Documente Academic
Documente Profesional
Documente Cultură
This psper was presented at the 27th Annual OTC in Houston, Texas, U. S.A.,1-4 May 1995.
This paper was selected for presentation by the OTC Pr~grem Committee following review of information conteined in an abstract submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Offshore Technology Conference and are subject to carrectlon by the euthor(s). The material, es presented, does not necessarily reflect
anv oositlon of the Offshore Technoloav Conference ar its officers, Permissionto coDvis restricted to an abstract of not mare than 300 words. Illustrations mav not be covied. The sbstract
sh&ld contein conspicuous acknow[;dgment of where snd by wham the psper ii ‘presented.
783
—
2 EXPERIMENTAL AND ANALYTICAL STUDIES OF WAVE IMPACT FORCES ON EKOFISK PLATFORM STRUCTURES OCT 7782
measure and interpret wave impact loads on the deck with water the effective mass can easily increase by an
structures can be of widespread interest. These order of magnitude,
measured impact loads included resonant effects from
the models that were separated and explained, The model design process becomes more complicated
as the number of structural components increases and
This paper is focused on the methodology of consequently the interactions among them. Modal
measurement and data interpretation rather than on analysis of the integrated model structure will identify
reporting findings of the study. The paper identifies a the resonant frequencies and related mode shapes
number of considerations related to the design of scaled which generally do not correspond to a particular rigid
models to be used in tests measuring impact loads. body motion but identify a distortion in the model,
These considerations can be of general use and However, these modes can be somewhat controlled by
information regarding knowledge of such forces. controlling the stiffness of linkages between rigid
components as in the case of a rigid component held in
MODEL JACKET DESIGN position by softer load cells.
Here M is the physical mass of the structure (or ■ 6 DOF loads on the deck structure,
particular member), A is the added mass , B is the ■ 6 DOF accelerations on the deck structure,
damping comprised of structural and hydrodynamic 9 6 DOF global loads on the jacket structure.
contributions, K is the effective stiffness and F is the
applied load. In cases of static or quasi-static loads the The model shown in Figure 1 is considered to consist of
Kx term is measured by a load transducer as a function two components, the deck and the jacket, and loads
of time and the hydrodynamic load, F, can be directly were required to be measured independently on each of
equated to this measurement. This assumes that the the two components. The deck is stiffened using
model’s dynamic response is negligible in relation to the aluminum tubing and covered with aluminum sheeting,
measurement. The solid sheeting can be removed from the deck and
replaced with porous sheets as part of a design
The structure’s dynamic response is described by the technique being investigated to reduce the impact
first two terms on the left-hand side of Equation (1), loads. The deck frame is attached to the jacket using a
The inertia and damping terms will be comprised of total of 8 slender flexures as illustrated in Figure 2. Five
both structural and hydrodynamic effects, When the flexures are mounted vertically and three are mounted
structure responds to an impact load its resonant horizontally.
response will be sensed on the measuring transducer. In
physical models, other than hydroelastic ones, resonant These flexures are manufactured from high tensile steel
responses are considered as contaminants to the with necked sections giving a bending to longitudinal
measurement and must be removed or ‘avoided, stiffness ratio of 103. Thus the flexure is sensitive to in-
Iine loads only. At the end of each flexure is a load cell
The underlying strategy is therefore to design a model which measures the load in the flexure. Knowing the
of sufficient stiffness such that the resonant locations of these load cells with respect to the mass
frequencies are above the lowest one of concern in the centre of the deck, the measured loads are resolved into
full scale structure. The lowest frequency of interest to six-degree-of-freedom loads, The deck is also equipped
a designer is specific to a particular structure and can with a system of seven accelerometers used to measure
depend on the global or local structural characteristics the six-degree-of-freedom accelerations of the deck
such as mass and stiffness. Higher stiffness in using the method described in Miles [11, The jacket
instrumentation will tend to increase the natural section, constructed of aluminum tubing, was mounted
frequency but at the expense of sensitivity. There are on a six-degree-of-freedom force dynamometer.
also design constraints related to the availability of Therefore, the total forces measured on the model
practical construction materials and fabrication including the deck and jacket were measured at the
procedures, Added mass effects on the model structure base of the model, Loads on the jacket alone were
significantly influence the natural frequency. Even when determined by subtracting the deck loads from the base
ultra-light weight materials are used, once in contact loads.
