Professional Documents
Culture Documents
0141-0296/80/010015-12/$02.00
1980IPC BusinessPress Eng. Struct., 1980,Vol. 2, January 15
Fluid-structure interaction in Morison's equation for offshore structure design: P. Ft. Fish et aL
may be described by N dynamic equations as follows: (ii) An inertia force (known as the Froude-Krylov force)
[MIni + [Clk + [K]x =F (1) due to the undisturbed pressure field around each body.
where: [M] is the mass matrix of the structure including (iii) An inertia force dependent on the relative acceleration
enclosed water and marine growth mass; [C] is the matrix between the water particles and each body. This force is due
due to structural and soil damping; and [K] is the matrix due to the disturbed pressure field around each body and repre-
to structural and soil stiffness. sents the force due to the time varying 'added hydrodynamic
The force vector F.aceounts for the hydrodynamic forces mass' of the body.
acting on each of the N bodies. Mofison et al. 9 proposed In this method, it is normally assumed that CM and CD are
that the force exerted by a flowing fluid on a rigid cylinder equal to the values applicable on a rigid, stationary cylinder.
of incremental length dS may be represented with reasonable
accuracy by the sum of a drag and an inertia force: Method B The non4nteractive form of Morison's
equation (2) is used to represent the fluid force, while the
dE = ~CDpDU 2 ds + C m p a U ds (2)
structural side of equation (1)is modified as follows:
then:
[M+(Cm-1)pVlx+[C]x+[K]x=r (4)
where: (Cm - 1) p V represents the added hydrodynamic
mass of the displaced fluid, which is assumed to be constant
0 at all points in the wave profile.
Method A, which employs equations (1) and (3), is the
where 77is the instantaneous water level at the axis of the
normal approach adopted for research studies, 1'2' lo
body.
especially when the structure is very responsive. However,
The drag and inertia coefficients Co and CM must be
due to the nonlinearity in the drag term, the equations are
determined from experience or experiment. Selection of the
often linearised as described above.4'6
correct coefficients is dependent on Reynolds number, sur-
Method B, based on equations (2) and (4), facilitates a
face roughness and the ratio of orbit radius to diameter
less costly solution procedure and is, therefore, the method
(Iversen's modulus) and, strictly, must be considered as vary-
adopted by most commercially available programs for the
ing along the member and with time. However, they are
dynamic analysis of offshore platforms, etc. 11
normally specified as an average for a particular member,
and this value is used at every calculation point within the If fiuid-structure interaction is important under any parti-
wave since the determination of time dependent coefficients cular set of conditions, then it is to be expected that the
is too complex. These time-averaged coefficients give a amplitudes of rotation and displacement predicted by
reasonable description of the time variation of the force, methods A and B will be different.
As a consequence of the amplitude comparison, it is
except when there is significant response due to vortex-
possible to deduce how much extra damping should be used
shedding. The latter may occur in waves for which the
in method B in order to simulate the effects of hydro-
Keulegan-Carpenter number (Um=x T/D) is in the range
6_29.1, 2 dynamic damping and so obtain the 'correct' amplitude as
predicted by method A. This extra damping is called the
'implied hydrodynamic damping in waves' and is very signi-
Morison's equation in design practice ficant for certain combinations of structural parameters and
sea state.
Now that the offshore jacket type of platforms are The vital question to be answered, therefore, is: Under
designed for waters of increasing depth, their dynamic what conditions does a difference in amplitude occur which
characteristics become progressively more important. Soil is significant with regard to the design of offshore structures?
and structural damping are then of less importance than the This question will be answered as far as possible in this
hydrodynamic damping due to the fluid in which the struc- paper, but the range of conditions investigated has been
ture is immersed. It is essential for safety reasons that damp- limited for practical reasons.
ing is not over estimated and wasteful in terms of construc-
tion materials if damping is underestimated.
Therefore, if the structure is responding to water waves C o m p u t e r program P L A T D Y N
then in a dynamic analysis using Morison's equation, hydro- Description o f program
dynamic damping may be included by one of two methods.
