Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
OMAE2016-54839
to define characteristics loads for the accidental limit state. 2 3070 6668 17.5 15.5 11.5 27.0 17.0
Figure 1 Relative-wave probe locations Table 1 Overview of the identified wave events
The next diagrams (in Figure 11) show, the best matches
between two subsets of selected measured events (no.: {3, 8,
10, 13} and {6, 9}) from Table 1 and the numerically simulated
wave at two different locations. The locations are given in
normalized distance where the normalization factor is the
diameter of the columns of the platform, i.e. xp = 2xD means
that the platform is moved with 2 diameters upstream from its
original position (see Figure 10). One cannot see perfect match
between the measured time series and the numerical time
series. However, the range of the differences is similar to the
variability that can be observed between the repetitions of the
same breaking wave events in the model test [8]. Based on this
observation we assume that the numerical match of the
measured wave events can be used in order to define the
similar, relevant physics of the green water event numerically
as one can observe in the model test.
The relative-wave levels from the measurements at
locations 7, 8, 9, 10 and 11 (Figure 1) are compared to CFD
results in Figure 12, using the numerical waves from Figure 11. Figure 11 Examples of the best matches between measured
We have monitored the same locations as in model test for wave events and the numerical wave (red l.)
direct comparison (still, we need to recall that TLP motion is
not included in the simulation). We see that for probes no. 7 & model tests).
8 near the column H10, similar levels are simulated as in the In order to estimate the effect of the mesh resolution on the
measurements, and the time series shape are also more or less global forces one simulation of a green water event was
similar with some exceptions. Also for points no. 9 11 there restarted with a finer mesh around the module. The name and
are similar high levels as in the measurements, but the position of the surface which are used in the convergence
systematically somewhat lower (notice that 9 & 10 are in the studies are shown in Figure 13. The comparison of the curves is
shadow of the module, so the CFD results for those probes presented in Figure 14. We can see less difference (less than
must be expected to be somewhat truncated relative to the 10%) between the curves than observed for the time step
convergence study. One can see the same tendency at the local
pressure peaks show in Figure 15. The peak value is about 10
% larger with the finer mesh.
The global forces time series are recorded during
simulations with a time step of 4 ms. Restarting the simulation
before the green water event with a time step of 1 ms is used to
estimate the influence of the time step on the global forces. This
time step is 2.5 times smaller than the time step, which was
used in [3] where the numerically estimated local and global
slamming forces are successfully validated against model test.
One can see similar maximum water particle velocity during
Figure 10 Illustration of the relative location the impact in [3] as in the simulations presented in this paper.
Further, the size of the platforms is also similar. Therefore, we
assume that this time step is sufficient to capture the slamming
event. As one can see in Figure 16, the reduction of the time
The structural response by means of Von Mises stress and Figure 22 The free surface elevation around the platform
vertical displacement is illustrated in Figure 24 and Figure 25.
The computed reaction forces at the location of the module
support points and at the connection points, where the main
frame support structure is mounted to the existing top side were
exported and delivered to our industrial partner (Aibel) for
further analysis of constructional and design details of the main
frame support structure.
Figure 20 Time history of integrated vertical load acting on Figure 23 Pressure distribution at the walls of the module
the platform structure
CONCLUSION
In this paper a new procedure is presented where CFD
generated synthetic numerical waves can be directly linked to
the extreme events observed in model tests. This means that the
simulated extreme events can be related to a probability of
occurrence, found from the model test.
The procedure depends on existing model test data. But the
size of the module in model scale makes the experimental
measurement of the acting forces on the modules challenging or
impossible. This combination of CFD with experiments offers
new possibilities for the design of structures subject to risk of
green water loading.
The computational time of running one complete CFD
simulation with 50 seconds duration (solution time) is about 12
hours using 96 parallel processors (threads) on a cluster. In
Figure 21 Velocity vectors at a cut at the approximately order to generate the ABAQUS input files the CFD simulations
center of the module 1 platform structure are restarted just before the impact with smaller time step and
finer mesh. The duration of the restarted simulations is about 5
ACKNOWLEDGMENTS
Statoil is gratefully acknowledged for the permission to
publish this paper.
REFERENCES