Proceedings of the Twenty-fifth (2015) International Ocean and Polar Engineering Conference www.isope.

org
Kona, Big Island, Hawaii, USA, June 21-26, 2015
Copyright © 2015 by the International Society of Offshore and Polar Engineers (ISOPE)
ISBN 978-1-880653-89-0; ISSN 1098-6189

Scale Effects on Hydrodynamic Manoeuvring Force Prediction

Yuting Jin, Shuhong Chai, Jonathan Duffy, Christopher Chin, Neil Bose
National Centre for Maritime Engineering and Hydrodynamics,
Australian Maritime College, University of Tasmania,
Launceston, Tasmania, Australia

ABSTRACT yaw moment existed throughout the whole range of tested Reynolds
number. Kim et al. (2003) investigated the scale effects in the
This paper investigates the influence of ship model size on the manoeuvring forces of a bare hull DTMB 5415 at steady drift angles
manoeuvring hydrodynamic coefficients utilising the Reynolds using RANS computation. The paper concluded the significance of
Averaged Navier-Stokes solver STAR-CCM+. The hydrodynamic scale effects for large drift angle and presented the discrepancy of
forces and yaw moment on a KRISO Very Large Crude Carrier vortical flow structures on the vessel due to the differences in model
(KVLCC2) model have been predicted for a static drift and a pure sway scale and full scale Reynolds number. Bhushan et al. (2009) applied
test and validated against benchmark model scale test. Good URANS simulations to investigate the hydrodynamics of self-propelled
correlations between the measured and computed force and moment full scale and model scale versions of the DTMB 5415 during zigzag
results were observed. On validating the numerical prediction manoeuvres. Their study compared the trajectories of the vessel
technique, a set of systematic simulations on different size ship models obtained from the simulations and noted the presence of Reynolds
has been carried out for static drift, pure sway and pure yaw tests to effects on the rudder effectiveness and propeller forces.
study the scale effects on the forces and moments obtained from the
captive manoeuvres. From the obtained results, scale effects are found The present study investigates the scale effects on manoeuvring forces
to influence the predictions of surge force and sway force but are not and moments of a rudder appended KVLCC2 vessel utilising URANS
significant in the yaw moment. computational method. Models of the KVLCC2 having different scale
factors and the full scale vessel undertaking captive motions are
KEY WORDS: Reynolds Averaged Navier-Stokes Equation, simulated using the virtual PMM techniques in CFD. Initially, static
KVLCC2, Manoeuvring forces and moments, Scale effects drift and pure sway simulations were undertaken and the results were
compared to the benchmark data provided by Toxopeus et al. (2013)
INTRODUCTION and Stern et al. (2011). Applying the same numerical setup, a series of
systematic simulations of different scaled models and the full scale
Prediction of hydrodynamic ship manoeuvring coefficients is vessel undertaking static drift, pure sway and pure yaw tests are carried
traditionally performed using captive model tests, which involves a out.
series of towing tank experiments utilising a rotating arm or a planar
motion mechanism (PMM). These model scale experimental fluid COMPUTATIONAL METHOD
dynamics (EFD) analyses are known to be vital for designing and
operating ships and their manoeuvring systems. However, due to the The computations were conducted using the RANS solver STAR-
great discrepancies in the Reynolds number when comparing with full CCM+. The code resolves the incompressible RANS equation in
scale conditions, concerns exists about how accurate these model scale integral form utilising finite volume technique. Transient calculations
measurements are when extrapolated to full scale ships. It would be were considered in this study as PMM motions have to be executed in
rather expensive to conduct this type study by means of physical the time domain.
experiments using different sizes of ship model.
The numerical model of the KVLCC2 is shown in Figure 1(a) with a
While the scale effects on ship resistance are taken into account in the body fixed coordinate system located at its centre of gravity. The main
conventional manner by flat plate correlations, the lateral forces and particulars of the vessel are presented in Table 1. In the present
yaw moments obtained by captive model tests generally have not been computations, the entire geometry of KVLCC2 was modelled to take
corrected. Mak and Renilson (1998) performed model testing to study account for asymmetrical flow over the hull due to the prescribed
the effects of Reynolds number on the manoeuvring coefficients for a model motions, which mimic those typically performed with a PMM.
modern full-form vessel. From the obtained results, it was not possible Overset grid technique was employed in the virtual PMM simulations.
to clearly detect significant scale effects in the linear manoeuvring The method required separate mesh enclosing the vessel in an
coefficients. However it was evident that variations of the non-linear overlapping domain and flow information was linearly interpolated