As part of the present study of fluid structure interaction,
Method A Equation (2) is used with relative terms for the program PLATDYN (PLATform DYNamics) was deve-
velocity and acceleration as follows: loped to analyse the dynamic response of fixed, tethered and
floating structures. The program calculates the response of a
dE= ~C~pD(U- ~)IU- ~IdS + pAO dS flexible fLxed structure by simulating it as a rigid body
Drag Froude- mounted on a number of springs and dashpots, whose stiff-
force Kdlov
force ness and damping are chosen to match the actual stiffness
and damping characteristics of the entire soil-structure
+ (cm - }) . A ( t ) - ~) dS
Inertia
system.
force (3) The action of waves and currents is included by specifying
a particular sea.state which may consist either of purely
which accounts for fluid-structure forces and their inter- regular waves, or of random waves to simulate a JONSWAP
action as follows: spectrum. A current is added vectorially to the wave4nduced
(i) A drag force dependent on the relative velocity between particle velocity before calculation of the hydrodynamic
water particles and each body. forces. The initial displacement and velocity conditions of
the centre of gravity may be defined. The ensuing motion of time-history analysis, defining 'truth values' of displace-
the structure is then computed by PLATDYN in the time ment. The version of PLATDYN using Morison's equation
domain, using a modified second order Runge-Kutta method in this form will be called PLATDYN 'A'. The other method,
applied to equations (1) and (3) in up to six degrees of given by equations (2) and (4), ignores :}2 in the drag term,
freedom and taking into account all the externally applied and the added hydrodynamic inertia is included in the mass
forces arising from hydrostatic and hydrodynamic considera- matrix instead of there being (correctly) a time-dependent
tions and from any mooring cables, springs and dashpots. inertia force on the right-hand side governed by mass coeffi-
The hydrodynamic forces are derived by application of cients CM. Version PLATDYN 'B' incorporates the latter
Morison's equation, using linear wave theory, to elemental method. Using the one program PLATDYN, with minor
parts of each constituent body in the structure. The true changes for both analyses, provided a consistent solution in
immersed volume and draft of each element is calculated at which the differences in answers could be directly attributed
each time step when evaluating the fluid-induced forces. to the effects of fluid-structure interaction.
The wave-induced components are calculated from the
particle velocity and acceleration of an undisturbed wave at Structure f o r analysis
the centre of the submerged volume allowing for the inclina-
tion of the body axis to the vertical. Reduction to simpler structure
The wave.induced forces include the drag component The structure chosen for analysis is the EEC phase 111
normal to the body axis and skin friction drag, each calcu- reference structure (see Figure 1) which stands in 156 m of
lated from the relative velocity of the body and fluid. For water. The sensitivity of this structure has been analysed by
inertia governed forces, the Froude-Krylov force due to the Lamb.12 Due to the complexity and expense that would be
acceleration of the water around the body and the pressure involved in analysing the full structure on PLATDYN, a
force due to the relative acceleration of body and fluid are simple two-dimensional reduction to 33 equivalent zones
computed. The dependence of the inertia-governed forces on was carried out in the x - z plane (see Figure 2), using a
the relative acceleration of the body and fluid, and vice- program called ASASMASS. This calculates the added mass
versa, requires an iterative approach to the dynamic equili- of fluid for each zone of the structure in the three mutually
brium of the force vectors. perpendicular directions of the global axis system and hence
PLATDYN facilitates specification of the structure in determines the added mass, the total area presented to the
three dimensions (the vertical axis is modified to obtain true fluid and the displaced volume for any desired number of
values of wave particle motion above the mean water level). zones. Each zone consists of one vertical cylinder defined by
The response of the structure in any or all of its six degrees its length, diameter, mass and coefficients CD and CM. The
of freedom (surge, sway, heave, roll, pitch and yaw) may be cylinder represents a number of different kinds of real
analysed allowing for any coupling effects. Chosen degrees members at various inclinations in the same physical zone of
of freedom are dynamically coupled by inserting appropriate the reference structure and it must be statically and dynamic-
off-diagonal terms into the mass, stiffness and damping
matrices.
