Sie sind auf Seite 1von 44

SIMMAN 2014

SHALLOW-WATER CFD COMPUTATIONS

00_066 WL Rapporten

Vlaamse overheid
SIMMAN 2014
Shallow-water CFD Computations

Van Hoydonck, W.; Delefortrie, G.; Eloot, K.; Peeters, P.; Mostaert, F.

August 2015

WL2015R00_066_3

fhr.cls
2015/08/18 – version 0.14
This publication must be cited as follows:

Van Hoydonck, W.; Delefortrie, G.; Eloot, K.; Peeters, P.; Mostaert, F. (2015). SIMMAN 2014: Shallow-water
CFD Computations. Version 3.0. WL Rapporten, 00_066. Flanders Hydraulics Research: Antwerp, Belgium.

Waterbouwkundig Laboratorium

Flanders Hydraulics Research

Berchemlei 115
B-2140 Antwerp
Tel.+32 (0)3 224 60 35
Fax +32 (0)3 224 60 36
E-mail: waterbouwkundiglabo@vlaanderen.be
www.watlab.be

Nothing from this publication may be duplicated and/or published by means of print, photocopy, microfilm or
otherwise, without the written consent of the publisher.

fhr.cls
2015/08/18 – version 0.14
Document identification

Title: SIMMAN 2014: Shallow-water CFD Computations

Customer: Waterbouwkundig Laboratorium Ref.: WL2015R00_066_3

Keywords (3-5): CFD, shallow water, SIMMAN 2014

Text (p.): 31 Appendices (p.): 5

Confidentiality: ☐ Yes Exceptions: ☐ Customer

☐ Internal

☐ Flemish government

Released as from: August 2015


☒ No ☐ Available online

Approval

Author Reviser Project Leader Research & Head of Division


Consulting Manager
Van Hoydonck, W Delefortrie, G. Delefortrie, G. Peeters, P. Mostaert, F.

Eloot, K.

Revisions

Nr. Date Definition Author(s)

1.0 01/07/2015 Concept version Van Hoydonck, W.

2.0 06/08/2015 Substantive revision Delefortrie, G.; Eloot, K.

3.0 13/08/2015 Final version Van Hoydonck, W.

Abstract

This report contains an overview of the computational work done for the 2014 SIMMAN workshop. It is based on
a paper submitted for the workshop proceedings and a poster that was presented at the workshop. Three test
cases (1b-1, 1b-3 and 1c-1) were selected for CFD computations. Of these, the latter two have a drift angle of 4
degrees, for case 1b-1, the ship sails straight ahead. A grid convergence study was executed for case 1c-1. All
computations were carried out blind (so without knowing the experimental values).
In addition to the contents of the paper, this report contains an analysis of the experimental data used as
a reference, a comparison of the CFD result with the EFD results, and some flow visualisations that were
not included in the paper due to page limitations. Apart from the lateral force for case 1b-1 (without drift), results
are in very good agreement with the experimental data.

F-WL-PP10-1 Versie 04
GELDIG VANAF: 12/11/2012
SIMMAN 2014: Shallow-water CFD Computations

Contents

1 Introduction ......................................................................................................... 1

2 Analysis of experimental data .................................................................................... 2


2.1 Rail alignment corrections .................................................................................... 2
2.2 Averages and standard deviations for case 1b-3 .......................................................... 4
2.3 Averages and standard deviations for case 1b-1 .......................................................... 5
2.4 Averages and standard deviations for case 1c-1 .......................................................... 6
2.5 Time history analysis .......................................................................................... 6

3 Computational Setup .............................................................................................. 10


3.1 Test conditions ................................................................................................. 10
3.2 KVLCC2 CAD Model .......................................................................................... 11
3.3 Domain Size and Ship Position .............................................................................. 11
3.4 Topology and General Refinement Settings ................................................................ 12
3.5 Solution Method and Flow Settings .......................................................................... 13

4 Grid convergence study for case 1c-1 ........................................................................... 14


4.1 Base meshes ................................................................................................... 14
4.2 Generation of derived meshes ............................................................................... 14
4.3 Results .......................................................................................................... 15

5 Cases 1b-1 and 1b-3 .............................................................................................. 18


5.1 Setup ............................................................................................................ 18
5.2 Results .......................................................................................................... 18

6 Comparison of CFD and EFD results............................................................................ 22

7 Flow visualisations ................................................................................................. 24


7.1 Water surface................................................................................................... 24
7.2 Hydrodynamic pressure on the hull and tank............................................................... 24

8 Conclusions ......................................................................................................... 30

References .............................................................................................................. 31

A Additional time analysis plots of the EFD data ................................................................. 32

B Additional views of the grid using cutplanes .................................................................... 33

Final version WL2015R00_066_3 iv


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

List of Figures

Figure 1 Fore and aft rail alignment corrections for T0Z02A03 and T0Z03A03 tests. T0Z03A03
shows suspicious oscillations during the first 30 meter and the last 10 meter of the test.... 4
Figure 2 Thrust versus torque for test cases T0Z02A03_C4700x. The numbers next to the black
points denote the last number of the test identification (see Table 1). ......................... 5
Figure 3 Thrust versus X-force for test cases T0Z02A03_C4700x. The numbers next to the black
points denote the last number of the test identification (see Table 1). ......................... 5
Figure 4 Measured sinkages for test T0Z03A03_C47000 showing multiple time intervals with con-
stant measured value................................................................................. 7
Figure 5 Convergence of total X-force for test T0Z02A03_C40000. ...................................... 8
Figure 6 Convergence of propeller torque for test T0Z02A03_C40000. .............................. 8
Figure 7 Convergence of total Y-force for test T0Z02A03_C40000. ...................................... 9
Figure 8 Convergence of midship sinkage for test T0Z02A03_C40000. ................................. 9

Figure 9 Bottom, front, back and isometric view of the Parasolid CAD model of the KVLCC2 as
used for the CFD computations. .................................................................... 11
Figure 10 Top view of domain box with dimensions and global axis system orientation. ................ 12
Figure 11 Overview of refinement boxes and refinement surfaces with absolute target cell sizes. .... 13

Figure 12 View of the grid near the stern in the symmetry plane of the ship for the six different cases. 15
Figure 13 Iterative convergence and final values for , and for 1c-1. ............................ 16
Figure 14 Iterative convergence and final values for sinkage, pitch and for case 1c-1. ............ 17

Figure 15 Graphical representation of the actuator disk properties listed in Table 11. ................... 19
Figure 16 Iterative convergence for , , , sinkage and pitch for case 1b-1. ....................... 20
Figure 17 Iterative convergence for , , , sinkage and pitch for case 1b-3. ....................... 21

Figure 18 Comparison of longitudinal forces. ................................................................. 22


Figure 19 Comparison of lateral forces......................................................................... 22
Figure 20 Comparison of roll moments......................................................................... 23
Figure 21 Comparison of yawing moments. ................................................................... 23

Figure 22 Visualisation of the water surface coloured with the elevation for case 1c-1 (drift angle 4
degrees)................................................................................................ 24
Figure 23 Visualisation of the water surface coloured with the elevation for case 1b-1 (straight ahead). 25

(drift angle 4 )..........................................................................................


