for accurate
numerical resolution of the very fast moving jets
where A (z) is the added mass per unit length as a func- precludes the calculation of the early stages of water en-
tion o f centre position z measured downwards and l is try for the cylinder at present - in particular the initially
the cylinder length. For impulsively-started constant large slamming forces are not calculated here, but, rather
velocity V o f entry or exit this reduces to the calculations are started when the body is already at
least partially submerged with centre at z0, as listed in
dA V2 = lp Cs 2a V 2 (2) Table I. In the calculations for both entry and exit, runs
F = 1 '~-z 2
were made with 84 points on the free surface, 40 (or in
where p = fluid density and C, is the so-called slamming some cases 60) points on the body and up to 180 points
coefficient and is shown dotted in the inset in Figure 1 in total. The free-surface points were initially densely
(the solid line is the experimental result for high-speed spaced near the cylinder, spreading out towards distant
entry o f Campbell and Weynberg3). The most striking fixed boundaries 10 radii away on each side o f the
features of this force term are V2 dependence and the cylinder. During the time o f the runs (up to 100 time
fact that it always acts upwards regardless of the direc- steps) the effect of these boundaries was always thought
tion o f motion i.e. it will tend to accelerate the cylinder to be o f no importance to the flow local to the cylinder,
during exit and retard it during entry. and in most cases no significant disturbance reached the
Clearly the above simplifications o f the free-surface boundary anyway. The runs typically took about 60
conditions will introduce errors: for example for high- seconds c.p.u, per time step on a VAX 785.
speed entry, the effect of free surface rise is known to
be important during the early stages o f penetration and
this may account in part for the disagreement in 2. ENTRY OF A C Y L I N D E R (EN)
theoretical and experimental Cs values there, see also Figure 2 shows the entry of a cylinder at Fr = 0.082 from
Greenhow and Yanbao 2. Furthermore the added-mass the starting positions given in Table 1. For EN1 and
theory tells us nothing about the fluid motion nor free- EN2, the water merely flows in over the top of the
surface deflections, and gravity (finite-Froude-number) submerging cylinder (particle paths are shown for
effects are missing. EN2). The rest of the free surface is elevated above its
Direct appeal to experiments to validate (or other- mean position. For EN3, the water near the cylinder
wise) equation (2) is difficult because o f viscous drag on initially runs up its side and then a wave propagates
the cylinder which is unknown during motion near the away. Finally water again runs in to cover the submerg:
free surface, especially when the cylinder is penetrating ing cylinder. Photographs o f this are similar.
it. An empirical correction for viscous drag, based on The upward vertical forces acting on the cylinder are
Morison's equation, gave a force of comparable compared in Figure 2d with the buoyancy force (assum-
magnitude to the force term in equation (2) in the ing a cylinder at the same position in undisturbed water)
experiments o f Greenhow and Yanbao 2, and this gave and this buoyancy force plus the correction due to the
fair agreement with experiments in some cases. Never- rate of change of added-mass term given by equation
theless considerable uncertainty remains in both force (2). In this case buoyancy is dominant, the cylinder be-
terms. In an effort to quantify the inviscM fluid loading ing totally submerged shortly after 6 = z/a = Vt/a = 1.
during entry and exit, the method of Vinje and Brevig ~ Because of the strong gravity effect, which supports
is here applied. Developments o f this method to treat wave-like water motion, the agreement between the

192 A p p l i e d Ocean Research, 1988, Vol. 10, No. 4

 1Vater-entry a#ld -exit o f a horizontal circular cylinder: Martin Greenhow

I , t , l

ENZ
S

EN3 "
_J

>

I I | I I "| I I I | I I
-I 0 i
g
Figure 2 Cylinder entry at Fr = 0.082. a} EN1, b} EN2, --- denotes particle paths, c) EN3, d} Vertical force during
entry. - - - - denotes buoyancy, --- denotes added-mass theory o f equation 2 plus buoyancy, - - denotes numerical
calculations.

