Beruflich Dokumente
Kultur Dokumente
Journal of Offshore Mechanics and Arctic Engineering AUGUST 2016, Vol. 138 / 041801-1
C 2016 by ASME
Copyright V
Journal of Offshore Mechanics and Arctic Engineering AUGUST 2016, Vol. 138 / 041801-3
Fig. 5 Meshed fluid domain with boundaries labeled for dynamic simulation
used in the present study is shear stress transport (Menter) k–x, which is a supercomputing facility available at Indian Institute of
which is widely used in similar studies by other researchers [7]. Technology, Madras. The simulations are executed in parallel on
Though Reynolds stress models could be a better choice while eight Intel E5-2670 8 C 2.6 GHz processors of the supercluster.
considering prediction of complex flow phenomena, these models An average simulation time of 24 hrs is observed for one oscilla-
are less robust and computationally expensive [8]. All the yþ tion cycle for dynamic PMM modes.
near-wall treatment is used for all the simulation, which uses the
blended wall law (yþ < 5), if the mesh is fine enough, and if the 3.2.1 Results and Discussion. The forces and moments are
mesh is coarse enough (yþ > 30), the wall law is equivalent to a computed with respect to the ship-fixed coordinate system in surge,
logarithmic profile. The free surface modeling is done using mul- sway, and yaw directions. The forces and moment time-history plots
tiphase flow model with the volume of fluid method, which cap- in 4DOF are obtained by performing dynamic (PMM) simulations
tures the movement of the interface between the fluid phases. The in pure yaw mode. It has been observed that the time history of the
high-resolution interface capturing convective discretization hydrodynamic forces has initial one to two cycles in the transient
scheme is used to track the air water interface. The time step is region [23], and hence, it is advisable to analyze the third or subse-
carefully chosen for the simulations and is based on solution sta- quent cycle to derive the hydrodynamic derivatives. This cycle of
bility as well as the grid motion per time step. Though an implicit force history is represented with a Fourier series. Equating this to
method of solution is used, time steps should be small enough so the right-hand side of the equations of motion (Eqs. (5)–(8)), expres-
that they do no lead to mesh failure during mesh motions for dif- sion for the derivatives is derived in terms of Fourier coefficients.
ferent modes. A convergence criterion of residual drop by order Subsequent evaluation and substitution of these coefficients in the
of 103 for internal iterations is chosen. The solution is termed above derived expressions yield the value for the respective hydro-
converged when there is a repetitive pattern for the forces and dynamic derivatives. The results reported here are those from the
moments on the ship hull over the oscillation cycles. The fluid- third cycle and are shown in Fig. 6.
domain dimensions, boundary conditions, and solver settings cho- The numerically obtained hydrodynamic derivatives are nondi-
sen for the simulations are the same as that used for the pure sway mensionalized using appropriate factors (Table 2) and compared
mode in dynamic simulations [21]. The meshed body along with
the specified boundary conditions is shown in Fig. 5. The cell
count of these meshes is approximately 2.62 106. Table 2 Hydrodynamic derivative results from dynamic simu-
The dynamic ship motion in pure yaw mode is numerically simu- lation in pure yaw mode
lated by assigning appropriate yaw rotation rate with respect to the
Nondimensionalized Numerical
ship-fixed coordinate system. A motion along the sway direction of hydrodynamic Nondimensionalization Experimental (present
the fluid domain with respect to the earth-fixed coordinate system is derivatives factor [20] CFD)
also given additionally, so that the ship maintains its centerline tan-
0
gent to the path. To accomplish the motions involved in dynamic sim- Xu_ 0:5qL3m 0.000238 0.0000072
ulation, a morphing procedure is utilized, which redistributes the 0
Xuu 0:5qL2m 0.000423 0.000758
mesh vertices at each time step to accommodate the motion of the 0
Xrr 0:5qL4m 0.0002 0.000065
ship. The motion is assigned to the ship model and inner cylindrical 0
Yr_ 0:5qL4m 0.000353 0.000012
domain by defining a translational velocity superimposed with angu- 0
Yr 0:5qL3m Um 0.00242 0.0025
lar velocity, which is expressed in terms of predefined field functions 0
available in the CFD tool. The PMM oscillation frequency is chosen Yrrr 0:5qL5m =Um 0.00177 0.0013
0
such that the nondimensionalized frequency, x0o ðx0o ¼ xo ðLm =Um ÞÞ, Nr_ 0:5qL5m 0.000419 0.00044
0
lies in the range of 2–4 [23] and based on the calculations is taken as Nr 0:5qL4m Um 0.00222 0.0022
0
0.9448 rad s1 with a sway amplitude of 0.3 m. The yaw rate (r) is Nrrr 0:5qL6m =Um 0.00229 0.00097
found to be 0.145 rad s1 as per the equation in Fig. 3.