Figure 24 Visualisation of the water surface coloured with the elevation for case 1b-3 with propulsion
26
Figure 25 Visualisation of the pressure coefficient on the submerged part of the ship and tank
walls for 1c-1. ......................................................................................... 27
Figure 26 Visualisation of the pressure coefficient on the submerged part of the ship and tank
walls for 1b-1. ......................................................................................... 28
Figure 27 Visualisation of the pressure coefficient on the submerged part of the ship and tank
walls for 1b-3. ......................................................................................... 29

Figure 28 Convergence of aft lateral force measurement for test T0Z02A03_C40000. ................. 32
Figure 29 Convergence of fore lateral force measurement for test T0Z02A03_C40000................. 32

Figure 30 View of the grid at the aft perpendicular (AP) of the ship. ....................................... 33
Figure 31 View of the grid at the mid section of the ship. .................................................... 34
Figure 32 View of the grid at the fore perpendicular (FP) of the ship. ...................................... 35
Figure 33 View of the grid near the bow in the symmetry plane of the ship. .............................. 36
Figure 34 Overview of the grid in the symmetry plane of the ship. ......................................... 36

Final version WL2015R00_066_3 v


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

List of Tables

Table 1 Overview of the EFD tests. ........................................................................... 3


Table 2 Averages, standard deviations and relative standard deviations for case 1b-3 rounded to
four significant digits. .................................................................................. 4
Table 3 Averages and standard deviations for case 1b-1 rounded to four significant digits. ........... 6
Table 4 Averages and standard deviations for case 1c-1 rounded to four significant digits. ........... 6

Table 5 Reference conditions .................................................................................. 10


Table 6 Settings for refinement surfaces and refinement boxes with absolute target cell sizes as
shown in Fig. 11. ....................................................................................... 12

/7. ................................................................................................ 14
Table 7 Number of cells for the coarse, medium and fine initial Cartesian mesh for the case based
on

/5. ................................................................................................ 14
Table 8 Number of cells for the coarse, medium and fine initial Cartesian mesh for the case based

/7 and
on

/5. .................................................................................................... 15
Table 9 Total cell count for the coarse, medium and fine meshes for case 1c-1 based on

Table 10 Thrust and torque values used as input for two of the CFD computations. ..................... 18
Table 11 Geometry properties of the actuator disk ........................................................... 18

Table 12 Overview of results .................................................................................... 30

Final version WL2015R00_066_3 vi


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

1 Introduction
Nautical research at Flanders Hydraulics Research (FHR) (Waterbouwkundig Laboratorium – Flanders Hy-
draulics Research n.d.) focuses on ships manoeuvring in shallow and confined water. For this research a
towing tank for manoeuvres in shallow water (cooperation FHR - Ghent University (Knowledge Centre Ma-
noeuvring in Shallow and Confined Water n.d.)) is available where experimental research is conducted. The
test results are used to generate mathematical manoeuvring models for fast-time and real-time simulations in
shallow and confined water. Amongst others, these mathematical models are used in the ship simulators of
FHR where maritime pilots of the ports of Antwerp, Zeebrugge and Ghent receive training. Traditionally, exper-
imental research has been the sole source of input for the mathematical manoeuvring models. In recent years,
Computational Fluid Dynamics (CFD) has seen significant improvements in the computational algorithms re-
lated to two-phase flows and in the mean time, the cost of the computational facilities to run the software in
an efficient manner has decreased significantly. For these reasons, FHR has started to use CFD as an alter-
native means to generate input for the mathematical manoeuvring models used in its real-time simulations. A
variety of shallow water test cases for the SIMMAN 2014 workshop were conducted at FHR (Delefortrie et al.,
2011) with models of the KCS and KVLCC2 and for some of these, CFD computations are performed for the
SIMMAN 2014 workshop. For the shallow water cases 1b-1, 1b-3 and 1c-1, computations are executed using
the CFD package FINE/Marine.
This report is based on the paper that was written for the workshop (Van Hoydonck and Eloot, 2014). The CFD
computations reported there were carried out blind, meaning that the experimental data could not be used to
compare results before the start of the workshop. The final workshop proceedings have not been published
so the decision was made to include a comparison with experimental values in this report.
This report starts with an analysis of the experimental data in Chapter 2. Afterwards, in Chapter 3 the setup
of the computations is discussed. This is followed in Chapter 4 by a grid convergence study for case 1c-1.
The resulting grid settings are used for cases 1b-1 and 1b-3 as discussed in Chapter 5. A comparison of
the numerical results with the experimental data is presented in Chapter 6 and the flow visualisations of the
solutions are shown in Chapter 7. Due to page limitations, these visualisations were not included in the paper
for the workshop. The report ends with conclusions.

Final version WL2015R00_066_3 1


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

2 Analysis of experimental data


Since the SIMMAN workshop proceedings with the EFD-CFD comparisons haven’t been released, a compar-
ison is added here. In this chapter, the EFD data is analysed.
Table 1 shows an overview of the experimental data (extracted from the DOC files) that is used as a reference.
For one of the computational cases, the towing tank experiment was repeated 10 times, which means that
apart from average values, standard deviations can be computed as well. The standard deviations for this
case will be used for the other two test cases as well. Inspection of Table 1 shows that the draft of the KVLCC2
in the experiments (0.2759 m) is different from the draft that was requested in the CFD instructions (0.2776 m).
As a result, the experiments were executed with a UKC value of 20.6% instead of 20.0% (that is, if the data
stored in the DOC files is correct). The draft of the KVLCC2 reported in Delefortrie et al. (2011) and Delefortrie
et al. (2014) differs from the values shown in the DOC files.
Averages for the captive stationary tests are computed from the data in the 30% and 95% time interval of the
test.
The following quantities are used in the comparison with CFD results and will be analysed here:
• Sinkage mid,
• Pitch,
• X-force,
• Y-force,
• X-moment,
• Z-moment,
• Propeller thrust,
• Propeller torque.
Of these, the X-moment is the only value that is directly recorded during the towing tank tests. The yawing
moment is computed from the lateral force measurements, the forces are summed from the fore and aft
measurements and the sinkage at the midship location is a linear interpolation using the fore and aft sinkage
measurements (after correction with the rail alignment data, see § 2.1). The pitch angle is also computed from
the fore and aft sinkages.