Each body comprising the structure for analysis is des-
cribed by its external dimensions, and by three drag coeffi- Z
cients and three mass coefficients applicable to Morison's
equation. Further, if the effect of wave radiation damping is
to be included (large bodies only), then three surface wave
damping coefficients must be specified.
Finally, the number of elements into which each body is
subdivided may be varied. Increasing the number of elements ion
will afford a more representative distribution of the hydro-
dynamic forces on the structure but will have the disadvan-
tage of requiring more computing time.
Several tests were run on PLATDYN, using the reduced
EEC structure (see section on structure for analysis), in order
to verify that, under restricted conditions, the program pro-
duced results in agreement with hand calculations. For
example, the structure was displaced in still water in which
hydrodynamic drag was eliminated by defining the drag Co-
efficients as zero. The computed frequency agreed exactly
with the theoretical value. The integration parameters were
varied in order to optimise numerical accuracy against com-
puting time. The time step was fixed at 0.1 s and an iteration
accuracy of 1% was sufficient to calculate displacements
accurate to the nearest 1 ram.
Modification to PLA TD YN
Two methods of applying Morison's equation were given
in the previous section; equations (1) and (3) were the fully-
interactive method including the nonlinear ~2 term and it is
assumed that the solution to this equation gives the 'correct' Figure 1 EEC phase III reference structure
Table I ASASMASS analysis of reference structure. 10.1 m marine growth thickness at Mean Water Level)
Depth, ~} 0.93
+2--
3.28 2.75 4.26
3.76 Mass in Gg
-- ---~o- ~
3.56 2.46 3.72
0.77
C 9 A
A 5.46 34.9
B 13.19 21.5
C 6.84 25.4
-l:~,x X X \ \\ Mannc
I I I I gr°wth
O 25 50 75 IOO thickness,
(ram) (No mass is carried by the conductors -- member D)
Figure 3 Assumed variation with depth of marine growth on reference
structure This Is illustrated in Figure 2
following equations: sea states and structural properties. The full details of the
input data for versions A and B are given in Tables 4 to 7.
COnS
[cl = Ira] (24) The runs are numbered 1 to 11, but a total of 18 production
ff runs were in fact performed as, in most cases, versions A and
B were used for each run.
[CI : [K] (25)
~Onff Hydrodynamic damping in still water - runs 1, 2
where the logarithmic decrement 8 = 2hi'. The base springs kH and kv were calculated using equa-
Mass proportional damping may or may not contain the
tions (18) and (19) to give natural periods in the sway mode
added mass effects of the entrained water. For the reduced
Ts= 2.165 s (Run 1) and 4.33 s (Run 2). The soft damping
reference structure, the latter equation was therefore used. matrix [C] was set to zero so that the only, contribution to
The damping components so obtained were between I0 and damping was due to the hydrodynamic x ' d r a g force term
30 times greater than those suggested by equation (24).
in Modson's equation. In still water, the response of the
Also, stiffness proportional damping was seen to enhance
reduced structure was found to be very sensitive to the
the relative importance of damping in the rotational degree-
choice of initial displacements. If the run was started w h e n
of.freedom as compared to the translational degree-of-
the structure was not exactly in its mode shape, then the
freedom. ensuing motion would be erratic. Therefore, the structure
The relation of the mode shape to the displacement of the
was given an initial deck level displacement of 0.4 m and a
the structure can be derived from Figure .5 and is given by:
mud level displacement of 0.01 m.
•X X I ( H0 -- Lo) + )`2Lo The logarithmic decrement was determined by computing
(26) the ratio of successive amplitudes in each period of response.
0 )-1 - ),2 Both runs 1 and 2 gave a hydrodynamic damping value of
Assuming that 7t ~ 72 "" 0.~ ),,/),2 represents the struc- 0.5% critical. As an example, F/gure 6 shows the time his-
tural flexibility in relation to the soft flexibility, it was tory o f run 2 for versions A and B, the latter being of course
estimated that a value of 40 is reasonable for the EEC steady-state in the complete absence of damping.
reference structure.