784
—
OCT 7782 J.J. MURRAY, P. KAPLAN AND W.C. YU 3
Samples of the deck response in air and in water as In the present test program, the storm duration was 90
measured by the vertical force transducer, are shown in minutes. Regular and irregular waves for the test
Figures 3 and 4, respectively. Figure 3 shows an almost program were calibrated in the basin without the
linear damped response with a dominant frequency of models installed for the three different water depths
6,3 Hz, This is expected since only structural damping corresponding to three different levels of subsidence.
is involved, The decay trace in Figure 4 shows a Samples of the ratio of maximum crest elevation and
significant increase in the decay period and damping, as the standard deviation of crest elevation measured for
expected from the hydrodynamic effects. This time two of the spectra during wave calibration are
trace also exhibits a beating frequency indicating a compared to those predicted using Equation (2) in
coupled response of two components with very close TABLE 2. There is very good agreement between
frequencies. In fact, the second modal frequency predicted and measured values.
predicted by the finite-element analysis is 1.8 Hz and is
a rotation about the base of the model jacket. All TABLE 2 Ratio of Maximum to Standard Deviation
frequencies are full scale equivalents determined from of Crest Elevations
the model measurements using Froude scaling laws.
Predicted Measured
Ratio Ratio
TABLE 1 Measured and Predicted Natural
Frequencies SPECTRUM 1 3.63 3.89
%
Mode Predict Measure Predict Measure
SPECTRUM 2 7.27 7.33
Air Air Water Water
(Hz) (Hz) (Hz) (Hz)
Increasing the time duration of the waves used to
1 5.8 6.3 1.6 1,4 model a particular spectrum above 90 minutes will
&
increase the ratios given in TABLE 2. The calculated
The measured values from the vibrations test compare values assume that the waves are linear. However,
very well with the predicted ones as indicated in TABLE because of the near shallow water conditions, the
1. The added mass effects on the deck as well as those waves are asymmetric. This is an important issue
on the jacket have a significant effect on reducing the because, considering clearance beneath a deck, the
maximum crest height is of primary importance as
natural frequencies.
compared to the maximum wave height.
785
—
—
— — —- . .— — __.
-— —_ —. -.=— ._L _
.—_. . . . ____ .—= .-. y.== —.— .—_ .... .-—
. _ -=. — — —.——— ——. -=_ ______ , ._ >_: __
— - — —.—— —
.
— -. — — — —_. _— ->- —- —-—
—=:: .—_u___ -_ ._ ._ —_
— -— .
4 EXPERIMENTAL AND ANALYTICAL STUDIES OF WAVE IMPACT FORCES ON EKOFISK PLATFORM STRUCTURES OTC 7782
The distribution of wave crests and heights are further Analysis of Impact Loads
investigated for the spectra. According to the Rayleigh
model, the probability of exceedence, PE, of a quantity Figure 8 compares the impact loads measured in the
x, is given by the expression, 2/4 R jacket model in the presence of the barrier and in
the far field. These results are related to the jacket in
x’ the NW position as illustrated in Figure 7. In the figure,
PE (X) =exp[— —
202] (3) F~Zis the vertical load measured at the base of the
model, F~Z is the vertical load measured on the deck and
F~x is the horizontal load measured in the direction of
Using a zero-crossing analysis, the wave heights and wave propagation. The time traces of the impact loads
wave crests generated in a sample of the modelled also illustrate distinct differences in magnitude as well
spectrum are compared to the distribution given in as shape. The impact loads are characterized by a sharp
Equation (3) . Results of these measurements are rise to a peak value with a more gradual decline as the
shown in Figure 5 and 6. The wave heights given in momentum of the water diminishes. The short rise time
Figure 5 appear to fit the model very well, indicating in the impact loads is comprised of many frequency
that the wave heights in the modelled spectrum are components, some of which are much higher than the
Rayleigh distributed. A comparison between observed frequency of the impacting wave.
and estimated crest heights shows underestimated
probability of exceeding the small range crest heights, The resonant response of the model is clearly evident in
but very good match for the large range of crest heights the time traces. Results in regular waves show
that could be of concern. increasing resonance effects with increasing steepness
of wave crests. This is because the level of dynamic
MODEL TESTS ON 2/4 R JACKET response depends on the rise time of the impact;
increased steepness results in increased impact load
Following wave calibration, the 1:36 scale model of the and larger dynamic response. Contributions from the
Ekofisk protective barrier, 2/4 R jacket and connecting model’s resonant response must be separated and
bridges were installed in the Offshore Engineering Basin removed, However, removing the high frequency
at IMD and subjected to waves incident from six components not associated with the model’s resonance
directions, N, NW, W, SW, S, SE as illustrated in Figure will risk underestimating the magnitude of the impact
7. This figure shows the position of the barrier relative load,
to the jacket, assuming the waves always propagate
from the same direction. The jacket model was rotated Closer examination of an impact load was carried out by
on the base dynamometer to follow the position of the isolating one impact from the far field F~Z loads, A
barrier. spectral analysis of this impact, given in Figure 9,
clearly shows a response within the model’s resonant
The jacket model was subjected to a total of 25 frequency 1.5 Hz, The forces measured by the vertical
different regular waves and 18 irregular wave spectra in load cells on the deck were low-pass filtered at 1,0 Hz,
the presence of the protective barrier. These runs The filtered signals are shown in Figure 10, As can be
covered the three subsidence levels. A series of tests seen in the figure, cells C2 and C3 sense the load first,
was also repeated on the 2/4 R alone to investigate the followed by C5 and finally Cl and C4. Refer to Figure 3
effects of the barrier on the wave field and for load cell locations. The time delay between the
consequently on the wave loads, These tests are different load cells suggests that the impact loads on
referred to as the “far field” condition, the deck in the far field are caused by a wave making
contact with the deck and traveling along its underside.