1577
between the background domain and the overset region. mesh transition between the refined hull surfaces and the outer
boundaries. Hexahedral trimmer and surface remesher were adopted to
Table 1: Geometrical properties of KVLCC2 generate global volume mesh as well as local refinements. To resolve
Length between perpendiculars (m) 320.0 the turbulent boundary layer, all y+ wall treatment was utilised as well
as the prism layer mesh, achieving wall y+ value of 25 to 30 along the
Beam (m) 58.0 hull. At free surface, anisotropic trimmer refinement was applied to
Draft (m) 20.80 capture ship generated wave patterns as shown in Figure 1(c).

Displacement (tonnes) 312622.0 BENCHMARK STUDY

Longitudinal centre of gravity forward of mid-ship 11.1
Benchmark model test description
(m)
The SIMMAN 2008 workshop on ship manoeuvring simulation
Moment of inertia about the x-axis (Ixx) (kg·m2) 1.68×1011 methods has provided a great amount of EFD data from captive model
Moment of inertia about the z-axis (Izz) (kg·m2) 1.99×1012 tests conducted by a wide range of institutions and organisations. In
order to validate the computational setup, the static drift test on a 1:
45.714 bare hull KVLCC2 model and a pure sway test on a 1:58 rudder
appended KVLCC2 model has been employed. During the test, the
vessel was set to be free for heave and pitch motions but constrained in
roll. The static drift and pure sway motion input parameters for the
PMM carriage are presented in Table 2 and Table 3. The tests were
carried out at Froude number of 0.142 representing realistic KVLCC2
operational speed of 15.5 knots (Stern et al., 2011).

(a) Table 2: Benchmark static drift test conditions

Bare hull
Static Drift Test Condition Heave and pitch unconstrained
Roll constrained
Froude Number (Fn) 0.142
Surge Velocity (U0) (m/s) 1.046
Static drift angle (β) (deg) 4

Table 3: Benchmark pure sway test conditions

Rudder appended
Pure Sway Test Condition Heave and pitch unconstrained
Roll constrained
(b)
Froude Number (Fn) 0.142
Surge Velocity (U0) (m/s) 1.046
Sway Amplitude (A) (m) 0.86
Sway Velocity (V) (m/s) 0.089
Sway/Surge Velocity Ratio (v') 0.0852
Sway Frequency ω (Hz) 0.0165

Benchmark simulation results

Comparisons were made between the computational and model test
(c) results for non-dimensionalised surge force X', sway force Y' and yaw
Figure 1: Overview of KVLCC2 geometry, computational domain and moment N'. The non-dimensional values were obtained by the
mesh distribution reduction equations (Equation (1) to Equation (3)).
FX
The dimensions of the control volume are illustrated in Figure 1(b). X' (1)
The background computational domain was designed to be
0.5UU 0 2T L
considerably large to overcome any potential blockage effect. The mass FY
centre of the vessel was located at an initial position of (0, 0, 0) in the Y' (2)
0.5UU 0 2T L
earth-fixed global coordinate system. The boundary conditions used
were velocity inlet at the flow inlet, lateral walls and domain top and MZ
N' (3)
bottom, pressure outlet at the flow outlet and non-slip wall at the hull 0.5UU 0 2T L2
surfaces. Damping boundaries were employed to minimise numerical
inaccuracies caused by any reflected free surface waves. For the benchmark static drift simulation, reasonable accuracy was
achieved for surge force, sway force and yaw moment as shown in
To generate mesh, slow cell growth rate was selected to create smooth Table 4.