2.1 Rail alignment corrections

The file with rail alignment correction data for CFD tests 1b-1 and 1b-3 is T0Z02A03_IJKR00.IJK. For test
1c-1, the file with rail alignment correction data is T0Z03A03_IJKR00.IJK. The data recorded in these files
is shown Fig. 1. The overall shape of the rail alignment corrections are similar for both IJK files, however,
T0Z03A03_IJKR00.IJK contains suspicious oscillations for the first 25 meter and the last 10 meter of the
measurement. It is very likely that these oscillations will be visible in a plot of the time series of the midship
sinkage of T0Z03A03_C47000. The original IJKR-test on which the data in T0Z03A03_IJKR00.IJK is based
also contains these oscillations.

Final version WL2015R00_066_3 2


fhr.cls
2015/08/18 – version 0.14
fhr.cls
Final version

2015/08/18 – version 0.14


Table 1 – Overview of the EFD tests.

CFD case doc file date waterdepth draft ukc heading drift lat. position req. prop. rate Fr(v) Fr(d) velocity sampling rate
/

1c-1 T0Z03A03_C47000.DOC 06/12/2010 0.3328 0.2759 20.6 4.0 4.0 -0.0001 0.0 0.064 0.23 0.416 40.0

1b-1 T0Z02A03_C40000.DOC 12/11/2010 0.3328 0.2759 20.6 0.0 0.0 - 400.104 0.064 0.23 0.416 40.0
SIMMAN 2014: Shallow-water CFD Computations

1b-3 T0Z02A03_C47000.DOC 10/11/2010 0.3328 0.2759 20.6 4.0 4.0 -0.0001 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47001.DOC 11/11/2010 0.3328 0.2759 20.6 4.0 4.0 0.0 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47002.DOC 11/11/2010 0.3328 0.2759 20.6 4.0 4.0 0.0 400.104 0.064 0.23 0.416 40.0

WL2015R00_066_3
1b-3 T0Z02A03_C47003.DOC 12/11/2010 0.3328 0.2759 20.6 4.0 4.0 -0.0001 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47004.DOC 12/11/2010 0.3328 0.2759 20.6 4.0 4.0 0.0 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47005.DOC 12/11/2010 0.3328 0.2759 20.6 4.0 4.0 0.0 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47006.DOC 13/11/2010 0.3328 0.2759 20.6 4.0 4.0 -0.0001 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47007.DOC 14/11/2010 0.3328 0.2759 20.6 4.0 4.0 0.0 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47008.DOC 14/11/2010 0.3328 0.2759 20.6 4.0 4.0 0.0 400.104 0.064 0.23 0.416 40.0
1b-3 T0Z02A03_C47009.DOC 14/11/2010 0.3328 0.2759 20.6 4.0 4.0 0.0 400.104 0.064 0.23 0.416 40.0

3
SIMMAN 2014: Shallow-water CFD Computations

Figure 1 – Fore and aft rail alignment corrections for T0Z02A03 and T0Z03A03 tests. T0Z03A03 shows suspicious
oscillations during the first 30 meter and the last 10 meter of the test.

2.2 Averages and standard deviations for case 1b-3

Between 10 and 14 November 2010, ten tests were executed for the condition of case 1b-3 (see Table 1). By
computing averages for the aforementioned quantities for each of these tests, the spread in the values can be
computed. This way, an uncertainty band can be determined for the reference values.
In Fig. 2, the thurst and torque values for the 10 tests are shown. The average value and its standard deviation
are shown in red. Figure 3 shows the values of as a function of thurst for the same 10 test cases. The
averages, standard deviations and relative standard deviations are shown in Table 2. Since these values
were computed using a set of Python scripts that utilize the original DOC files as input, it is possible that slight
differences are present when the values are compared to average values computed using the derived KRT
files using ZeeMan.

Table 2 – Averages, standard deviations and relative standard deviations for case 1b-3 rounded to four significant digits.

Quantity unit average std. rel. std. (%)

N -1.400 0.07797 5.569


N 13.98 0.08694 0.6218
Nm -0.6224 0.02452 3.939
Nm 14.92 0.05939 0.3980
Sinkage mm 3.899 0.01905 0.4885
Pitch deg -0.04324 0.0003857 0.8920
N 2.616 0.1719 6.572
Nmm 47.61 2.253 4.731

Final version WL2015R00_066_3 4


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

Figure 2 – Thrust versus torque for test cases T0Z02A03_C4700x. The numbers next to the black points denote the last
number of the test identification (see Table 1).

Figure 3 – Thrust versus X-force for test cases T0Z02A03_C4700x. The numbers next to the black points denote the last
number of the test identification (see Table 1).

2.3 Averages and standard deviations for case 1b-1

For case 1b-1, only one test was executed in the towing tank at FHR (T0Z02A03_C40000.DOC). Hence,
the reference values are computed from this one test. Standard deviations are estimated using the relative
standard deviations computed for case 1b-3. The resulting are presented in Table 3.

Final version WL2015R00_066_3 5


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

Table 3 – Averages and standard deviations for case 1b-1 rounded to four significant digits.

Quantity unit average std.

N -0.6588 0.03669
N 1.127 0.007006
Nm 0.01025 0.0004035
Nm 0.9256 0.003684
Sinkage mm 3.45 0.01685
Pitch deg -0.03877 0.0003458
N 3.198 0.2101
Nmm 54.86 2.596

2.4 Averages and standard deviations for case 1c-1

Similar to case 1b-1, only one test was executed in the towing tank at FHR for case 1c-1 (T0Z03A03_C47000.DOC).
Hence, the reference values are computed from this one test. Standard deviations are estimated using the
relative standard deviations computed for case 1b-3. The resulting values are presented in Table 4. This case
was executed using a bare hull, hence thrust and torque values are not included.

Table 4 – Averages and standard deviations for case 1c-1 rounded to four significant digits.

Quantity unit average std.

