ITTC - Recommended Procedures and Guidelines: Seakeeping Experiments

ITTC – Recommended 7.
5-02
07-02.1
Procedures and Guidelines Page 1 of 22
Effective Date Revision

Seakeeping Experiments 2011 04
Table of Contents
1. PURPOSE OF PROCEDURE.............. 2 4. VALIDATION ....................................... 9
4.1 Uncertainty Analysis ......................... 9

2. SEAKEEPING EXPERIMENTS......... 2
4.2 Benchmark Tests ............................... 9
2.1 Model Size .......................................... 2
5. REFERENCES .................................... 10
2.2 Model Completeness .......................... 3
2.3 Model Weight Distribution ............... 3 APPENDIX A ............................................... 10
2.4 Guidance System ............................... 4 A.1 BACKGROUND TO ISO-GUM.......... 10
2.5 Free Running Tests............................ 4
A.1.1 Type A uncertainty ........................ 11
2.6 Measurement of Wave Loads ........... 4
A.1.2 Type B uncertainty......................... 11
2.7 Measurement of Added Resistance .. 4
A.1.3 Standard uncertainty ..................... 11
2.8 Measurement of Impact Loads......... 4
A.1.4 Combined uncertainty ................... 12
2.9 Parameters to be Measured .............. 5
A.1.5 Expanded uncertainty.................... 12
2.10 Headings ............................................. 5
2.11 Regular Waves ................................... 5 A.2 SOURCES OF UNCERTAINTY ........ 12
2.12 Transient Waves ................................ 6 A.2.1 Type A uncertainty ........................ 12
2.13 Irregular Waves ................................. 6 A.2.2 Type B uncertainty......................... 12
2.14 Data Presentation .............................. 7 Data presentation ..................................... 18
3. PARAMETERS ..................................... 8 A.2.3 Example........................................... 20
A.2.4 Summary ......................................... 20
3.1 Parameters to be Considered............ 8
3.2 Recommendations of ITTC for
Parameters ......................................... 8
Updated / Edited by Approved
26th ITTC Seakeeping Committee 26th ITTC
Date 04/2011 Date 09/2011

ITTC – Recommended 7.5-02
07-02.1

Seakeeping Experiments
0.7
1. PURPOSE OF PROCEDURE 0.6
Fr ωe (LM/g)1/2
This procedure outlines the recommended 0.5
state-of-the-art practice of model seakeeping 0.4
experiments for the evaluation of ship hull 0.3
performance in predefined operational and

0.2
environmental conditions.
0.1
The procedure describes requirements 0
relevant to the selection of model size, 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
BT/LM
completeness of its geometry, ballasting and
mass distribution and possible model
configurations. It provides recommendations Figure 1. Maximum frequency at which tank
for model response data measurements, and interference occurs in head waves
operational and environment parameters that
should be included in the test plan.
BT/LM Fr ⋅ ω LM / g
The procedure also outlines the
0.50 0.635
recommended approach to data analysis and
presentation formats as well as the preferred 0.75 0.458
approach to uncertainty analysis including 1.00 0.378
theoretical background and practical examples. 1.25 0.335
1.50 0.309
1.75 0.292
2. SEAKEEPING EXPERIMENTS 2.00 0.280
2.25 0.271
2.1 Model Size 2.50 0.265
2.75 0.260
The size of the model should be such that 3.00 0.255
tank wall interference is avoided for the range
3.25 0.252
of wave frequencies and model speeds to be
3.50 0.249
tested. Figure 1 and Table 1 give, in
3.75 0.247
dimensionless form, a relationship between
4.00 0.245
model length LM tank breadth BT , Froude
number Fr and the highest wave frequency ω Table 1. Maximum frequency at which tank
at which interference effects may occur in interference occurs in head waves
head waves.
Those calculations are made by estimating
the potential generated by a source with
07-02.1

Fr. ωe √(LM/g)
harmonic strength. Calculations using the 10
unified-slender ship theory were made by 9

Frh=0,7
8
Kashiwagi & Ohkusu (1991). 7 Frh=0,9
6
Frh=1,5
Figure 2 shows where tank-wall effects 5
are expected for a prolate spheroid of beam - 4 Fig. 1
length ratio 1/8. With K = ω2/g. The dotted

3
lines in Figure 2 show the results of Figure 1. 1
0
0 0.2 0.4 0.6 0.8 1
Non published work of Fernandez shows BT/LM
that the finite depth must be taken into

account in tank-wall effects for: Figure 3. Maximum frequency at which tank
interference occurs in head waves and finite
Fr ⋅ ω e LM / g ≤ 1 2 depth.
with ωe, the encounter circular frequency. 2.2 Model Completeness
These estimations use calculations of the It is desirable that the model is complete
potential generated by a source with harmonic up to the uppermost weather deck, including
strength in finite depth. Figure 3 shows results forecastle and bulwarks. A more complete
in the same format as Figure 1. modelling of deck fittings, deck houses and
freeing ports may be necessary if parameters
such as deck wetness are to be measured.
All appendages should be fitted, and the

report should state which appendages were
fitted during the experiments.
2.3 Model Weight Distribution
If bending moments, shears, and torsion

experienced by the model in waves are to be
measured, the longitudinal and transverse
distributions of mass must be reproduced as
correct as possible, and must be correctly
reported. In other cases, only the radii of
Figure 2. Estimation of tank-wall effects gyration need to be simulated. For tests in
using unified slender theory. head or following waves with a model
restrained in rolling, it is not necessary to
simulate the transverse weight distribution.
If the longitudinal radii of gyration for

pitch or yaw are unknown, a value of 0.25 LPP
should be used. If the transverse radius of
07-02.1