1578
Table 4. Computational results of benchmark static drift test against
X' Y' N'
model test data
S3-S2 3.41 -2.14 -2.65
Condition Model Test (D) CFD (S) Difference % Variation (%)
S2-S1 -2.22 1.45 -0.88
Static Drift X' -0.0162 -0.0176 8.6
Scale factor pG Oscillatory Oscillatory 3.2
Y' 0.0159 0.0140 11.9
λ=45.71
UG 0.00030 0.00015 0.00020
N' 0.0116 0.0113 2.7
UG (% of S1) 1.67 1.09 1.75
The benchmark pure sway results are presented in Figure 2. The CFD
simulation performed precise prediction for the non-dimensional surge To validate the numerical simulation results against the EFD data, it is
force X' and yaw moment N' with average discrepancies of 3.7% and necessary to check if the absolute comparison error E is smaller than
4.5% respectively. A reasonable numerical accuracy of 11.1% was the validation uncertainty UV calculated using Equation (4), U D is the
achieved for the average sway force. The total average comparison
experimental uncertainty.
error Eave was 6.4%.
UV U G2  U D2 (4)

The experimental results and data uncertainties obtained from

Toxopeus et al. (2013) are presented in Table 7. It can be seen that the
maximum discrepancy between the computations and experimental
measurements is around 10% which is reasonably accurate. However,
only the yaw moment calculation has been validated. The numerical
errors for the surge force and sway force are more or less 2% off from
the experimental uncertainties. However, taking into account that
overcoming this 2% difference would possibly require a change in grid
resolution from medium to fine, which is likely to double the
computational time, it was decided that the 3.7 million mesh size was
suitable for the following computations. This is also due to the present
study only focusing on performing systematic calculations and
Figure 2. Non-dimensional forces and moments from pure sway test for obtaining relatively large data sets.
KVLCC2
Table 7. Validation of numerical results against EFD data
Verification and validation study
D UD UV UV %D
A verification and validation (V&V) study was performed for the X' -0.0162 0.000891 0.000940 5.8
benchmark static drift computation in order to estimate the numerical EFD
Y' 0.0159 0.001574 0.001581 9.9
uncertainties within the simulation setup. The V&V study followed the
procedures provided by Wilson et al. (2001) and Stern et al. (2001). N' 0.0116 0.000696 0.000724 6.2
Mesh sensitivity was of particular interest. Three different grid sizes
S E=S-D E %D
adopting refinement ratio rG 2 were employed in order to evaluate
X' -0.0176 -0.0014 - 8.6
the numerical uncertainties within the obtained force and moment data.
The non-dimensionalised forces and moments for each different grid Y' 0.0140 -0.0019 - 11.9
CFD
condition are presented in Table 5. N' 0.0113 -0.0003 - 2.7
Table 5. Non-dimensionalised hydrodynamic forces and moments with SCALE EFFECTS ON MANOEUVRING FORCES AND MOMENTS
different grids for static drift simulation, β=4°, Fn=0.142
Mesh Condition Mesh Size X' Y' N' Based on the validated numerical modelling techniques, scale effects
on the manoeuvring forces and moments of KVLCC2 were
Coarse (3) 2516240 -0.0182 0.0137 0.0110 investigated. Four different size ship hulls (λ=1, 58, 100 and 225)
Medium (2) 3659369 -0.0176 0.0140 0.0113 undergoing prescribed static drift, pure sway and pure yaw motion have
been simulated. Comparisons were made between the non-dimensional
Fine (1) 5019853 -0.0180 0.0138 0.0114 surge force, sway force and yaw moment for different size ship models.

Referring to Simonsen et al. (2012), the grid uncertainty U G of Scale effects in static drift
computations with unstructured mesh can be estimated based on On validating numerical simulation techniques, systematic static drift
Richardson extrapolation. The estimated order of accuracy pG and grid computations on different size rudder appended KVLCC2 hulls (see
Table 8) were performed and are presented in this section. The
uncertainty are shown in Table 6. It shows the grid uncertainty of β=4° hydrodynamic forces and yaw moment are non-dimensionalised by the
for static drift simulation, ranges from 1.09% to 1.75% of the fine grid reduction relationships given in Equations. (1) to (3) and have been
solution. Both X' and Y' reached oscillatory convergence while N' compared for different size KVLCC2 hulls as illustrated in Figure 3.
achieved an order of accuracy of 3.2.
Table 6. Estimated order of accuracy and grid uncertainty of β=4° for
static drift simulation