0
Kr 0:5qL4m Um 0.000063 0.000024
R
All the simulations in the present work are run on the VIRGOV 0
Krrr 0:5qL6m =Um 0.000046 0.000025
supercluster of the high-performance computing environment,
Journal of Offshore Mechanics and Arctic Engineering AUGUST 2016, Vol. 138 / 041801-5
Method of determination
Nondimensionalized hydrodynamic derivatives Experimental [20] Value Error (%) Value Error (%) Value Error (%)
0
X uu 0.00042 0.00078 84.6 0.00075 79.4 0.00049 16
Fig. 8 Meshed fluid domain with boundaries labeled for static simulation
Journal of Offshore Mechanics and Arctic Engineering AUGUST 2016, Vol. 138 / 041801-7
0 0 0
Fig. 12 Plots to determine Yr uu ; Nr uu ; and Kr uu
0 0 0
derivatives, namely, Yrru ; Nrru , and Krru . Two different values of
yaw rate, 0.12 rad/s and 0.17 rad/s, are chosen for performing the
simulations which are based on the yaw rate(r) equation given in Table 4
0 0
Experimental and numerical values of Yr uu , Nr uu , and
Fig. 3. Separate meshes are generated for heel angle, u ¼ 0 deg to 0
Kr uu
10 deg, in steps of 2 deg. The fluid-domain dimensions,
boundary conditions, and solver settings chosen for the simula- Nondimensionalized Experimental Numerical
tions are the same as that used for the pure yaw mode in dynamic hydrodynamic derivatives [20] (present CFD)
simulations (Sec. 3.2). The motion parameters are calculated as 0
per the yaw rate considered for each set of simulation. The model Yruu 0.001368 0.0147
0
in a heeled position and the fluid domain surrounding it (inner Nruu 0.00242 0.0029
0
domain) are subject to oscillations in pure yaw mode. Twelve Kruu 0.000036 0.0002
meshes (six heel angles for each yaw rate and two yaw rates
0 0 0 0
Fig. 14 Plots to determine Xuu ; Yu ; Nu ; and Ku from pure yaw with roll simulations
considered) are generated and used for the CFD analysis. The cell
0 0
Table 5 Experimental and numerical values of Xuu ; Yu ; Nu ; and
0
count of these meshes is approximately 2.63 106.