Implied hydrodynamic damping in waves
C a l c u l a t i o n s a n d discussion The purposes of running versions A and B of PLATDYN
Once the new program PLATDYN had been validated, a was to generate some values for the 'implied hydrodynamic
number of 'production runs' were conducted for specified damping in waves'. Essentially, version A represents calcu-
Computed
Wave Wave Wave Nat. period Nat. freq. Soil hydrodynamic
Run Sea length, height, period, of struc, of strut. damping % damping
no. state L, (m] H, (m] Tw, (s) Ts, (s) Ns, Hz critical CH % crit.
Table 5 Mass values for PLATDYN production runs lation methods which allow for fluid.structure interaction
in dynamic analysis, while version B is representative of
Mass matrix
those calculation methods in which it is assumed that the
Version MI1 MI2 M22 wave force is not modified by the motion of the structure.
Gg Gg • m Gg • m 2
At least four runs of PLATDYN are necessary in order to
A 42.0 --2304.0 192861.2 obtain one estimate of implied hydrodynamic damping. This
involves the solution of Morison's equation with and with-
B 71.2 --5283.3 535322.0
out fluid.structure interaction for at least two soil damping
values.
There are some difficulties in applying this concept, in
particular:
Table 6 Stiffness values for PLATDYN production runs (a) The implied hydrodynamic damping will depend on
Horizontal Vertical Stiffness matrix
sea state and structural dynamic properties.
springs springs (b) The amplitude difference varies during the response
Run kH kv Ks= K t == K2t K=2
number (GN/m) (GN/m) (GN/m) (GN/red) GN.m/red)
time-history in regular waves and would need to be time-
averaged for random waves.
1 3198.1 3016.7 6396.2 --969668.0 152431992.0
As only a limitedrange of sea states and structural proper-
2--11 799.5 754.2 1599.1 --242415.7 38107798.0 ties have been tested, it is not possible to define all the con-
ditions under which a significant difference in amplitude
Damping matrix
Run Soil Cll C1= = C=1 C==
number damping GN • s/m GN • s/red GN.m.s/red
~%
P
22 Eng. S t r u c t . , 1 9 8 0 , V o l . 2, J a n u a r y
Fluid-structure interaction in Morison's equation f o r offshore structure design: P. R. Fish et al.
I °,
o•
O.OG t
• °o • =,
•° •,
• •l
• o ,•
• °o •°•
=1.
I I I I I I I I I I I I I I I I I I I I
O.OO I.OO ?-OO 3LOO4.OO SO0 ~O0 7JOO 8.O0 9.O0 IQO0U.O0 I~-O0 13.O0 14.O0 I~OO 16.O0 17.OO 18.O0 19J00 20.O0
TimeJs|
Figure 7 Demonstration of non-monotonic response of reduced structure in regular waves (run 3). Expected evelope of oscillation after long
integration period shown in top right hand corner
occurs. Point (a) means that a further study should therefore Run 4: regular waves This run was conducted with a
include careful variations in these quantities. Point Co) is regular wave length of 7.5 m, and, as the members A, B and
important because of the implication that picking off peak C of the reduced structure were equispaced by 30 m (Figure
values o f the response amplitude and using these to deter- 2), this meant that the wave peaks occurred simultaneously
mine 'implied hydrodynamic damping' is not an entirely on members A, B and C. However, the wave period Tw =
satisfactory procedure. A comparison of the r.m.s, of the 2.165 s was not coincident with the natural period of the
displacements obtained from PLATDYN versions A and B structure Ts = 4.330 s and it was found that, for a soil damp-
would be more appropriate for both regular and random ing value of 1%, the difference in the amplitudes of hori-
wave loads, but to do this would necessitate long runs on zontal displacement and rotation, as predicted by PLATDYN
the computer. It was therefore decided to use only the peak versions A and B, was negligible. This is to be expected as
displacements for comparison in this analysis, but the time- the wave height is only 0.75 m and the amplitudes of res-
history of response would be studied carefully in each run ponse are small enough to subdue the effects of fluid-struc-
performed ture interaction. It was therefore decided to adjust the wave
period to coincide with the natural period of the structure,
Run 3: regular waves Morison's equation contains wave since the effects of damping are most clearly observed at
velocity and acceleration terms in U IUI and U which means resonance.