gyration is unknown, a value between 0.35B Care has to be taken to reduce any
and 0.40B, depending on the ship type, should influence of cables or safety lines on the
be used. (These values are those without model’s motions to a minimum.
including the effect of added mass).
It is recommended that rpm and rudder
For experiments during which rolling is action are continuously recorded.
not restrained, the metacentric height should
be simulated. If the vertical position of the 2.6 Measurement of Wave Loads
centre of gravity is unknown, it should be
established and reported. As an alternative to Segmented models for measuring global
ballasting the model to a specified transverse loads should have natural frequencies far from
radius of gyration, the natural period of the wave frequency range. These frequencies
rolling of the full-scale ship may be simulated. have to be measured and documented.
When measuring loads on catamarans, The mass, COG and inertias of each
cross products of inertia have to be taken into separate segment have to be known (measured
account. or calculated) and reported. Preferably, the
loads due to the mass and inertia of the
2.4 Guidance System segments should be separated from the total
loads during analysis to get the wave-induced
The guidance system should be such as to loads.
impose the minimum restraint on the motions
of the model. It is desirable that even in head For global bending moment, sagging and
or following waves the model should have the hogging loads should be reported.
freedom to roll. In oblique waves, care also
must be taken to minimize restraint on sway 2.7 Measurement of Added Resistance
and yaw motions.
The power increase in waves can be
The report should describe in detail the measured directly with free running models or
characteristics of the guidance system. determined indirectly from measurements of
added resistance on captive models (refer to
2.5 Free Running Tests ITTC recommended procedure 7.5-02-06-0.1).
Testing with a free running self-propelled 2.8 Measurement of Impact Loads

model is preferred method for seakeeping
experiments. Experiments are usually run at The guidelines for the measurement of
predefined speeds. Preliminary tests can be impact loads are presented in procedure 7.5-
necessary to adjust the rpm in order to reach 02-07-02.3 Loads and Responses Seakeeping,
the desired speed in waves. Alternatively, the Experiments on Rarely Occurring Events.
rpm can be automatically controlled.
The autopilot parameters should be chosen

to reflect a realistic full-scale response of the
model. These parameters should be reported.
07-02.1

2.9 Parameters to be Measured - Propeller revolutions. Whenever a self-

propelled model is used, the shaft
The hull motions, motion rates and revolutions should be recorded.
accelerations in the desired degrees of
freedom should be measured. - Visual records. Tests should be recorded
visually, preferably in a way allowing
Wave height measurements should be scaling of time.
made with a probe mounted close to the
model, but not causing interference. The Additionally, the following parameters
probe should preferably be fixed to the may be measured depending on the test
carriage, but measurements may be made at a requirements:
fixed point in the tank. In the latter case, the
measuring point should be selected in the - It is recommended that propeller torque
position where waves are fully formed and thrust be also continuously recorded.
without being affected by the waves reflected
at the wave maker and the tank walls & - Encounter (heading) angle. The angle
beaches. between the mean model heading and the
wave direction.
Non-contact probes are preferable for
wave measurements moving with the model, - Leeway (or drift) angle. The angle
especially at high speeds. between the mean model heading and the
tangent to the path of CG.
The capability to measure the following
additional parameters should be provided: 2.10 Headings
- Relative motion. Measurements of the When performing tests in oblique seas, the
relative motion between the model and the range of encounter angles between zero and
water surface at points that allow 180 degrees should be selected in accordance
correlation with wave and other motion with the stated test objectives. The 180
data. degrees heading represents head seas.
- Rudder angle. In cases where active 2.11 Regular Waves

rudder control is employed, the rudder
control signal and actual rudder angle For conventional ship forms, a sufficient
should be continuously monitored. number of tests should be carried out at each
speed to provide adequate data for a minimum
- Impact pressures on the hull or on deck at range of wavelengths from at least 0.5 LPP to
selected locations. 2.0 LPP. More tests with closely spaced
wavelengths can be necessary to ensure a
- Still water resistance and added resistance good definition in the resonance region.
in waves (if not freely running). Either the ratio of the wave height to LPP or
the ratio of wave height to wavelength should
- Water on deck. be maintained constant. (The recommended
07-02.1

value of the ratio of wave height to fetch-limited seas. When generating irregular
wavelength is around 1/50.) waves in a tank, the input signal to the wave
maker should be produced such that the
In determining the motions, it is generated waves are non-repeatable.
recommended that the average amplitude and
Irregular wave generation in experimental
period of at least 10 cycles be obtained.
tanks is subjected to voluntary or involuntary
Alternatively, a spectral analysis following
truncation of idealized spectrum as a result of
the procedures for irregular waves outlined
mechanical limits of wave making facilities.
below could be followed to obtain the mean
amplitude and period of waves and responses. The truncation frequency is facility specific
and depends on characteristics of the wave
Guidelines for regular wave data analysis are
given in the ITTC Recommended Procedure maker and model scale selected for the
experiment. Selection of too low cut-off
7.5-02-07-03.2 “Analysis Procedure for
frequency affects properties of resultant
Model Tests in Regular Waves”.
spectrum and values of target significant
2.12 Transient Waves wave height HW1/3 and modal period TP. If
n=fT/fP: fT is truncated frequency and fP is
The transient wave technique is an peak frequency of idealized spectra, the
experimental technique in which a wave train recommended cut-off frequency for most
that contains wave components of all the facilities is n>2, and preferable n approaching
relevant frequencies is produced in such a 3.
way that the component waves reach a certain Data should preferably be digitised before
place in the test tank simultaneously so that a analysis, using sample rates appropriate for
single large wave packet is formed. If a model the avoidance of aliasing with the individual
structure is positioned at the place where the measured parameters. Care must be taken for
single large wave packet accumulates, the duration of the data acquisition so that
response characteristics to regular waves of enough data are recorded for the objective of
all the frequencies contained in the wave the test.
packet are obtained in one single experiment
(provided the linear superposition assumption The test duration is represented by total
holds). This technique proves to be very number of waves (encounters) N. The N=50
efficient as a standard tool for evaluating should be taken as a lower limit. Larger
RAO’s of stationary offshore structures or values are to be preferred and it is more usual
towed/self propelled ships. to take N=100 as the standard; N=200 or
above is considered excellent practice. For the
2.13 Irregular Waves following sea case, 30 minutes of equivalent
full scale is considered sufficient.
Tests should be carried out in waves
corresponding to the sea conditions in which The time interval between test runs is also
the vessel may be required to operate. In the important and can be tank specific. In most
absence of specific wave spectrum data the cases 20 minutes between runs is acceptable
ITTC should be used for open ocean and for a typical facility. The residuary tank
JONSWAP spectrum should be used for
07-02.1