Apart from instrumentation on the jacket model, the The speed at which the force moves along the
barrier was instrumented to measure wave elevations underside of the deck can be determined by comparing
on the outer wall and local wave impact loads at a the phases between loads measured at individual loads
number of locations within the barrier, A number of cells attached to the deck.
wave probes were also placed at locations in the wave
field that correspond to those where the jacket would The part of each signal filtered off from each of the load
be located as it was positioned around the barrier, cell measurements , i.e. the high frequency
During each test run all instrumentation was sampled at components, are compared in Figure 11. These are
200 Hz. Only the results of the wave impact loading on essentially the result of high-pass filtering at 1.0 Hz. As
the 2/4 R deck are discussed in this paper, shown in the figure, all these signals are in phase with
each other indicating that they are not phase-locked to
the components below the 1.0 Hz frequency, The sum
786
—
OTC 7782 J.J. MURRAY, P. KAPLAN AND W.C. YU 5
of the five components shown in Figure 11 is given in mass and added moments of inertia. In the
Figure 12. This figure also compares the vertical inertia actual tests the coverage area and uniformity of
force computed from the global acceleration measured water in contact with the deck during the
by the accelerometers and multiplied by the mass passing of a wave crest is continually changing,
(including added mass estimates). This comparison making an accurate assessment of the non-
shows that the response around 1.5 Hz is caused by constant added mass on the deck difflcuk.
the inertia effect of the deck vibrating. The inertia force
is multiplied by -1.00 for comparison purposes. Method 3 Low-Pass Filter
The analysis concluded that impact loads measured The third method considered was to use a low-
above 1.0 Hz were attributed to the model’s resonance pass non-recursive time domain filter. A Kaiser
response and could be removed, Three techniques to filter [41 was selected with a cut-off frequency
remove resonant effects from the measured signals of 1.0 Hz. Referring to Figure 13, the thin line
were investigated. These were: passing through the decay trace is the output
from the selected filter. This shows all effects
Method 1 Inverse Transfer Function of the deck’s response to the step function have
been removed.
Kaplan [31 presented a procedure using the
model’s impulse response to determine an Figure 14 compares the results of METHOD 2 and
inverse transfer function. The transfer function METHOD 3 when applied to a measured impact
is found from the derivative of the unit step response. The figure shows that the inertia removal
response for a particular degree of freedom. In technique reduces the peak of the load by
the present case this step function was approximately 50%. However, there remains
determined by applying a static load to the considerable ripple in the signal assumed to be caused
model and instantaneously releasing it. by variations in the added mass which in this case is
Following this procedure the experimentally assumed to be constant. The 1.0 Hz filter results show
measured step response of the deck with the a smoother signal with all high frequency components
jacket mounted on the base dynamometer was removed.
measured. The results are shown in Figure 13.
The discussion on vertical load considerations with
The trace shows a modulating frequency regard to model resonance effects applies to the
indicating coupling between two modes having horizontal measurements as well, although the added
close frequencies. The method of inverse mass effects were not as significant as in the vertical
transfer functions is not easily applied when direction. The above described analysis concluded that
coupling is involved unless a matrix form responses around 1.5 Hz were related to model inertia
including all cross coupling terms is included. responses and could be removed to provide smoother
The matrix is extremely difficult, if not time traces of the measurements. All measured data
impossible, to assemble. Some approximations were subsequently low-passed at 1.0 Hz using a Kaiser
were made assuming a single degree of freedom filter,
system but produced very poor results. This
method was abandoned. TEST RESULTS
Method 2 Remove Measured Inertia Forces Effects of Protective Barrier on Surrounding Wave
Elevation
Assuming damping contributions defined in
Equation (1) to be negligible, the intended The protective barrier has a significant effect on the
method of compensating for model resonance wave field around it due to diffraction. Wave diffraction
effects was to estimate inertia forces using patterns around the barrier result in locations that
measured accelerations and subtract them from produce higher and lower wave crest elevations than
the measurement. This technique assumes the that of the incident wave. In linear waves, diffraction
deck to be a rigid plate. Phase differences patterns around a circular cylinder can be readily
between the vertical accelerations measured at predicted of the model of MacCamy and Fuchs [51.
different locations on the deck were checked to However, in intermediate and shallow water conditions
confirm this assumption. The main problem this theory can underestimate the maximum crest
associated with estimating the inertia effects height attained by a wave due to the second-order
using the accelerometers is related to the added contributiorw Murray and WU [61 Presented a stretching
787
——
—
.