of course that the wave-induced forces will contain two or
more frequency components, depending on the choice of Runs 5, 6, 7: regular waves These runs represent the
wave theory used to define the water particle velocities resonant condition in which the wavelength was 30 m,
(HaUam et al.). 7 Therefore the dynamic response of the exactly equal to the distance between each of the members
structure will not be sinusoidal with constant amplitude and
the effect of this on the time history is shown in Figure 7 in
which a particularly long tun time of 20 seconds was allowed.
Even this was not long enough to show the complete effect
which is similar to a high frequency wave within a low fre-
quency evenlope. The sketch on Figure 7 illustrates this and
raises the problem of deciding at which point in the time
history to compare amplitudes of displacement. Probably
the most significant point is where the maximum amplitude
occurs, but to find this would require in most cases, run
times well in excess of 20 s. As this would be prohibitive in
view of the substantial computing costs involved, it was
decided to compare displacements as predicted by versions
-1.COt I I I I I I I I I t
A and B at successive amplitudes (separated by one response GO0 2DO 4C0 ~ 8C0 K:K~O
period) just to illustrate the way in which the amplitude T~m¢. (scc|
difference between A and B varied during the response time Figure 8 Time-history of response of the reduced structure when
history. exposed to regular waves (run 6). (*), version A; (--), version B
OE
~
-
, ~ 6°/o Impliedhydrodynamic
~/donpincj
2nd peok
1.4% Implied hydrodynomic
pin¢j
I I I I I I
0 2 46 8 I0 12
Scxl daring, ~ qo
O15 Figure 11 Hydrodynamic damping in a typical design wave (runs 8,
A 9 and 10). First peak is separated from second peak by one period
o f the response. Wave height, H = 24 m; wave length, L = 240 m;
wave period, Tw = 12.25 s; structures natural period, Ts = 4.33 s
I1% Implied hydrodynamic
,,J
times the distance between members A and C, so that the
c reduced structure lay completely within one quadrant of
III
010 - Ist p e a k '
the wave. The response is entirely transient and, again, it
A would have been desirable to have pursued the time history
i:3 (Run 9 is shown in Figure 10) to near rest, but computing
expense prevented this. Instead, the 1st and 2nd Peaks have
been compared in Figure 11, and a considerable difference
is found between them. In view of the large perturbation
caused by the wave, the calculation procedure will take
longer to stabillse and therefore the 2nd Peak results are of
most significance. Figure 11 indicates that there may be as
much as 12% implied hydrodynamic damping at the first
response peak and 6% at the second peak.
To obtain the value of 12% implied hydrodynamic damp-
I I I I I ing, it was necessary to extrapolate the line through the data
0 I 2 3 4 5
points for version B of PLATDYN. In fact, it is not obvious
sc~ dc~o~ ~O/o
that the lines for A and B should be parallel, and indeed, it
Figure 9 Hydrodynamic damping in regular waves (runs 5, 6, 7).
First peak is separated from second peak by one period of the
is more probable that the lines will converge for higher
response. Wave height, H = 3 m; wave length, L ,= 30 m; wave period, values of damping because the diminishing response would
Tw = 4.33 s; structures natural period, Ts = 4.33 s be reducing the effects of implied hydrodynamic damping.
He Distance from mud line to deck level (m) 0 Rotation in x - z plane (rad)
0 Amplitude of 0-rotation (rad)
Significant wave height (m)
klk 2 Maximum displacement of structure at deck and
/el Moment of inertia of each zone (Gg. m 2)
mud levels (m)
IaAt Added moment of inertia of each zone (Gg. m 2)
p Density of water
ku Horizontal base spring of reduced structure (GN/m)
con Natural frequency of structure with added mass of
kv Vertical base spring of reduced structure (GN/m) fluid (rad/s)