added-mass theory and the numerical calculations is in behind it; the initial intersection point particle (at the
most cases not good, except at the start of the calcula- intersection of free and cylinder surfaces) is projected
tions when the free surface is indeed fiat. vertically and thus leaves the cylinder surface. The new
In the program, on the vertical line of symmetry intersection point is chosen according to an algorithm
through the cylinder axis a thin plate was "inserted" described in Greenhow 7 , which then moves in to start
projecting upwards from the cylinder for the entry covering the cylinder top. For EN8, the intersection
calculations and upwards and downwards for the exit point particle again leaves the cylinder surface to form
calculation. Thus there is, in principle, no restriction on a fast moving jet of fluid, as seen in experiments o f
calculating flows after complete submergence: never- slamming (Figure 3g) and is similar to that calculated
theless this proved very difficult. After inflow over the for wedge entry by Greenhow 7. The calculations in EN7
top of the cylinder, the free surface particles were and EN8 could be carried further in time with finer
typically quite spread out, and thus not able to resolve resolution and showed no tendency to numerical i n -
the splash caused by the meeting of the two inflows (one stability. This is encouraging and raises the possibility
from either side of the cylinder) which occurs in the of calculating entry from just after the instant of slam
experiments. Some success was achieved in the case of until complete submergence.
EN4, where the cylinder was already submerged initial- The forces are shown in Figure 3f, the broken lines
ly, but even this flow became numerically unstable. It being as before. Here the added-mass theory forces a r e '
may be possible to resolve these splashes with finer much larger and generally it could be claimed that these
resolution but this has not yet been achieved. agree quite well with the numerical calculations despite
Figure 3 shows results for EN4 to EN8 at the higher the large free-surface deflections. EN8 shows an in-
Froude number Fr = 0.51. A number of interesting new teresting initial rise in force before the jets are formed
features are observed. For EN5, the free surface drew which then relieve the high pressures. While it is true
back to uncover the top of the cylinder in the last stages that the relatively coarse numerical spacing o f free-
o f the calculation. For EN6, the inflow over the cylinder surface particles inhibits the early stages of jet forma-
is large and would result in a vigorous splash. For EN7, tion, this force effect may also be partly physical.
the initially half-submerged cylinder forms a cavity Finally Figure3h gives the pressure distribution

Applied Ocean Research, 1988, Vol. 10, No. 4 193

Water-entry and-exit o f a horizontal circular cylinder: Martin Greenhow

EN4 EN5 -

EN6

EN8

,,-,I
~C
I'O

, . . . .

J
/
~~Z/Sf
s
/
d

o v' I
$

194 Applied Ocean Research, 1988, Vol. 10, No. 4

 IVater-entry and -exit o f a horizontal circular cylhlder: Marthl Greenhow

)V x

Figure 3 Cylinder entry at Fr = 0.51. a) EN4, b) EN5, c) EN6, d) EN7, e) EN8, f ) Vertical force during entry with
Ihws as hi ftgure 2, g) Photograph o f cylinder slammhlg showhlg jets as in EN8, h) pressure distribution around the
cylinder f o r EN8 when jets have formed. Two reference pressures 89p V2 and pga are shown. The hydrostatic pressure
is marked ..

around the cylinder for EN8 when the jet is well pletely covering the cylinder and is actually drawn under
developed. Since both gravity and hydrodynamic effects the cylinder during the later stages. In Figure 4a, the
are important here, two reference pressures are given: free surface and cylinder positions are labelled by the
~p V2 is a reference hydrodynamic pressure while pga is time-step number, and free surface particle spacing is
a reference hydrostatic pressure. Although the free sur- also shown. Just after step 50 the calculations break
face profiles for EN8 and the slamming experiment of down. This is not due to particle spacing or other
figure 3g are very similar, the pressure maxima observed numerical problems (several other spacings and time in-
by Campbell and Weynberg 3 are missing for EN8, the crements were tried) but appears to be physical. In the
pressure decreasing monotonically as one moves from experiments the free surface breaks spontaneously - an
the cylinder bottom to the free surface. effect which will be familar to anyone who has got out
o f the bath quickly. The cause appears to be that a
region o f strongly negative pressure develops on the
3. EXIT OF A CYLINDER (EX)
cylinder (between the points m a r k e d , in Figure 5a),
Figure 4 shows exit of a cylinder at F r = 0 . 0 8 2 for beginning at time step 24. At step 50, the negative
various starting positions. For EX1, the water above the pressure region is large in both magnitude and area, as
cylinder is lifted with the cylinder, and a thin layer o f shown in Figure 5h, and thus across the free surface a
water forms which eventually causes numerical break- pressure inversion must occur. In this case the free sur-
down due to poor resolution (the fluid particles on the face would suffer a Rayleigh-Taylor instability see e.g.
free surface tend to spread out). The majority o f the Baker, McCrory, Verdon and Orszag s. Figure 5g from
free surface near the cylinder moves downwards in all Greenhow and Lin 9 for a similar flow at higher Froude
three cases shown in Figure 4, so as to conserve fluid number shows this effect very clearly. It is o f course
mass. For EX2, the fluid particle at the intersection of impossible to resolve the flow beyond the onset of this
free and cylinder surfaces is lifted at first by the cylinder instability and the calculations should indeed break
and then moves rapidly downwards before again down here.
moving upwards which results in a localised breaking Similar comments apply to the experiments of EX5
and this stops the calculations. This is also seen in but here the calculations indicate the top of the cylinder
experiments. For EX3, a similar draw-down and rush- becomes exposed and the thin layer became numerically
up is seen, again leading to breaking near the intersec- unstable near the intersection point. (The photographs
tion point. Forces are shown in Figure 4d. The agree- o f breaking, not shown here, are interesting. The free
ment between added-mass theory and the numerical surface breaks momentarily, ventilating bubbles into
calculations is not good. the fluid which move down into the region of low
For exit at higher speed, Fr= 0.51, a number o f new pressure on the cylinder). For EX6, the calculations
effects are observed. Perhaps the most interesting case again broke down because the thin layer which is lifted
is EX4 where the free surface forms a thin layer corn- by the cylinder cannot flow down fast enough and