N -3.693 0.2056
N 12.55 0.07802
Nm -0.4456 0.01755
Nm 16.33 0.06497
Sinkage mm 3.654 0.01785
Pitch deg -0.04983 0.0004445

2.5 Time history analysis

In this section, some time history plots of the experimental quantities of interest will be shown1 . For each
quantity shown, the normalized cumulative moving average (nCMA) and the normalized cumulative standard
deviation (nCSTD) are computed for the part of the test that is used to compute average values. The derivatives
of these two quantities can be used to judge the convergence of the time history of the quantity: if the derivative
of nCMA decreases as time progresses, the cumulative moving average converges towards a fixed value. The
same holds true for the standard deviation: if its derivative decreases as time progresses, the magnitude of
the fluctuations converges as well. In addition, the frequency content of the data is shown using an FFT plot.
This data can be used to find trends in the data: if the lowest frequencies have the highest value, the quantity
is not stationary in time but contains a trend.
It should be noted that while processing the DOC files, it was noticed that the measured values for the fore
sinkage measurement remain exactly the same for longer time periods (in some cases 10 seconds or more).
This is only observed for the fore sinkage, not the aft sinkage. In a single measurement, it can occur multiple
times, and the measured value is exactly the same. It appears as if the measurement device is stuck during

imately 10 seconds and starts at = 112 . There is another one that starts at = 173 , right before the start of
these time intervals. The top graph in Fig. 4 shows multiple intervals like these, the longest one takes approx-

1
Eight plots (one for each quantity) are created for each of the data sets, which amounts to 96 plots. At two plots per page, this would
add 48 pages to the report. The plots are available upon request.

Final version WL2015R00_066_3 6


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

the deceleration phase of the test. After correcting the measured sinkages with the rail alignment corrections,
these intervals with constant measured values disappear.

Figure 4 – Measured sinkages for test T0Z03A03_C47000 showing multiple time intervals with constant measured value.

Figure 5 shows the convergence characteristics of the total measured X-force (the sum of the forward and aft
measurements) for test T0Z02A03_C40000. During the last twenty seconds of the measurement interval, the
nCMA value remains within 2% of the final value. The fluctuations around the average remain within 2% of
the final value for the last 40 seconds. The dominant frequency for this time signal is 2.4 Hz.
For the same test, Fig. 6 shows the convergence of the propeller torque and in Fig. 7 the convergence analysis
of total lateral force is shown. According to the DOC file for this case (see Table 1), the requested propeller
rate was about 400 rpm or 6.66 Hz. This value is the dominant frequency in the lower plot of Fig. 6. The
two-per-rev frequency (at 13.3 Hz) is also clearly visible. The three-per-rev frequency is just visible at 20 Hz,
the Nyquist frequency of the recording (the sampling rate for the experiments equals 40 Hz). Convergence
for the torque is very good, both the average value and the fluctuations.
The CMA convergence of the lateral force (Fig. 7) is not as good as the convergence for the longitudinal
force. The time history plot shows signficant fluctuations in the amplitude, which is reflected in the plot of the
normalised cumulative standard deviation. Also, the average value converges only very slowly towards its
final value, with deviations up to 20% from the final value during the last minute of the measurement interval.
The frequency plot shows that the dominant frequency is located at 19.6 Hz (this may not be clear from the
figure itself but a close inspection of the interactive figure shows that the peak at 19.6 Hz is indeed the highest
one). A broad peak centered at f = 3.2 Hz is only slightly lower in magnitude. The signals with the lowest
frequencies also have a significant power. Inspection of the time series of the fore and aft components from
which the lateral force is computed, shows that the average of the fore measurement is close to the average
of the total lateral force. The average of the aft measurement is close to zero. The plots for the fore and aft
measurements are displayed in Appendix A.
The sinkage at the midship location is computed from the fore and aft sinkage measurements and shown in
Fig. 8. Low-frequency oscillations dominate the time signal. Convergence towards the final mean value is

ment interval. Although the convergence of the nCMA appear erratic, the derivatives are smaller than 10 4 for
reasonable: the nCMA of the sinkage remains within 2% of the final value for the last minute of the measure-

the last 110 seconds of the measurement interval, which is significantly better than the convergence of e.g.
the X-force (Fig. 5).

Final version WL2015R00_066_3 7


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

Figure 5 – Convergence of total X-force for test T0Z02A03_C40000.

Figure 6 – Convergence of propeller torque for test T0Z02A03_C40000.

Final version WL2015R00_066_3 8


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

Figure 7 – Convergence of total Y-force for test T0Z02A03_C40000.

Figure 8 – Convergence of midship sinkage for test T0Z02A03_C40000.

Final version WL2015R00_066_3 9


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

3 Computational Setup

3.1 Test conditions

This section contains the relevant data related to the CFD computations documented in this report that was
provided by the organizing committee of SIMMAN 2014. Cases 1b-1 and 1b-3 are characterized by:
• captive simulation in shallow still water condition;
• Pitch and heave free; otherwise restrained;
• Speed ratio / 0 = 1.0 with an approach speed 0 =7 at full scale;
• model length = 4.2667 ;
• water depth: / = 1.2;
• width of basin: 7.0 m;
• Tests performed at model self-propulsion point;
• Appendages: Propeller and rudder;
• Available EFD data:
– Static tests – Forces/Moments including rudder forces;
– Dynamic tests – Time histories of Forces/Moments including rudder forces;
For case 1c-1, the characteristics are:
• Bare hull captive simulation in shallow still water condition;
• Pitch and heave free; otherwise restrained;
• Speed ratio / 0 = 1.0 with an approach speed 0 =7 ;
• model length = 4.2667 ;
• water depth: / = 1.2;
• width of basin: 7.0 m;
• Appendages: none
• Available EFD data:
– Static tests – Forces/Moments;
A summary of the test conditions is listed in Table 5.

Table 5 – Reference conditions

Test
No - - m/s rps deg

2 106
2 106
1b-1 0.063 0.416 6.67 0 0

2 106
1b-3 0.063 0.416 6.67 4.0 0.070
1c-1 0.063 0.416 - 4.0 0.070

Final version WL2015R00_066_3 10


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

3.2 KVLCC2 CAD Model

A cleaned version of the IGES file that can be found on the SIMMAN website is available at FHR. It can be
found in the /cfd/cfd/cad_models/ships/T0Z/parasolid/ folder on the cluster. The cleaned Parasolid
file (VLCCmodelscale_cf.x_t) is used by a set of domain creation scripts that automate the majority of tasks
needed for creating a setup of a CFD computation using FINE/Marine. The CAD model of the hull is shown in
Fig. 9. For the appended cases, a simplified rudder model is attached to the hull. For the bare hull case, the
small part of the rudder attached to the hull is included as well.

Figure 9 – Bottom, front, back and isometric view of the Parasolid CAD model of the KVLCC2 as used for the CFD
computations.

3.3 Domain Size and Ship Position

The three cases follow the same general setup. The computational domain is as wide as the towing tank
where the experiments were executed (7 m). It extends 1.5 ahead of the ship bow and 2.5 aft of the

and conditions n.d.), which means that the water depth for a depth to draft ratio / = 1.2 is 0.3328 m.
ship stern (see Fig. 10). The ship draft in the experiments is 0.2773 m (MOERI KVLCC2 Tanker – Geometry

Above the water level, the domain extends 0.5 upward. In absolute dimensions, the length, width and
height of the domain box are 21.96 m, 7 m, and 2.46615 m, respectively. To prevent the cells below the hull
from compressing too much in the final position, the ship is meshed three millimeters below the hydrostatic
equilibrium.