6 EXPERIMENTAL AND ANALYTICAL STUDIES OF WAVE IMPACT FORCES ON EKOFISK PLATFORM STRUCTURES OTC 7782
technique to account for these second-order effects by there was no deck impact even at the maximum
applying an enhancement factor to the predictions of subsidence level, presumably due to the diffraction
linear theory. effects from the barrier causing a depression of the
water elevation in the diffraction pattern at the location
Wave diffraction is sensitive to wave period. Therefore of the platform,
incident waves of shorter periods, although smaller in
height than a wave of a longer period, may result in THEORETiCAL ANALYSIS
deck impact whereas the longer wave may not contact
the deck. Also the position of the deck relative to the A theoretical analysis procedure for predicting platform
barrier wail may have an influence on deck impact since deck impact forces has been developed and described
the structure can be placed at a low elevation point in in [71, The method was based upon an extension of the
the diffraction field. procedures described in [3], where the forces arise due
to the sum of inertial effects ( associated with the time
Vertical Impact Loads on 2/4 R Deck in Regular and rate of change of fluid momentum) and drag
Irregular Waves contributions. Both vertical and horizontal plane forces
on deck structures are considered in [7], with major
The effects of jacket position and incident wave emphasis on far field conditions. Further extensions and
direction are illustrated in Figures 15 and 16. The top applications to conditions including the diffracted wave
part of each figure shows the deck elevation and field arising from the barrier structure are indicated
location relative to the barrier wall together with the there, but no detailed description of the analysis is
wave surface elevation caused by diffraction from the provided,
barrier. The ordinates of the graphs show the water
elevation at distances along a line passing through the Correlation between theoretical predicted forces and
centre of the barrier and the geometric centre of the those measured in the present test series shows
jacket. These elevations, computed using the method generally good agreement. The influence of the porous
described in [61, are plotted from the barrier to a deck in reducing the vertical forces is explained in terms
distance 125 m in outward radial directions. of changes in added mass and cross-section area of the
porous region, with the necessary modifications in
The elevations at the wall should represent the wave modelling procedures described in [71.
runup on the barrier at these circumferential locations.
In each figure the surface elevation along a line passing CONCLUSIONS
below the centre of the structure’s deck is given for
particular wave period and direction of incidence, The The model test study discussed in this paper deals
deck elevation and location, relative to the barrier wall, primarily with procedures for measurement and
are indicated in the figure as a solid horizontal line. The interpretation of impact forces, particularly in terms of
bottom part of the figure shows the vertical loads identifying and removing extraneous responses. The
measured in these wave conditions using the solid and paper also discussed the effects of wave asymmetry
porous deck coverings. and its effect on crest height both in the diffracted field
and in the far field.
In general, the level of subsidence and position with
respect to the direction of wave incidence were the The following conclusions result from the analysis:
most significant influences on vertical impact loading on
the 2/4 R deck, The largest vertical deck loads The distribution of crest heights in a particular
measured in regular waves occurred in the water depth spectrum are not Rayleigh distributed even
corresponding to the largest subsidence level in the N when the wave heights are. Crest heights can
and W positions. Wave loading responses on both the be significantly higher than those assumed
deck and jacket in these positions are very similar. This under linear theory,
similarity is expected since the positions are nominally
symmetrical about the wave direction. Figures 15 and The added mass effects during wave impact
16 illustrate the high correlation between crest height significantly reduce the natural frequencies as
as well as position and impact load, compared to those measured in air.
The vertical loads showed a dramatic reduction when Characteristics of vertical wave impact are
the porous deck was used as compared to the solid significantly different in the far field than in the
deck, In most cases the porous deck loads were diffracted field both in shape and magnitude,
significantly reduced, In certain regular wave conditions
788
—
OTC 7782 J.J. MURRAY, P. KAPLAN AND W.C. YU 7
REFERENCES
789
—
Figure 2. Deck Frame
f I
F~z
time time
790
PE
i I I I
Wave Height
PE
I I I I
Crest Height
(n N
Directions
Figure 7.
791
—
.
-+
o 0
0
I
N N x
Lim d Lo
. >
0 0
0
N x
Id”
792
—
,...
f I 1, I I I I I
J 1 I I
N
Ii? N
Ufn
..
1 t 1 r . .
\
793
/’ measured response
Time
Figure 13. Step Response
~
measured load
1.0 Hz filter
FDZ
I
i
Time
794
a-
x
u.
‘r m 0
I I
I18M J~!JM3a
x
u
795