1579
Table 8. Static drift tests for different model hull sizes significant influence on the sway force coefficients during static drift
tests. However, due to the large numerical uncertainty detected from
Parameter Value
the verification and validation study (11.9%), it can only be stated that
λ 1 58 100 225 scale effects are marginally present in the sway force. For the yaw
moment (Figure 3(c)), scale effects are shown to be negligible as the
U0 (m/s) 7.97 1.05 0.80 0.53 captured differences are within the numerical uncertainty limit.
9 6 6
Re 2.86×10 6.48×10 2.86×10 8.48×105
Scale effects in pure sway
β (deg) 0, 4, 8, 12, 16 The pure sway tests were performed for four different size rudder
appended KVLCC2 hulls to study possible scale effects in this
manoeuvre. The computational setup conditions are presented in Table
9. The surge force, sway force and yaw moment are computed as a
function of time as shown in Figure 4. Figure 4(a) demonstrates
significant differences in the surge force for different model scale
computations. The calculated non-dimensional surge force of the 1:225
KVLCC2 model is more than twice of that derived from the full scale
computation. This is consistent with the surge force data gathered from
the static drift computations as discussed previously. These observed
variations are due to the difference in Reynolds number between the
different size vessels. For model scale conditions, the viscous force
components contribute to a greater proportion of the total surge force
on the vessel. These viscous components do not follow the Froude
scaling law but are related to Reynolds number. Therefore, the non-
(a) dimensionalised model scale surge force will not be able to fully
represent full scale conditions without correlation. For the sway force, a
similar trend can be found in Figure 4(b) between different hull sizes.
There is a 27.4% discrepancy found between the peak sway force for
the 1:225 KVLCC2 model and the full scale vessel. Again, the
magnitude of the force is smaller at full scale, but the difference is not
as great as that observed in the surge force comparison. The yaw
moment on each of the different size KVLCC2 are presented in Figure
4(c). As observed, there is no significant scale effect that exists in this
hydrodynamic component.

Table 9. Froude scaling of pure sway test conditions

Parameter Value

(b) λ 1 58 100 225

Fn 0.142 0.142 0.142 0.142
9 6 6
Re 2.86×10 6.48×10 2.86×10 8.48×105
U0 (m/s) 7.96 1.05 0.80 0.53
A (m) 50.0 0.86 0.50 0.22
V (m/s) 0.67 0.089 0.068 0.045
v’ 0.0852 0.0852 0.0852 0.0852
ω (Hz) 0.0022 0.0165 0.0216 0.0324

(c)
Figure 3. Non-dimensional hydrodynamic forces and yaw moment
from static drift computations for different size KVLCC2 hulls: (a)
surge force; (b) sway force; (c) yaw moment

From the surge force comparison as shown in Figure 3(a), the smallest
hull experiences the largest surge force. From Figure 3(b), scale effects
are evident in the sway forces of the static drift computations. A
maximum percentage difference of 21.3 is found at β=4° between the
1:225 model and the full scale vessel. As the drift angle increases, scale
effects become more significant and lead to greater discrepancies in the
sway force magnitude. This agrees with the experimental outcomes (a)
from Mak and Renilson (1998), showing scale effects are likely to have

1580
 Max yaw angle
(ψ) (rad) 0.55 0.55 0.55 0.55
A (m) 360 6.21 3.60 1.60

(b)

(a)

(c)
Figure 4. Non-dimensional hydrodynamic forces and yaw moment
from pure sway computations for different size KVLCC2 hulls: (a)
surge force; (b) sway force; (c) yaw moment

Scale effects in pure yaw (b)

In this section, different size rudder appended KVLCC2 hulls
undergoing prescribed pure yaw motion with the same Froude number
and constant non-dimensionalised yaw rate were performed. The pure
yaw test conditions based on Froude scaling are shown in Table 10.

The non-dimensional forces and moment have been predicted for a

pure yaw period as illustrated in Figure 5. The surge force was
relatively larger for the pure yaw computations when compared with
pure sway and static drift simulations. Large discrepancies in the surge
force for different size hulls are observed from Figure 5(a) indicating
the existence of scale effects. For the maximum sway force in Figure
5(b), a 25% difference is found in the peak force between the 1:225
model scale and the full scale case. No significant scale effects are
observed in the yaw moment as the differences are within numerical (c)
uncertainty limits, seen from Figure 5(c). Figure 5. Non-dimensional hydrodynamic forces and yaw moment
from pure yaw computations for different scale KVLCC2 hulls: (a)
Table 10. Pure yaw test conditions for four different scale factors surge force; (b) sway force; (c) yaw moment