/2 on
In longitudinal direction, the aft perpendicular is located at the origin of the global axis system (as shown in
Fig. 10). For cases 1c-1 and 1b-3, the ship is rotated around its midship location, which is located at
the x-axis of the global axis system.
The initial location of the centre of gravity in the global axis system is 2.2825, 0, 0.3298, which is 3 mm below
the water level. Using domhydro, a mass of 740.7429 kg was computed based on the given draft.

Final version WL2015R00_066_3 11


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

21.96 m

7m 2.5 Lpp y 1.5 Lpp


x

2.46615 m
z

Figure 10 – Top view of domain box with dimensions and global axis system orientation.

3.4 Topology and General Refinement Settings

The hull of the KVLCC2 is simplified to the point where it consists of the deck surface, transom surface, two
hull halves and the propeller shaft surface. For the appended cases (1b-1 and 1b-3), the rudder has a blunt
trailing edge that is approximately 0.5 mm wide.
The global refinement diffusion factor is increased to five (from the default of two) to improve the prediction of
drag.
All rudder surfaces have a refinement of 8, as does the propeller shaft surface. The transom and hull halves
get six refinements and the deck four. A refinement surface that covers the complete domain is used to capture
the complete water surface (Fig. 11a); a smaller surface located close to the ship increases the resolution of
the water surface near the ship (Fig. 11b). The relevant refinement settings for these surfaces are displayed
in Table 6. For the cases with propulsion, a refinement surface is added at the propeller location to increase
the cell count in this region.
Two refinement boxes are used to control the cell sizes of the water volume. The first one starts at the domain
inlet and ends halfway between the ship transom and the domain outlet (Fig. 11c). The last one encloses the
rest of the water volume (Fig. 11d). Both of these boxes are volumic.
Three non-volumic refinement boxes are used to refine cells close to the ship hull. One of them covers the
last quarter of the hull, the second one covers the front eighth part of the hull. The last one covers the lower
part of the hull and the tank bottom, and is as wide as the hull. Seven refinement levels are used for these
boxes, one more than the refinement level for the two hull halves.
By default, the target cell sizes as listed in Table 6 are used for the water surface refinement planes and the
water volume refinement boxes with absolute target cell sizes.

Table 6 – Settings for refinement surfaces and refinement boxes with absolute target cell sizes as shown in Fig. 11.

refinement surface/box

/8 /8 /500
/16 /16 /1000
large refinement surface (Fig. 11a)

/12.5 /50 /100


small refinement surface (Fig. 11b)

/25 /100 /200


front refinement box (Fig. 11c)
aft refinement box (Fig. 11d) 3 3 3

The final cell size for refinement surfaces depends on the initial cells size, which is used as a parameter in the
grid independence study as explained in the next section.

Final version WL2015R00_066_3 12


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) Large water surface refinement plane. (b) Small water surface refinement plane.

(c) Front water volume refinement box. (d) Aft water volume refinement box.

Figure 11 – Overview of refinement boxes and refinement surfaces with absolute target cell sizes.

(and the rudder for 1b-1 and 1b-3), the target + value is smaller than 1, i.e., no wall functions are employed.
High quality viscous layers are inserted on the hull and tank bottom to account for viscous effects. On the hull

For the tank bottom, wall functions are employed to model the velocity profile in the viscous layer.

3.5 Solution Method and Flow Settings

The density and dynamic viscosity of water (999.147 / 3 and 0.00114696 ) are based on standard
sweet water (Anon., 2011) with a temperature of 14.7 degrees Celsius.
In all cases, the ship starts from rest, and is accelerated to the final velocity using a velocity profile based on
a cosine function. At the start of a simulation, the ship degrees of freedom (sinkage and trim) are released as
well. The consequence of meshing the ship below the hydrostatic position at rest is that the initial motion of
the ship will be upward, stretching the mesh below the ship.

velocity is 0.416 m/s. For the computations reported here, the Reynolds number is about 1.5 106 , which
The reference length is 4.2667 m, the length between perpendiculars of the ship model, and the reference

corresponds to fully turbulent flow. Results are computed by solving the incompressible, two-phase Reynolds-
Averaged Navier-Stokes (RANS) equations, which means that the complete spectrum of turbulence (from the
largest scales down to the Kolmogorov scales) is modelled. The interface between the air and water phase
is tracked with a Volume-of-Fluid (VOF) method. The turbulence closure model used is the Explicit Algebraic
Stress Model (EASM) (Anon., 2014). The boundary condition applied on the tank lateral sides and the ship
deck is a slip condition, while on all other ship surfaces, a no-slip condition is applied. For the upper surface
of the domain, the pressure is prescribed (updated hydrostatic pressure). The boundary condition for the inlet
and outlet surfaces is zero far field velocity. Turbulence initialization levels are left at their default values.

Final version WL2015R00_066_3 13


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

4 Grid convergence study for case 1c-1

4.1 Base meshes

By default, HEXPRESS uses cell sizes that are approximately cubic with a total that is as close to 1000 as pos-
sible. Here, the initial cell size is coupled to the size of the ship in terms of its length between perpendiculars,
.

/7 as possible1 . In addition, a base mesh is created with cubic cells


The domain box is divided along the three axes such that the resulting cells are approximately cubic and have

/5. Using these two base meshes, six derived meshes are created
linear dimensions that are as close to
that have linear dimensions close to
that are used for the grid convergence study.

4.2 Generation of derived meshes

For the first case (based on /7), this gives 36, 11 and 4 cells in the x, y and z-direction, respectively, for a
total of 1584 cells. After an initial mesh is generated, the complete project is duplicated and all absolute cell
size settings are reduced by a factor 1.4 to get a finer mesh. For the third mesh the initial cell size settings are

/7 and in Table 8 for


increased by a factor 1.4 from their initial values to get a coarse mesh. The number of cells for the coarse,

/5. By accident, the number cells for the fine mesh based on /5 is the same as the
medium and fine initial Cartesian meshes are listed in Table 7 for the case based on

/7, in addition, both coarse meshes have the same number of cells.
the case based on
medium mesh for

Table 7 – Number of cells for the coarse, medium and fine initial Cartesian mesh for the case based on /7.

x y z total

coarse 18 6 2 216
medium 36 11 4 1584
fine 50 15 6 4500

Table 8 – Number of cells for the coarse, medium and fine initial Cartesian mesh for the case based on /5.

x y z total

coarse 18 6 2 216
medium 26 8 3 624
fine 36 11 4 1584

The target + values for the viscous layers on the ship and tank bottom were changed in a manner similar to
the changes in absolute target cell sizes.
For all cases, the grid near the stern in the lateral symmetry plane of the ship is shown in Fig. 12. Other views
are shown in Appendix B.