disturbance of less than 1% of the next target Rudder angles may be non-dimensionalized
wave height is a valid alternative. by wave slope or be presented in other
appropriate non-dimensional form.
The sample rate in the data acquisition
needs to be fast enough in order to achieve
Translations x1, 2,3
sufficient resolution. A sampling rate
corresponding to about 4 Hz at full scale is ςA
enough for most measurements but much Rotations x 4 , 5, 6
higher rates (in the order of kHz) are
κς A
necessary to detect peaks of slamming loads.
Energy spectra of waves and relevant Dimensional presentations can sometimes

responses should be produced through be more appropriate depending on the
spectral analysis using either the indirect objectives of the experiment. Phase angles
method of Fourier transformation of the should be given in degrees and increases in
autocorrelation function, or the direct method resistance and propulsion parameters should
of splitting the record into suitable blocks and be presented in the non-dimensional form.
subjecting these to Fast Fourier Transform. Accelerations should be made non-
dimensional by LPP /ζ( g A ) . It is
In addition to the spectral analysis, recommended that the results are plotted to a
base of ω (LPP /g )1/2 or ωe (LPP /g ) , although,
statistical analysis should be performed to 1/2
produce at least the mean, maximum,
minimum, and the mean of 1/3 highest values. depending on the objectives of the experiment,
In the presentation of the results the other bases such as wavelength - ship length
techniques utilised to smoothen spectral ratio or wavelength may be appropriate. The
shapes, such as block overlapping, should be limit of tank wall interference effects should
documented. When reporting statistics, the be indicated on the plots.
number of events and number of encounters
should also be reported together with the For tests in irregular waves, the
overall statistics. corresponding wave-energy spectrum should
be defined.
For the measurement and analysis of
rarely occurring events such as slamming or When appropriate, performance in
wetness refer to ITTC recommended irregular waves should be presented in non-
procedure 7.5-02-07-02.3. dimensional form involving a characteristic
wave period or frequency and a characteristic
2.14 Data Presentation wave height.
The coordinate system in which data are The results of statistical analyses may be
presented should be defined. Motion presented to depict probability of exceedance
components should also be defined. Linear and as cumulative probability distribution for
translations and rotations may be presented in selected responses.
non-dimensional form as being divided by
wave elevation and wave slope respectively.
07-02.1

Tabular presentation of results is involving wave characteristic period and

recommended in addition to plots. characteristic wave height.
1978 Recommendation for open ocean

3. PARAMETERS spectral formulation:
S (ω ) = 5 e − B / ω
A 4
3.1 Parameters to be Considered (1)

ω
The following parameters defining the where
A = 173(ζ~W )1 / 3 / T14
2
tests are to be taken into account (as
applicable): B = 691 / T14
T1 = 2πm0 / m1
• Scale
• Model dimensions 1984 Recommendation for long crested
• Ratios of model to tank dimensions limited fetch sea spectral formulation:
• Hull configuration (lines, appendages,
superstructures, ...)
( ) ~ 2
ζ W 1/ 3  944 
S J (ω ) = 155 4 5 exp − 4 4 3.3γ (2)
T1 ω  T1 ω 
• Loading conditions (displacement and
draft) where:
• Mass distribution (COG, inertias, ...)  (0.191ωT1 − 1)2 
•
γ = exp  − 
Towing and/or restraining device
 2σ 2 
characteristics (specially DOF) (3)
• Speeds and headings 0.07 ω < 5.24 / T1
σ =
• Wave characteristics (heights, periods, 0.09 ω > 5.24 / T1
spectra, dispersions, ...)
• Autopilot control law and gains This formulation can be used with other
• Speed control characteristics characteristic periods by use of the following
• Run duration approximate relations:
• Number of runs per test condition T1 = 0.924T−1 = 0.834T0 = 1.073T2
• Positions of sensors (accelerometers, Where T−1 is the energy average period
relative motion, encountered wave, ...) ( 2πm−1 / m0 ), T0 is the spectral peak period, T1
• Resonance frequencies for segmented
models is the average period ( 2πm0 / m1 ) and T2 is the
• Sampling frequency average zero crossing period estimated from
• Sensor calibrations and accuracy the spectrum ( 2π m0 / m2 ).
3.2 Recommendations of ITTC for