Applied Ocean Research, 1988, Vol. 10, No. 4 195

Water-entry and -exit of a horizontal circular cylinder: Martin Greenhow

t I a . . . .

EXI

.~ - - w . . . . .

...J

i,--
la..i
>
/
, 9 i i | i s | a

-I 0 I

Figure 4 Cyfinder exit at 17, = 0.082. a) EXI, b) EX2,

c) EX3, d) Vertical force during exit with fines as ht
figure 2.

EX5

EX4

J I a'"

196 Applied Ocean Research, 1988, Vol. 10, No. 4

 IVater-entry and -exit of a horizontal circular cylinder: Martin Greenhow

i i i i | i | i i i t

.,~ ! I !
-I o i 1,
g

Figure 5 Cylinder exit at Fr = 0.51. a) EX4, b) EX5, c)

EX6, d) EX7, e) EX8, f ) Vertical force durhlg exit with the cylinder for EX4 just before nunlerical breakdown.
lines as in figure 2, g) Oblique photographs showhtg Two reference pressures ~p V 2 and pga are shown. The
spontaneous breakhtg, It) pressure distribution around hydrostatic pressure is marked ..

becomes almost separated from the rest of the water. In possible to calculate the subsequent fluid motion, and
experiments this thin layer broke up into droplets, some early stages are shown here.
presumably because of surface tension. The very fine The forces are given in Figure 5f. One important
resolution needed in this case is shown in Figure 5e. practical conclusion is that the added-mass theory gives
EX7 shows very pronounced draw-down and rush-up a contribution which is generally in poor agreement with
(as in EX3) giving excellent agreement with the numerical calculations and may even be of the
experimental photographs. The rush-up culminates in wrong sign, see EX4 and EX5. Another point of interest
localised breaking at the intersection point in the is the large increase in upward force during the rush-up
experiments and numerical breakdown in the calcula- in EX7 which may have significance for crane opera-
tions. EX8 shows an example of complete exit of the tions and fatigue on oil rig cross-members. (The
cylinder. As described earlier, the calculations can con- magnitude of the peak is not easy to calculate accurately
tinue beyond the time of complete exit because of the because of the difficulty of calculating the rush-up
thin plate insertion along the line of symmetry. It is thus exactly, but the effect was also noted in pressure

Applied Ocean Research, 1988, Vol. 10, No. 4 197

IVater-entry and -exit o f a horizontal circular cylinder: Martin Greenhow

measurements in a corresponding experiment. H o w - totally submerged (centre position z / > a) then:

ever, further experimental work is needed to confirm the A(Z) (1-q2)2 /./~~ 1 KZ(m,)
effect). pTra2 = q2 ~.~'- + ~ (1 + nl~ ) zc----y--
A l t h o u g h the motivation for the present work was the
need to calculate the force time histories during entry E(nll) K(n11) 1 2 K2(n12)
and exit, the time histories o f the pressures a r o u n d the 7r2 6 3 (1 + m ~ ) 7r--------T---
b o d y were printed out in all cases. These were useful for
u n d e r s t a n d i n g the behaviour o f the intersection point, 2E(m2)K(m2)]
especially when it left the cylinder surface (EN7 and 7 -j-1
EN8) and f o r understanding the s p o n t a n e o u s breaking
during exit (EX4, see Figure 5h). Further i n f o r m a t i o n where E ( m ) and K(m) are complete elliptic integrals as
and p h o t o g r a p h s are available f r o m the author. given in A b r a m o w i t z and Stegun it, chapter 17,
ml + nil = 1
q = exp [ - 7rK(nl{ )[K(ml )]
4. C O N C L U S I O N
Z z _ a2= a2(1 -qZ)Z]4q2
This paper presents accurate numerical calculations
o f the entry and exit o f a cylinder which include b o t h and m2 is given implicitly in terms o f mt by
gravity and non-linearity in the free surface and b o d y K(m{) 1 K(m~ )
b o u n d a r y conditions. For entry, inflow, cavity and jet K(ntl) - 2 K(m2)
f o r m a t i o n were calculated. For exit, thin layers o f fluid,
d r a w - d o w n and rush-up and the development o f An equivalent expression may be found in
negative pressures were calculated, the negative pressure Venkatesan ~z. When the cylinder pierces the line z = 0
causing a breaking o f the free surface seen in i.e. z ~< a then:
experiments. It is notable that the p r o g r a m calculates a A ( z ) _Tr z ( 1 - c o s O ) ( 1 - c o s 0 ) sin0-0
wide diversity o f flows without modification to the code, pTra z 3 ~: 0--~+ 6 I - - - 2re
and that agreement with experimental free surfaces is
g o o d , especially in regard to the onset o f breaking. where 0 = c o s - t ( - z ] a ) . This expression is derived f r o m
T h e forces on the b o d y were c o m p a r e d with a simple Taylor 13.
b u o y a n c y and added-mass theory. Generally the agree-
ment with such a simplified t h e o r y was not g o o d and the
complexity and diversity o f the flows m a y preclude the REFERENCES
practical use o f any such simple model. However, the
numerical m e t h o d is itself a practical tool which couid Vinje, T. and Brevig, P.: Nonlinear two-dimensional ship
motions Proc. 3rd int. Conf. Num. Ship Hydro. Paris, June
be used for m a n y other geometries and proved to be pp. 257-266 (1981) (see also Nonlinear two-dimensional ship
r e m a r k a b l y robust (no explicit s m o o t h i n g or redistribu- motions S.I.S. rep. no 112.81, NHL, Trondheim, for further
tion o f points was used in any calculation). details)
Greenhow, M. and Li Yanbao: Added masses for circular
cylinders near or penetrating fluid boundaries - review, extension
and application to water - entry, - exit and slamming, Ocean
Engineering 1987, 14, 4 pp. 325-348
5. A C K N O W L E D G E M E N T Campbell, I. M. C. and Weynberg, P. A.: Measurement of
This w o r k was partly supported by the Royal parameters affecting slamming. Final report, Wolfson Unit for
Marine Tech. Rep. No. 440, Technology Reports Centre No OT-
Norwegian Council for Industrial and Scientific R-8042, 1980
Research ( N T N F ) under the Marine Operations Lin, W-M., Newman, J. N. and Yue, D. K.: Nonlinear forced
Programme. motions of floating bodies, Proc. 15th Symp. Naval Hydro,
Hamburg, Germany, Session I, pp. 1-15, 1984
Greenhow, M. and Lin W-KI.: Numerical simulation of nonlinear
free surface flows generated by wedge entry and wavemaker
motions. Proc. 4th Num. Conf. Ship Hydro, Washington D.C.
6. N O T E A D D E D IN P R O O F Sept., 1985
Yim B: Numerical simulation for two-dimensional wedge slamm-
Recently Telste t~ has p r o d u c e d results for a m o r e ing with a nonlinear free-surface condition. Proc 4th Int. Conf.
deeply submerged cylinder rising to the free surface Ship Hydro. pp. 107-116, Washington, Sept., 1985
after an initial acceleration period, using a generalised Greenhow, M.: Wedge entry into initially calm water Appl.
vortex m e t h o d . A l t h o u g h no direct c o m p a r i s o n can be Ocean Res (1987) 9, no 4, 214-223
Baker, G. R., McCrory, R. L., Verdon, C. P. and Orszag, S. A.:
m a d e with the present results, his results are qualita- Rayleigh-Taylor instability of fluid layers. J. Fhdd Mech. (1987)
tively similar to case EX4 in this paper, and to the 178, 161-175
experiments o f G r e e n h o w and Lin 9. Greenhow, M. and Lin W-M.: Nonlinear free surface effects.
Experiments and theory KILT. Dept. Ocean Engng. Rep. No
83-19, Unpublished Rep., 1983
10 Telste J. G.: Inviscid flow about a cylinder rising to a free sur-
face, J. FluM Mech. (1987) 182, 149-168
APPENDIX 11 Abramowitz, M. and Stegun, I. A.: Handbook of Mathematical
F o r convenience, we here give results taken f r o m Functions, Dover Publications Inc., 9th ed., (1970)
12 Venkatesan, S. K.: Added mass of two cylinders, J. Ship Res.
G r e e n h o w and Y a n b a o 2 for the a d d e d mass A ( z ) o f a (1985) 29, 4, 234-240
horizontal circular cylinder m o v i n g vertically near to or 13 Taylor, J. L.: Some hydrodynamic inertia coefficients. Phil.
t h r o u g h a line o f equipotential ~b = 0 on z - - 0 . W h e n Mag. (1930) 9, series 7, 161-183

198 Applied Ocean Research, 1988, Vol. 10, No. 4