1
With this setting, six levels of refinement give target cell sizes slightly smaller than 0.01 m. Ten levels of refinement give target cell
sizes smaller than 0.0006 m

Final version WL2015R00_066_3 14


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

Table 9 – Total cell count for the coarse, medium and fine meshes for case 1c-1 based on /7 and /5.

/7 /5

coarse 2 673 963 2 673 963


medium 6 567 806 4 509 538
fine 11 221 780 7 667 172

(a) /5, coarse (b) /7, coarse

(c) /5, medium (d) /7, medium

(e) /5, fine (f) /7, fine

Figure 12 – View of the grid near the stern in the symmetry plane of the ship for the six different cases.

4.3 Results

= 0.05 ) is performed where the trim and sinkage of the ship are the only degrees of freedom. In
For each of the six cases as discussed in the previous section, a steady simulation (10000 iterations, time
step
Fig. 13, convergence of , and versus time are shown, as are values averaged over the last 500
iterations. Close inspection of the convergence histories (Figs. 13a, 13c and 13e) shows that for the coarse
meshes, initially, results are converging, but approximately halfway, the trends become unstable. This may
indicate that the mesh is too coarse for the problem at hand. Differences between results on the medium and
fine meshes are smaller than differences between results on the coarse and medium meshes, which indicates
that as the cell sizes are reduced, the solution converges to a single value.

Final version WL2015R00_066_3 15


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) vs time. (b) vs # cells.

(c) vs time. (d) vs # cells.

(e) vs time. (f) vs # cells.

Figure 13 – Iterative convergence and final values for , and for 1c-1.

No uncertainty analysis has been performed as iterative convergence is not satisfactory for results computed
with the coarse meshes. Similar conclusions can be drawn after inspection of Fig. 14.
For all quantities, the differences between the medium and fine meshes are sufficiently small to assume that

/7 will be used.
computations based on the medium grid give adequate results. For the rest of the computations, the settings
based on

Final version WL2015R00_066_3 16


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) Sinkage vs time. (b) Sinkage vs # cells.

(c) Pitch vs time. (d) Pitch vs # cells.

(e) vs time. (f) vs # cells.

Figure 14 – Iterative convergence and final values for sinkage, pitch and for case 1c-1.

Final version WL2015R00_066_3 17


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

5 Cases 1b-1 and 1b-3

5.1 Setup

Based on the grid study for case 1c-1, it was decided to generate meshes for cases 1b-1 and 1b-3 starting
from the settings as used for the medium mesh of 1c-1. Some modifications are necessary, as cases 1b-1
and 1b-3 include propulsion. A refinement surface is added at the location of the propeller to increase the cell
count in this area. The selected target cells size for this refinement surface is 0.005 m in the three dimensions.
Furthermore, the target cell sizes for the refinement surface shown in Fig. 11b were reduced to 0.1, 0.1 and
0.003 m. The consequence of these changes is that the cell count for 1b-1 and 1b-3 increases to 15.4 and
16.4 million cells, respectively.
The thrust and torque values are required as input for propulsion modeling using an actuator disk. The values
measured during the experiments are used for this (Delefortrie et al., 2014), they are shown in Table 10.

Table 10 – Thrust and torque values used as input for two of the CFD computations.

case thrust, torque,

1b-1 3.198 54.858


1b-3 2.616 47.61

Past experience with the use of FINE/Marine has shown that in order to run a simulation with an actuator disk,
first a simulation should be run without propulsion (but with free trim and sinkage) so that the initial transients
subside. In a second computation that starts from the end result of the first computation, the actuator disk
model is activated.
The geometric properties of the actuator disk (inner and outer radius, thickness and origin) are listed in Table 11.
For case 1b-3, the location is modified to account for the drift angle, in a similar way that the ship orientation
itself is modified.

Table 11 – Geometry properties of the actuator disk

property value unit

inner radius 0.0107 m


outer radius 0.75 m
thickness 0.02 m
center x-coordinate 0.0869 m
center y-coordinate 0.0 m
center z-coordinate 0.1298 m

5.2 Results

Iterative convergence plots are shown in Fig. 16 for 1b-1 and in Fig. 17 for case 1b-3. In all these figures,
the first 500 seconds show results without propulsion, and the last 500 seconds show results with propulsion
activated. In the cases with propulsion, as shown in Figs. 16a and 17a is the sum of the pressure and
viscous force acting on the ship in the longitudinal direction, so this does not include the contribution due
to the propeller. The resultant force (that would be measured in the experimental results), can be found by
adding the thrust (shown in Table 10) to . In both cases, iterative convergence for all quantities is sufficient
to consider the results converged. It should be noted that despite the large number of cells for these cases,

Final version WL2015R00_066_3 18


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

Figure 15 – Graphical representation of the actuator disk properties listed in Table 11.

the use of an actuator disk to model propulsion may have a negative impact on the accuracy of the results
when compared to cases without propulsion for a similar number of cells. It is the simplest (and least accurate
method) to take propulsion into account in CFD computations.

Final version WL2015R00_066_3 19


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

0 1.4
1.2
1 1.0
0.8
2 0.6

Fy, N
Fx, N

3 0.4
0.2
4 0.0
0.2
50 200 400 600 800 1000 0.40 200 400 600 800 1000
Time, s Time, s
(a) vs time. (b) vs time.

1.5 0.0040
0.0035
1.0 0.0030
Sinkage, mm

0.0025
Mz, N

0.5 0.0020
0.0015
0.0 0.0010
0.0005
0.50 200 400 600 800 1000 0.00000 200 400 600 800 1000
Time, s Time, s
(c) vs time. (d) Sinkage vs time.

0.04
0.02
0.00
Pitch, deg

0.02
0.04
0.06
0.08
0.100 200 400 600 800 1000
Time, s
(e) Pitch vs time.

Figure 16 – Iterative convergence for , , , sinkage and pitch for case 1b-1.

Final version WL2015R00_066_3 20


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

0 2
0
1
2
2 4

Fy, N
Fx, N

6
3 8
10
4
12
50 100 200 300 400 500 600 140 100 200 300 400 500 600
Time, s Time, s
(a) vs time. (b) vs time.

16 0.0045
14 0.0040
12 0.0035
10 0.0030
Sinkage, mm

8 0.0025
Mz, N

6 0.0020
4 0.0015
2 0.0010
0 0.0005
20 100 200 300 400 500 600 0.00000 100 200 300 400 500 600
Time, s Time, s
(c) vs time. (d) Sinkage vs time.