Parameters
1975 Performance in irregular waves

should be presented in non-dimensional form
07-02.1

4. VALIDATION 4-1) Comparative Tests of a Series 60 Ship

Model in Regular Waves (11th ITTC, 1966,
4.1 Uncertainty Analysis pp.411-415). Series 60 with CB=0.60.
The detailed procedure of uncertainty 4-2) Experiments on Heaving and Pitching

analysis following the principles behind the Motions of a Ship Model in Regular
ISO-GUM is shown in the Appendix A. Longitudinal Waves (11th ITTC, 1966,
pp.415-418). Series 60 with CB=0.60.
4.2 Benchmark Tests
4-3) Experiments on the Series 60 with
1) Seagoing Quality of Ships. (7th ITTC, CB=0.60 and 0.70 Ship Models in Regular
1955, pp.247-293). A model of the Todd- Head Waves (11th ITTC, 1966, pp.418-
Forest Series 60 with CB=0.60. Results 420)
from 7 tanks are presented.
Fr = 0, 0.18 ,0.21 ,0.24 ,0.27 and 0.30 4-4) Comparison of Measured Ship Motions
LPP / H = 36, 48, 60, 72 and Thrust Increase of Series 60 Ship
λ / pp = 0.75, 1.0, 1.25, 1.5 Models in Regular Head Waves (11th
ITTC, 1966, pp. 420-426).
2) Comparative Tests at Three Experimental 4-5) Estimation of Ship Behaviour at Sea
Establishments with the Same Model. from Limited Observation (11th ITTC,
(11th ITTC, 1966, pp.332-342) 1966, pp.426-428)
British Towing Tank Panel: A 10 ft. Fibre-
glass model of the S.S. Cairndhu. 5) Analysis of the S-175 Comparative Study
A series of experiments on a ship model in (17th ITTC, 1984, pp.503-511).
regular waves using different test
techniques. 6) S-175 Comparative Model Experiments
Data obtained in irregular and transient (18th ITTC, 1987, pp.415-427)
waves and some result predicted by the
theory (based on Korvin Kroukovsky's 7) Rare Events (19th ITTC, 1990, pp.434-
work and employing the added mass and 442). Comparison of results from tests at
damping coefficients calculated by Grim). 12 establishments in irregular waves.
Absolute and relative motions. S-175 at
3) Full Scale Destroyer Motion Tests in Head Fr =0.275.
Seas (11th ITTC, l966, pp.342-350).
Comparison among motion response 8) The ITTC Database of Seakeeping
obtained from full scale tests, model Experiments (20th ITTC,1993, pp.449-
experiments and computer calculations for 451).
destroyer H.M. "Groningen” of the Royal
Netherlands Navy 8-1) Tests of Two Dimensional Models.
Added mass, damping and wave exciting
4) Experiments in Head Seas For Series 60. forces
07-02.1

8-2) Tests of a Wigley hull form. Added Kishev, R., (1998), “Uncertainty Analysis of
masses, damping, exciting forces and Dynamic Model tests and Its Application
seakeeping motions and loads. to Capsizing Simulation in Waves”,
Proceedings of International Workshop on
8-3) Tests for S-175. Modeling of Ocean Environments in
Waves & Current Basin, February, 1998,
9) The ITTC Database of Seakeeping Taejon, Korea.
Experiments (21st ITTC, 1996, pp.43). S-
175, high speed marine vehicle Yum, D.J., Lee, H.Y. and Lee, C.M., 1993.
Uncertainty Analysis for Seakeeping
10) Numerical and Experimental Investigation Model Tests. Journal of the Society of
to Evaluate Wave-Induced Global Design Naval Architects of Korea, 30 (3): 75-89.
Loads for Fast Ships (Schellin et al, 2003).
Two segmented models of fast ships (Fr
up to 0.63) were tested in head seas.
Motions and global loads are reported.
The results are compared with several Appendix A
non-linear codes.
A.1 BACKGROUND TO ISO-GUM
5. REFERENCES
The recommendation of the ITTC 2008
Kashiwagi, N. and Ohkusu, M., 1991, “A new was to adopt the ISO-GUM (International
theory for side-wall interference effects on Organization for Standardization, Guide to
forward-speed radiation and diffraction the Expression of Uncertainty in
forces”, Schiffstechnik, vol. 38. Measurements, ISO 1995) approach to
conducting uncertainty analysis of
Schellin, T.E. et al, 2003, “Numerical and experimental results. The ISO GUM
Experimental Investigation to Evaluate recognises two groups of uncertainty, type A
Wave-Induced Global Design Loads for and type B, which are based on way in which
Fast Ships”, Transactions SNAME, vol. the uncertainty is evaluated. Type A
111, pp. 437-461. represents the random category of uncertainty
evaluated by using statistical analysis of
ISO/IEC 2008, “Guide 98-3, Uncertainty of repeated measurements of, nominally, the
measurements – Part 3: Guide to the same observation; type B components are
expression of uncertainty in measurements estimated by means other than repeated
(GUM: 1995)”, 2008 Switzerland. observations. The “other means” may include
previous measurements, past experience or
ITTC (2008), “Discussion to the 25th ITTC general knowledge, handbook information,
Seakeeping Committee by Joe Longo, manufacturer specification or data provided as
Claus Simonsen, Fred Stern”, Proceeding a certificate. A detailed approach to
of 25th ITTC-Volume III, September 2008, uncertainty analysis in experimental
Fukuoka, Japan, pp. 697-701. hydrodynamics can be found in ITTC
procedure 7.5-02-01-01.
07-02.1