Parameter Value CONCLUDING REMARKS

λ 1 58 100 225
Non-dimensionalised Based on the validated CFD model, a numerical study was carried out
yaw rate (r') 0.3 0.3 0.3 0.3 to investigate the scale effects in the static drift, pure sway and pure
yaw manoeuvres. A verification and validation study was first
Fn 0.142 0.142 0.142 0.142 conducted for one static drift test case to quantify numerical
Re 2.86×10 9
6.48×10 6
2.86×10 6
8.48×105 uncertainties for performed computations. Systematic computations
were executed to calculate manoeuvring forces and yaw moment for
U0 (m/s) 7.97 1.05 0.80 0.53 four different scale KVLCC2 hulls undergoing prescribed motions, as
Max yaw rate typically performed by planar motion mechanism. Variations in the
max(rPMM) (rad/s) 0.0075 0.0569 0.0747 0.1121 non-dimensional forces and moment were quantified. From the
Yaw frequency obtained results, the following statements can be made:
(ω) (Hz) 0.0022 0.0165 0.0216 0.0324

1581
x The surge force derived from the performed static drift, pure sway REFERENCES
and pure yaw computations were significantly influenced by the
scale effects. Bhushan, S., Xing, T., Carrica, P., & Stern, F. (2009). Model- and Full-
Scale URANS Simulations of Athena Resistance, Powering,
x Scale effects are also evident in the sway force predictions,
Seakeeping, and 5415 Maneuvering. Journal of Ship Research,
introducing 20%-30% difference in the numerical prediction 53(4), 179-198.
between the KVLCC2 model scale and full scale condition. Kim, K. H., Gorski, J., Miller, R., Wilson, R., Stern, F., Hyman, M., &
x Yaw moment was not greatly affected by the variation of Burg, C. (2003, 9-13 June 2003). Simulation of surface ship
Reynolds number throughout the performed computations. dynamics. Paper presented at the User Group Conference, 2003.
Proceedings.
x Hence, based on the cases investigated in this study, care must be
Mak, T., & Renilson, M. R. (1998). Investigation into scale effects on
taken when extrapolating surge force and sway force from captive static hydrodynamic manoeuvring coefficients for a full-form modern
model scale manoeuvring tests to full scale. Correlation vessel: [Launceston, Tas.] : Australian Maritime Engineering CRC
techniques are necessary to account for the scale effects. Ltd. [1999].
Simonsen, C. D., Otzen, J. F., Klimt, K., Larsen, N. L., & Stern, F.
In future work, more manoeuvres such as static rudder, static drift and (2012). Maneuvring predictions in the early design phase using CFD
rudder, yaw and rudder, yaw and drift will be computed for different generated PMM data. Paper presented at the 29th Symposium on
scaled models to investigate possible Reynolds number impact on the Naval Hydrodynamics, Gothenburg, Sweden.
manoeuvring coefficients. Computations of free running manoeuvres Stern, F., Agdrup, K., Kim, S. Y., Hochbaum, A. C., Rhee, K. P.,
such as turning circle and zigzag motions for different scaled models Quadvlieg, F., . . . Gorski, J. (2011). Experience from SIMMAN
can be carried out for better illustration of scale effects on ship 2008-The first workshop on verification and validation of ship
manoeuvring performance. maneuvering simulation methods. Journal of Ship Research, 55,
135-147.
ACKNOWLEDGEMENTS Stern, F., Wilson, R. V., Coleman, H. W., & Paterson, E. G. (2001).
Comprehensive Approach to Verification and Validation of CFD
The authors acknowledge the scholarship support provided by the Simulations—Part 1: Methodology and Procedures. Journal of
Australian Maritime College, University of Tasmania. Fluids Engineering, 123(4), 793-802. doi: 10.1115/1.1412235
Toxopeus, S. L., Simonsen, C. D., Guilmineau, E., Visonneau, M.,
Xing, T., & Stern, F. (2013). Investigation of water depth and basin
wall effects on KVLCC2 in manoeuvring motion using viscous-flow
calculations. Journal of Marine Science and Technology (Japan),
18(4), 471-496. doi: 10.1007/s00773-013-0221-6
Wilson, R. V., Stern, F., Coleman, H. W., & Paterson, E. G. (2001).
Comprehensive Approach to Verification and Validation of CFD
Simulations—Part 2: Application for Rans Simulation of a
Cargo/Container Ship. Journal of Fluids Engineering, 123(4), 803-
810. doi: 10.1115/1.1412236

1582