0.04
0.02
0.00
0.02
Pitch, deg

0.04
0.06
0.08
0.10
0.120 100 200 300 400 500 600
Time, s
(e) Pitch vs time.

Figure 17 – Iterative convergence for , , , sinkage and pitch for case 1b-3.

Final version WL2015R00_066_3 21


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

6 Comparison of CFD and EFD results


Results of CFD computations discussed in Chapters 4 and 5 are compared with the experimental values as
presented in Chapter 2. For each quantity that is compared, first absolute values are shown and then relative
errors. These relative errors are computed from the absolute values as follows

(% ) = (1)

where is the CFD result and is the experimental reference value.


For the absolute plots, the standard deviations are shown using vertical bars. For the longitudinal force com-
ponent, the resultant X-force on the hull (sum of pressure and viscous forces) is compared with the resultant
force minus the measured thrust (see Fig. 18). The standard deviations shown in Fig. 18a consist of the stan-
dard deviations of the resultant longitudinal force for case 1c-1 (see Table 4), and for cases 1b-1 and 1b-3, the
standard deviations shown are the sum of the standard deviations for the longitudinal force and the standard
deviations of the measured thrust (see Tables 2 and 3).

(a) Absolute values. (b) Relative errors.

Figure 18 – Comparison of longitudinal forces.

In all three cases, the CFD predictions of the longitudinal force on the hull are very close to the experimental
values. For two cases (1c-1 and 1b-3), the error is smaller than or equal to the standard deviation associated
with the measurement. For case 1b-1, the error is about 0.5 N larger than the standard deviation of the
experiment. Figure 18b shows that on a relative scale, the errors vary between -6% and +7%, which is very
good. The best value is achieved for case 1b-3, with an error smaller than 4%.
The lateral force comparison is presented in Fig. 19. The magnitude of the standard deviations of the exper-
imental values are so small that the error bars are not discernible. The result for case 1b-1 is noteworthy as
the magnitude is almost right, but the CFD prediction has the wrong sign. As a result, the relative error for
this case is close to 200%. The ship sails straight ahead in this case (no drift angle), so the asymmetry in
the lateral force can only be caused by the interaction between the propeller and the rudder. The actuator
disk model in FINE/Marine uses both a thrust force, and a torque as input. It is possible that this model is
too simplistic for the case at hand and one should use more advanced methods, such as a potential panel
method where the actual geometry of the propeller is modelled. For cases 1c-1 and 1b-3, the lateral forces
are predicted with errors smaller than -2% and -4%, respectively.

(a) Absolute values. (b) Relative errors.

Figure 19 – Comparison of lateral forces.

Final version WL2015R00_066_3 22


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

The small reference values for the roll moments mean that the errors for these cases are larger (Fig. 20). The
best result is achieved for case 1b-3 with an error of 9%. The relative errors shown in Fig. 20b are misleading:
on an absolute scale, the prediction for case 1b-1 is the best, with a difference of approximately 0.02 Nm; the
relative error is more than 200% for this case.
The comparison of the yawing moment is shown in Fig. 21. On an absolute scale, the results are again very
good. The yawing moment for 1c-1 is predicted with an error smaller than 2%. For case 1b-3, the error is 6%
and for 1b-1, it is approximately 20%. On an absolute scale, the error for the latter case is 0.2 Nm, while for
1c-1, it is 0.26 Nm and for 1b-3 it is slighly less than 1 Nm.

(a) Absolute values. (b) Relative errors.

Figure 20 – Comparison of roll moments.

(a) Absolute values. (b) Relative errors.

Figure 21 – Comparison of yawing moments.

Final version WL2015R00_066_3 23


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

7 Flow visualisations

7.1 Water surface

The water surface elevation around the KVLCC2 hull for the three cases is shown in Figs. 22, 23 and 24. For
the latter two, the water surface elevation is shown for both the computation without propulsion and for the
restart computation with propulsion.
The mesh for case 1c-1 uses a refinement surface for the water surface with coarser cells. Hence, less detail
is visible in the wave pattern produced by the vessel (Fig. 22). By using automatic grid refinement (AGR), it
is possible to improve the water surface without adding cells in places where they are not needed. The AGR
functionality in the ISIS CFD solver has not been used up to this point.

Figure 22 – Visualisation of the water surface coloured with the elevation for case 1c-1 (drift angle 4 degrees).

7.2 Hydrodynamic pressure on the hull and tank

For the three cases, the pressure coefficient is shown where the mass fraction is larger than or equal to
0.5 (water). The pressure coefficient is defined as

= 2
1
(2)
2

where is the hydrodynamic pressure, is the water density and is the reference velocity. In stagnation
points, the value of equals one.
There are significant differences between the pressure distributions for the two cases with non-zero drift angle
(Fig. 25 and 27). For the latter case, signficant fluctuations can be observed on the leeward side of the hull,
aft of the midship location near the water surface. Due to the differences in grid resolution between 1c-1
and 1b-3 it is unclear what the cause is of these differences (either the grid resolution or the influence of the
propeller).

Final version WL2015R00_066_3 24


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) Without propulsion.

(b) With propulsion.

Figure 23 – Visualisation of the water surface coloured with the elevation for case 1b-1 (straight ahead).

Final version WL2015R00_066_3 25


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) Without propulsion.

(b) With propulsion.

Figure 24 – Visualisation of the water surface coloured with the elevation for case 1b-3 with propulsion (drift angle 4 ).

Final version WL2015R00_066_3 26


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) Overview.

(b) Close-up.

Figure 25 – Visualisation of the pressure coefficient on the submerged part of the ship and tank walls for 1c-1.

Final version WL2015R00_066_3 27


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) Overview.

(b) Close-up.

Figure 26 – Visualisation of the pressure coefficient on the submerged part of the ship and tank walls for 1b-1.

Final version WL2015R00_066_3 28


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) Overview.

(b) Close-up.

Figure 27 – Visualisation of the pressure coefficient on the submerged part of the ship and tank walls for 1b-3.