A.1.1 Type A uncertainty quoted values of uncertainty, assumed

statistical distribution of the parameters and
The fundamental form of uncertainty factors depending on a level of confidence in
associated with a measurement is type A, the measurement. Generally, the
u (q ) , which can be expressed as a standard experimenter can assume that the type B
deviation. Type A uncertainty is typically uncertainty is normally distributed around
based upon the analysis of repeated some mean, however, in some specific cases
measurements which characterizes the is may be pertinent to consider alternatives
randomness of the experimental process. The such as triangular or rectangular distributions.
most common approach to estimating type A For type B uncertainty that are assumed to be
uncertainty is by undertaking end-to-end normally distributed Table A1 shows the
multiple repeated runs; care should be taken factors that need to be applied for some
to ensure that as many factors as possible that examples of confidence.
affect repeatability of experiment are
accounted for. Numbers of repeats should be Confidence Factor
as large as practicable in order to minimize Level [%]
type A uncertainty; however 10 repeats 50 0.6757
indicates good experimental practice. 68.27 1.
However, in most seakeeping tests it is not 90 1.645
practicable to carry out multiple repeats for all 95 1.96
experimental conditions. It may be more 99 2.576
feasible to select only characteristic or unique 99.73 3
test conditions (due to environment and/or
operations) for which repeat runs should be Table A1. Confidence factors for normally
undertaken and reported. Historic database of distributed type B uncertainties
information on Type A uncertainty could be
created (occasionally confirmed) and used to For example, this means that, statistically,
report uncertainty for routine experiments. you can have 95% confidence that a
measurement lies within a value of ±1.96 ui .
A.1.3 Standard uncertainty
The standard uncertainty, u(y), in a

A.1.2 Type B uncertainty measured value is the summation of type A
and all of the type B uncertainties and can be
Type B uncertainty, characterised as ui2 or calculated using the uncertainty propagation
ui , may be considered as an approximation to formula:
1
the experimental variance or standard  N K  2
deviation respectively. In the same way as u(y) =  ∑ u 2 ( q) + ∑ u i2  (A1)
type A uncertainty, type B is assumed to be  i =1 j =1 
equal to the standard deviation ui . Typically
type B uncertainty can be estimated from
07-02.1

A.1.4 Combined uncertainty within the range y-U and y+U; the value of U
is defined by k. For cases where the
A further step is required when result of uncertainty can be assumed to be normally
an experiment is derived from values of a distributed the confidence factors presented in
number of other measurement variables (xi). Table 1 can be used. For example, a value
The most common situation where this is k=2.576 value gives confidence level of 99%.
undertaken in seakeeping experiments is
when the results are non-dimensionalised. In
this case, the combined uncertainty uC(y) is A.2 SOURCES OF UNCERTAINTY
applied to express uncertainty in the derived
result. A typical requirement from a seakeeping
1 experiment is to obtain the basic rigid body
 N  ∂f  2 N −1 N
∂f ∂f  2
uC (y) =  ∑  u(xi , x j 
 u 2 (xi ) + 2 ∑ ∑ motions (surge, sway, heave, roll, pitch and
 i =1  ∂xi  i =1 j =i +1 ∂xi ∂x j
 yaw), accelerations and relative motions at
 
specific locations, waves, model speed, and
propulsion and steering systems
(A2)
characteristics (propeller revolutions, rudder
The second term in the combined angle). All of these measured parameters are
uncertainty formula represents the cross subjected to type A and type B uncertainties
correlation between two or more variables. that need to be estimated as a part of the
Theses terms are zero when variables are experimentation procedure.
∂f
considered to be independent. The term A.2.1 Type A uncertainty
∂xi
is the partial derivative with respect to Type A uncertainty is evaluated by taking
variable xi, also known as the sensitivity repeated measurements of the same
coefficient and u(xi) is the standard experimental condition (recommended
uncertainty of variable xi. number of repeated runs is 10). Since
repeating the entire set of test runs in a
A.1.5 Expanded uncertainty seakeeping experiment makes the programme
prohibitively long (and hence expensive) it is
When presenting the results of recommended that only a few selected
experiments along with interval expressing representative test conditions should be
some level of confidence in that measurement repeated to obtain some understanding of the
then the expanded uncertainty U is applied. type A uncertainty.
U = kuC (y) (A3) A.2.2 Type B uncertainty
Where, k represents the confidence factor, There are elemental type B uncertainties
and the result of the measurement can be that are an inherent part of each sensor, its
interpreted as y-U ≤Y≤ y+U. calibration, the data acquisition system,
processing and analysis.
So, Y can be interpreted as the best
estimate that the resultant measurement lies
07-02.1

All of these elemental type B uncertainties Before used in experiments, all

should be accounted for, using same type of instruments need to be calibrated; either
formulation in equation A2, to determine the bench or in-situ calibration or else factory
type B uncertainty for each measured calibration constants are applied.
parameter.
Calibration characterises an instrument’s
Sensors. uncertainty but does not eliminate it; indeed,
the calibration process itself is subject to
Measurements of the rigid body motions uncertainties. Generally, a system level, in-
of the model, accelerations and relative situ end-to-end calibration is advisable that
motions, propulsion and control parameters includes as many of the possible elemental
are usually primary requirements of sources of uncertainty in the calibration
seakeeping experiments. Specifications procedure. A few, additional, elemental
provided by the manufacturers of the sensors sources of uncertainty need to be considered
used in experiments, coupled with past when estimating uncertainty: calibration
experience in the use of such sensors, allows standards (quality of calibration specimens or
an estimation of the relevant type B injection source), calibration curve fitting,
uncertainty to be made. The manufacturer calibration set up (misalignments) and A/D
may present sensor uncertainty information as conversion.
standard deviations (or multiples of) or as an
expanded uncertainty with a specified The uncertainty associated with the
confidence level. This information can be quality of the calibration standard and
translated to a standard deviation and can be calibration device/jig set-up misalignments
used to obtain the standard type B uncertainty can be estimated from the manufacturer’s
for that particular element. For example, a specification. Uncertainty due to calibration
sensor specification stating that roll and pitch standards BCS can be estimated using:
angles are measured to a dynamic accuracy of
BCS = ∑ (A ⋅ Wi )
2
0.5 degrees rms can be interpreted as a 0.5 CG (A4)
degree standard uncertainty in roll and pitch.
In most cases individual sources of ACG accuracy of calibration
uncertainty need to be identified from specimens; e.g., weight, distance, angle…
available specification documents and the
uncertainty propagation formula should be Wi physical values of calibration
used to obtain the standard uncertainty given points; weight, distance, angle…
in (A1).
The uncertainty associated with
Elemental sources of uncertainty that are misalignment in the calibration set-up
usually identified from manufacturer’s uncertainty can be evaluated from:
specification may include: non-linearity,
hysteresis, non-repeatability, zero offset drift, BCM = ∑ (W (1 − cos α j ) (A5)
spam temperature coefficient, and resolution.
W nominal measurement value
Calibrations.
07-02.1