The hydrodynamic derivatives Yu0 ; Nu0 , and Ku0 are obtained
0
Ku
from each set of simulations at different yaw rates. The procedure
Nondimensionalized Experimental Numerical for obtaining the hydrodynamic derivatives is as mentioned in
hydrodynamic derivatives [20] (present CFD) Sec. 3.4.2. The values of forces and moments at zero yaw rate
0
after two cycles of operation are taken at each heel angle. These
Xuu 0.0002 0.0008 forces and moments are nondimensionalized using appropriate
0
Yu 0.00006 0.0006 factors and plotted against heel angles as shown in Fig. 15. The
0
Nu 0.00014 0.0003 points are fitted with straight line in case of sway, yaw, and roll to
Ku
0
0.000021 0.0007 obtain the hydrodynamic derivatives, Yu0 ; Nu0 , and Ku0 . The fitted
equations of the force/moment curves are differentiated once, and
Journal of Offshore Mechanics and Arctic Engineering AUGUST 2016, Vol. 138 / 041801-9
0 0 0
Fig. 16 Plots to determine Yrr u , Nrr u , and Krr u
0 0
Table 6 Experimental and numerical values of Yrr u , Nrr u , and
0
Krr u
Nondimensionalized Numerical
hydrodynamic Nondimensionalization Experimental (present
derivatives factor [20] CFD)
0
Xvr 0:5qL3m 0.00311 0.00036
0
Yvrr 0:5qL4m =Um 0.0405 0.048
0
Yvvr 0:5qL3m =Um 0.0214 0.0181
0
Nvrr 0:5qL5m =Um 0.00156 0.0043
0
Nvvr 0:5qL4m =Um 0.0424 0.0102
0
Kvrr 0:5qL5m =Um 0.001057 0.001
0
Kvvr 0:5qL4m =Um 0.000558 0.0019
Journal of Offshore Mechanics and Arctic Engineering AUGUST 2016, Vol. 138 / 041801-11
Turning circle test parameters (m) Experiment [20] Value Error (%)
derivative which has high impact on the maneuvering characteris- zag maneuver are simulated by solving the equations of motion
0 R
tics of the ship [20] in combined sway–yaw mode is Nvvr and using MATLABV.
0 0
those having moderate influence are Yvrr and Yvvr . These deriva-
tives have been predicted satisfactorily. The percentage deviation 4.1 Results and Discussion. Nineteen derivatives are obtained
0
of the derivative, Nvvr , from the experimental values is more from the sway velocity-dependent derivatives obtained from
(76%). However, Sakamoto et al. [11] reported similar trends of dynamic simulations in pure sway mode and from static simula-
percentage deviation for the particular derivative. Though the tions with roll [21]. These 19 derivatives along with the 23 deriva-
0
hydrodynamic derivatives like Kvvr deviate more, the sensitivity tives obtained from dynamic simulations in pure yaw mode,
analysis shows that it has a small effect on ship maneuvering. yaw–roll coupled mode, and the combined sway–yaw mode and
Considering the complexity of the motions in the combined 0
Xuu obtained from static simulation from present study are used to
sway–yaw motion mode of PMM and noting the results from the simulate the standard maneuvering tests. The tests are also per-
sensitivity analysis, the results are encouraging. formed using all derivatives obtained by the experimental meth-
ods for comparison. A comparison of parameters of the turning
4 Simulation of Trajectory circle tests as well as zig-zag tests simulated using derivatives
obtained by both experimental [13] and CFD methods is shown in
The maneuvering capabilities of the vessel are studied by
Figs. 21 and 22 and also compared in Tables 8 and 9. The results
simulating standard maneuvering tests like the turning circle and
of the various test parameters match with reasonable accuracy
zig-zag tests. Turning tests are performed to understand the ves-
with the experiments for both turning circle tests and zig-zag
sel’s steady turning ability and course changing characteristics.
maneuver.
Figure 19 shows the details of the turning circle tests which are
characterized by standard measures like steady turning radius, tac-
tical diameter, and advance and transfer. Zig-zag tests are per- 5 Summary and Conclusions
formed to learn about the control characteristics, course keeping
ability, and yaw checking ability. Figure 20 shows the details of This paper presents the study of maneuvering qualities of a
the zig-zag maneuvers which are characterized by the principal container ship (S175) where the hydrodynamic derivatives
numerical measures, namely, time to reach second overshoot, appearing in the equations of motion of the mathematical model
overshoot yaw angle, and over shoot width of path [19].The turn- proposed by Son and Nomoto [13] are numerically estimated.
ing circle tests at rudder angle, d ¼ 35 deg and 20 deg/20 deg zig- RANSE-based CFD approach is used for the determination
Journal of Offshore Mechanics and Arctic Engineering AUGUST 2016, Vol. 138 / 041801-13