Final version WL2015R00_066_3 29


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

8 Conclusions
This report documents the CFD computations that were executed for the SIMMAN 2014 workshop in Denmark.
FHR executed tests with the KVLCC2 and KCS ship models in the towing tank for shallow and confined water
for this workshop. A subset of these tests were used as reference for blind CFD computations. For three of
these tests (1c-1, 1b-1 and 1b-3) CFD computations were carried out using the FINE/Marine CFD software.
This report is based on a paper that will be included in the workshop proceedings and a poster that was
presented at the poster session of the workshop. The final workshop proceedings haven’t been released,
which means that it is unknown how close the CFD computations are to the EFD data. For this reason, the
current report contains an analysis of the experimental data, a comparison of the CFD results with the EFD
results and some flow visualisations.
The EFD analysis showed that there may be issues with the rail alignment data for case 1c-1. Compared to
the rail alignment data for the other two cases, the rail alignment data for 1c-1 contains suspicious oscillations
during the first 25 meter of the tests and the last 10 meter of the test. It was also noticed that the raw data of
the front sinkage shows suspiciously constant values for longer time periods, something that is not observed
for the aft sinkage measurements. For one case (1b-3), the experiment was repeated 10 times. Apart from
average values, standard deviations were computed for this case as well. It was found that relative standard
deviations are largest for the longitudinal force, the thrust, torque and roll moment, at 5.6%, 6.5%, 4.7% and
4.0%, respectively. For the other quantities of interest, the relative standard deviations are all smaller than
1%. These relative standard deviations were used to estimate standard deviations for the other two test
cases.

/7. For
For case 1c-1, a grid convergence study was performed to investigate the influence of the initial cell size on
the final result. It was decided to use an initial grid with cells that have a linear dimension close to
case 1c-1, the grid contains 11.2 million cells. The grids of 1b-1 and 1b-3 contain 15.4 and 16.4 million cells,
due to using smaller target cells sizes for the water surface refinement. The smaller cell size near the water
surface results in more detailed wave patterns.
The comparison of the EFD and CFD results shows that (apart from the lateral force for case 1b-1) the results
are in very good agreement with the experimental values (these are shown again in Table 12). The errors for
are smaller than 8%. The best result is achieved for 1b-3, with an error less than 4% for . Case 1b-1 is
the only one without drift. For this case, the reference values of , and are rather smaller than for the
other two cases which results in large relative errors (up to 200%). Although both thrust and torque are used
as input for the actuator disk model, it is possible that this model is too simple for this specific case.

Table 12 – Overview of results

case
EFD, N CFD, N E(%D) EFD, N CFD, N E(%D)

1c-1 -3.693 -3.481 -5.731 12.55 12.35 -1.572


1b-1 -0.6588 -4.152 7.66 1.127 -1.245 -210.5
1b-3 -1.4 -4.174 3.918 13.98 13.4 -4.197

case
EFD, Nm CFD, Nm E(%D) EFD, Nm CFD, Nm E(%D)

1c-1 -0.4456 -0.6508 46.06 16.33 16.59 1.595


1b-1 0.01025 0.03259 218.1 0.9256 1.117 20.62
1b-3 -0.6224 -0.6773 8.822 14.92 15.85 6.196

Final version WL2015R00_066_3 30


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

References
Anon. (2011). Fresh Water and Seawater Properties. http://ittc.sname.org/CD2011/pdfProcedures2011/
7.5-02-01-03.pdf

— (2014). Theoretical Manual ISIS-CFD v3.1

Delefortrie, G.; Eloot, K.; Mostaert, F. (2011). SIMMAN 2012: Execution of model tests with KCS and
KVLCC2. WL Rapporten, 846_01. Version 2.0. Antwerp, Belgium

Delefortrie, G.; Eloot, K.; Peeters, P.; Mostaert, F. (2014). SIMMAN 2014: Revised Shallow Water Model
Tests with KCS and KVLCC2. WL Rapporten, 00_066. Version 3.0. Antwerp, Belgium

Knowledge Centre Manoeuvring in Shallow and Confined Water. http://www.shallowwater.be/

MOERI KVLCC2 Tanker – Geometry and conditions. http://www.simman2014.dk/cms/site.aspx?p=


13327

Van Hoydonck, W.; Eloot, K. (2014). Shallow Water CFD Computations for SIMMAN 2014. In: SIMMAN
2014 Workshop on Verification and Validation of Ship Manoeuvring Simulation Models. Preprints of Workshop
Proceedings. Ed. by J. F. Otzen; C. D. Simonsen

Waterbouwkundig Laboratorium – Flanders Hydraulics Research. http://www.watlab.be

Final version WL2015R00_066_3 31


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

A Additional time analysis plots of the EFD data


Figures 28 and 29 show the aft and fore lateral force measurements that are used to compute the lateral force
(Fig. 7) for test case T0Z02A03_C40000. The mean of the aft measurement equals 0.09722 N, the mean of the
fore measurement is 1.030 N. The standard deviations are 0.6152 and 1.005, respectively: the measurement
with the smallest mean value has the largest standard deviation, and vice versa. From the time histories of
the normalized CMA, it is apparent that during the whole test, the CMA oscillates around its final value for the
fore measurement, whereas for the aft measurement, the CMA value oscillates around 2, and only starts to
decrease towards 1 around t = 155 s. The FFT plots of both measurements are very similar.

Figure 28 – Convergence of aft lateral force measurement for test T0Z02A03_C40000.

Figure 29 – Convergence of fore lateral force measurement for test T0Z02A03_C40000.

Final version WL2015R00_066_3 32


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

B Additional views of the grid using cutplanes

(a) /5, coarse (b) /7, coarse

(c) /5, medium (d) /7, medium

(e) /5, fine (f) /7, fine

Figure 30 – View of the grid at the aft perpendicular (AP) of the ship.

Final version WL2015R00_066_3 33


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) /5, coarse (b) /7, coarse

(c) /5, medium (d) /7, medium

(e) /5, fine (f) /7, fine

Figure 31 – View of the grid at the mid section of the ship.

Final version WL2015R00_066_3 34


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) /5, coarse (b) /7, coarse

(c) /5, medium (d) /7, medium

(e) /5, fine (f) /7, fine

Figure 32 – View of the grid at the fore perpendicular (FP) of the ship.

Final version WL2015R00_066_3 35


fhr.cls
2015/08/18 – version 0.14
SIMMAN 2014: Shallow-water CFD Computations

(a) /5, coarse (b) /7, coarse

(c) /5, medium (d) /7, medium

(e) /5, fine (f) /7, fine

Figure 33 – View of the grid near the bow in the symmetry plane of the ship.

(a) /5, coarse (b) /7, coarse

(c) /5, medium (d) /7, medium

(e) /5, fine (f) /7, fine

Figure 34 – Overview of the grid in the symmetry plane of the ship.

Final version WL2015R00_066_3 36


fhr.cls
2015/08/18 – version 0.14
Waterbouwkundig Laboratorium

Flanders Hydraulics Research

Berchemlei 115
B-2140 Antwerp
Tel.+32 (0)3 224 60 35
Fax +32 (0)3 224 60 36
E-mail: waterbouwkundiglabo@vlaanderen.be
www.watlab.be

fhr.cls
2015/08/18 – version 0.14

Das könnte Ihnen auch gefallen