α angle of misalignment in In the case of measuring instruments that

relevant plane. are provided with manufacturer calibration
data (most modern digital instruments)
The curve fitting uncertainty can be calibration standards are reflecting standards
estimated using the standard error of of high precision source (voltage) that,
estimation (SEE) formula: normally, is expected to be considerably more
1/2
 1 accurate than accuracy that can be achieved in
(y n − y LS,k )2 
n
SEE =  ∑ (A6) a physical bench calibration.
 n(n − 1) 1 
It is advisable, if practical, to conduct in-
In the formula n is the number of calibration situ end-to-end (with all model systems being
samples, yn is calibration data point, and yLS,n active) calibration of the sensors that are to be
is fitted value. In most cases n≥ 7 is used in the experiment. In such a situation,
recommended. It can be assumed that the SEE the calibration process should include all or
value is approximately equal to the standard most elemental type B uncertainty sources,
uncertainty. which are difficult to estimate individually.
However, for in-situ calibration they don’t
Generally, the majority of data acquisition have to be individually identified and
systems that are currently in use employ a 16- estimated.
bit (or better) analogue to digital (A/D)
converters. However, some specific This approach does not exclude the need
equipment may still use 12-bit A/Ds to for uncertainty analyses due to calibration
acquire model data. standards, set up, curve fitting and other
related sources of uncertainty but hopefully
The type B uncertainty associated with the overall simplifies the procedure.
A/D conversion is equivalent to ½ the
resultant resolution and can be estimated from:
1 TotalVoltageRange Data Acquisition System.

BCAD = ⋅ CalibrationFactor
2 A / Dbits
(A7) In case when in-situ end-to-end
calibration procedure is applied all data
The TotalVoltageRange is typically equal acquisition system elemental error sources are
to either ±10 Volts or ±5 Volts; the A/Dbits included in the process except for noise due to
value is 216 or 212 for 16 and 12 bit convertors variation in surrounding external environment
respectively, the CalibrationFactor is a (temperature, humidly) and other used
calibration constant that translates voltage to devices (propulsion motor). Good testing
physical units. Typically, uncertainty due to practice requires screening of all noise
resolution of 16-bit system would be sources, but when this appears to be difficult
negligible, but for 12-bit system it could be those effects should be estimated.
significant for higher precision instrument.
07-02.1

Data processing. n / 5000πD fπD

v= = (A8)
t 5000
Type B uncertainty due to data reduction
and analysis should include any uncertainty where D (m) is the diameter of carriage
related to data integration, differentiation, wheel and n is the number of light pulses
filtering and other methods of data sensed by the photo coupler during the time
manipulation. It can be evaluated based on period t . The 5000 number is facility specific
previous experience of working with data and indicates number of pulses per single turn
processing systems. Uncertainty due to any of carriage wheel. The measured quantities
data reduction associated with the calculation and error sources for the estimation of model
of basic statistics (mean, standard deviation) speed and error limit are the diameter of
should be considered negligible, however, for carriage wheel and the pulse frequency f (=
more complex data manipulations resultant
uncertainty may need to be considered. These n t ).
uncertainties can be estimated by using the
same data manipulation process with a known The combined uncertainty becomes, in
signal with known analytical solution (sin or this case:
cosine) comparing the processed and analytic 1
outputs.  ∂v 
2 2 
2
 ∂v  2
u C (v) =   u 2 (f) +   u (D) (A9)
Data analysis.  ∂f   ∂D  
Model speed and heading. - Model speed 1

2
uncertainty is subjected to both type A and  πD  2 
 πf  2
2 2
type B uncertainties. The type A component u C (v) =   u (f) +   u (D)

is calculated using equation for the standard  5000   5000  
uncertainty (the uncertainty propagation
formulae A1) and the combined uncertainty (A10)
(A2). The Type B uncertainty component is
dependent upon the method in which the
model speed is obtained. If, during the
experiments, the model is attached to or If a free running model is used in the
follows the carriage and speed of the model experiments and, for example, an optical
can be assumed to be equal to the speed of the tracking system is used for to determine
carriage, then the method presented in ITTC model position, then v=s/t should be applied,
2008 7.5-02.07-02.1 and that suggested by and the instantaneous and/or mean speed can
Fogash (1992) can be used. be calculated (s is distance between two
consecutive sampled positions, and t is time
Under the assumption that model speed, v , between two consecutive samples). In this
through the water is equal to the speed of case, the combined uncertainty formula can
towing carriage, the model speed is be used to obtain model speed uncertainty:
determined from
07-02.1

1
2 (A13)
 ∂v  2
 ∂v 
2

uC (v) =   u 2 (s) +   u 2 (t) (A11) 1
 ∂s   ∂t   ∆Y  1  2  ∆Y 
2
2
) =   u (∆Y ) +  −  u 2 (∆X )
2
u(
∆X  ∆X   (∆X )
2
 
1
2
 1 
2
 s 
2

u C (v) =   u 2 (s) +  − 2  u 2 (t) (A12) (A14)
 t   t  
Nominal ∆X and ∆Y values are calculated
Standard uncertainties u(s) and u(t) need to be from the mean heading angle, and the
estimated based on information provided on appropriate uncertainty can be used to
model positions and accuracy of sample time. calculate uncertainty in the ratio.
Nominal values of s and t should be applied to
In case when model heading is obtained
obtain combined uncertainty.
after double integration of yaw rate
Similarly if a captive model is used or free measurement, both uncertainty of yaw rate
running model follows the carriage the measurement and accuracy of integration
heading angle is assumed to be equal to the procedure need to be included in combined
heading of the carriage with respect to the uncertainty estimate.
oncoming waves. For free running, self-
Model geometry and mass distribution -
propelled models when an optical system is
sources of uncertainty in model geometry are
used to obtain model positions the
model length (LPP), width (B) and draft (T).
instantaneous (and mean) heading angle can
For seakeeping experiments the position of
be estimated from consecutive longitudinal
centre of gravity (KG) and longitudinal radius
and lateral positions of the model. The
of gyration (kyy) are also important and their
estimate of combined uncertainty in heading
respective uncertainties need to be determined.
is then based on the uncertainty in the lateral
and longitudinal position of the previous and Typical suggested tolerances on the
next location of the model, and the nominal principal parameters associated with model
longitudinal and lateral distance between geometry are +/-0.05% on linear dimensions
those two points. The arctangent is applied to larger than 2m, and +/-1mm on dimensions
estimate the uncertainty in the angle based on less than 2 m, and +/-1% on model
the uncertainty in the ratio of the lateral (∆Y) displacement. In all cases they are the type B
and longitudinal (∆X) consecutive positions. uncertainties that are constant for the duration
Uncertainty in the ratio can be calculated of experiment. Examples of achieved and/or
from: suggested uncertainties of model main
1
  ∆Y   2 2
2 parameters and mass properties as well as
  ∆Y  
 ∂   ∂   presented results are shown in Kishev (1998)
∆Y ∆X   2 ∆X   2
u( ) =   u (∆Y ) +   u (∆X ) and ITTC (2008).
∆X  ∂ (∆Y )   ∂ (∆X )  
    
     Uncertainties in model geometry can be
determined using past experience in model
construction. For instance, if a model
07-02.1

manufacturer states that a 5-metre long model 2

 ∂GM  2  ∂GM  2
2
is accurate to within +/-2.5 mm with 90% uC (GM )= [   u ( w) +   u (d ) +

confidence, then one can assume that the  ∂w   ∂d  (A17)
2 2
 ∂GM  2  ∂GM  2
1
expanded uncertainty of the model length is
 u (ϕ ) ]
2
  u (W ) + 
+/-2.5 mm. The standard uncertainty can be  ∂W   ∂ϕ 
estimated, using the confidence factor in
Table 1, from expression 2.5/1.645=1.52  d  2
2
 w  2
2
assuming that a normal distribution can be uC (GM )= [   u ( w) +   u (d ) +

 W ⋅ tgϕ   W ⋅ tgϕ  (A18)
applied to represent the stated value. So, the 2 2
 w⋅ d  2  w⋅ d  2
1
 u (ϕ ) ]
2
resultant standard uncertainty of the length is − 2  u (W ) +  −
 W tgϕ   W ⋅ sin 2 ϕ 
~1.5 mm.
To estimate the uncertainty in the model
Standard uncertainty of KB and BM can
KG and kyy,, the propagation of uncertainty
be evaluated by assuming a simplified
needs to be applied to the formula used to
geometry of hull form and using known
calculate these respective values. For example
standard uncertainties of main parameters.
if KG of a model is estimated based on
inclining experiments and the following For example, the transverse BM for a
formula is employed: KG =KB + BM − GM . triangle-prism shaped vessel with a
The vertical centre of buoyancy (KB) and rectangular water plane area can be calculated
transverse metacenter ( BM ) are geometry from:
dependent, when metacentric height ( GM )
can be obtained from inclining experiment. I LB 3 LB 3 B 2 (A19)
BM= = = =
The combined uncertainty in KG can be V 12∇ 12 ⋅ 1 LBT 6T
evaluated from: 2
1
 ∂ KG 2 2
 ∂ KG  2
2
 ∂ KG  2 2
uC ( KG ) =   u ( KB) + 
2
 u ( BM ) +   u (GM ) 
 ∂ KB   ∂ BM   ∂GM   Where, I is second moment of rectangular
(A15) water plane area about its centreline, ∇
The standard uncertainty of GM can be volume of displacement, and L, B and T are
estimated by applying combined uncertainty length and breadth of water plane respectively
formula to: and T is draft of the vessel.
w ⋅d
GM = (A16) The combined uncertainty of BM is:
W ⋅ tan(ϕ ) 1
 ∂ BM 2  ∂ BM  2  (A20)
2 2
uC ( BM ) 
=  u ( B) + 
2
 u (T ) 
Where, w is inclining weight, d is distance  ∂B   ∂T  
the inclining weight is moved, W model 1
 B  2  B2  2 
2 2
displacement, φ is heel angle when inclined. =
uC ( BM )   u ( B ) −  2  u (T )  (A21)
2
 3T   6T  
The combined uncertainty of GM can be
presented as: Nominal B and T values, and their
respective uncertainties need to be applied to
07-02.1

calculate combined uncertainty in transverse analysis can be also be employed to determine

BM . significant height of irregular waves
H W1/3 = 4 m0 , where m0 is area under the
A similar procedure can be used to
energy spectrum curve.
evaluate the combined uncertainty in the
vertical location of centre of gravity and Total standard uncertainty in wave
longitudinal radius of gyration kyy that could amplitude or height measurements should be
be obtained a pendulum experiment. evaluated using the uncertainty propagation
formula. Type A uncertainty can be evaluated
Wave parameters – the uncertainty in
from repeated observations, although this can
wave measurements (regular and irregular) is
be impracticable for seakeeping experiments,
one of major sources of uncertainty in
and type B uncertainty established from
experiments. Limitations of wave generators,
properties of measuring device and data
the deterioration of wave properties
process.
propagating forward of experimental facilities
and reflections from beach devices contribute Wave direction is also a significant
to uncertainty in the wave environmental. parameter when undertaking experiments in
Those uncertainties are difficult to estimate oblique waves. Verification of waves
and are usually neglected. Target irregular propagation direction can be carried out using
wave properties are normally defined as many instruments. One possible choice for the
significant wave height, modal period and validation of wave direction with respect to
type of spectrum. Target regular wave the tank could be by using a 3D acoustic
properties are described by wave amplitude Doppler velocimeter. Periodic repeated wave
and frequency. Wave matching is normally measurements for selected wave directions
conducted based on a measurement in one can be carried out to determine standard
selected-representative location, and deviation and standard uncertainty. The
supported by measurements in a few other direction can be verified during wave
locations to check for consistency. Two matching for a specific experiment.
sources of error for which uncertainty could
be estimated are difference between the target Data presentation
and matched wave(s) and uncertainty due to
measuring and processing errors. It is customary to present the final
experimental results in a standardized format -
Either regular or irregular wave properties usually non-dimensionalised. The linear
are generally obtained from measurements of translations from regular waves tests are
wave displacement using devices such as a typically non-dimensionalised by wave
sonic wave probe and/or capacitance wave amplitude; rotations by wave slope and
probes. Basic statistics from measurements amplitude, and accelerations by LPP/(g·ζA).
provide rms that can be used as a first Generally these non-dimensional responses
estimate of the amplitude of regular waves are presented to a base of the non-dimensional
( ς A = 2 ⋅ rms ) and significant height of L
encounter wave frequency given as ωe PP
irregular waves ( H W1/ 3= 4 ⋅ rms ). Spectral g
07-02.1

Therefore, the combined uncertainty uc of For example, formulae for non-

the non-dimensional heave displacement dimensional pitch motion and the resulting
z' = z/ζ A can be calculated from the following: combined uncertainty are as follow:
θ ⋅ L pp
Cθ = (A26)
 ∂z ' 2
 2  ∂z '
2
 2 
1/2
2π ⋅ H W1 3
u C (z ' ) =   u (z) + 
  ∂ζ
 u (ζ A )

(A22)
 ∂z 
2
 ∂Cθ  2
2
  A   ∂Cθ  2  ∂C  2
2
  uC (cθ ) =  u (θ ) +   u ( HW 1/3 ) +  ϑ  u ( LPP )
 ∂θ   ∂HW 1 3   ∂LPP 
1/2
 1  2  z 
2

uC ( z )   u 2 ( z ) +  − 2  u 2 (ζ A )  (A23)
= ' Where:
 ζ A   ζA  
∂Cθ LPP
=
Where, z and ζA are heave displacement ∂θ 2π ⋅ H W1/3
and regular wave amplitude, u(z) and u(ζA) ∂Cθ θLPP
= −
are respective total standard uncertainties of π
∂H W1/3 2 ⋅ H W1/3
2
measured heave displacement and wave ∂Cθ θ

amplitude including all type A and type B =
∂L ppπ 2 ⋅ H W1 3
elemental error sources.
Similarly, combined uncertainty of non- θ is the significant pitch angle response in

dimensional encounter frequency irregular wave.
L
ωe' = ωe PP can be evaluated from: Combined uncertainty in estimation of
g
1/2 Froude number ( Fr = V / gLPP ) can be
 ω '  2 2
 ωe'  2 
uC (ωe )   u (ωe ) + 
= ' e 2
 u ( LPP )  (A24) expressed as follow:
ω
 e   LPP  
2
 ∂Fr  2  ∂Fr  2
2
   2  2 1/2
=
uC ( Fr )   u (V ) +   u ( LPP )
L  ωe  ∂V   ∂LPP 
uC (ω )  PP  u 2 (ωe ) +  −
= '
 u ( LPP ) 
 g 
e
 2(L ⋅ g ) 
1/2

  PP   (A27)
(A25)
Where,
Again, ωe, LPP and g are respective
nominal values, and u(ωe) and u(LPP) are ∂Fr 1
respective standard uncertainties. =
∂V gLPP
Motions in irregular seas are typically
presented as plots of non-dimensional or ∂Fr V
= −
significant values versus velocity, Froude ∂LPP 2 gL3PP
number or sea state.
07-02.1

Once the experimental data have been contained below 1%, and that the type B
collected and reduced to non-dimensional uncertainty is dominating model motions
format for a particular wave encounter measurements.
frequency and/or Froude number, they can be
presented in a tabular format or we may want A.2.4 Summary
to obtain a mathematical expression to
represent the data. In this case regression can The above presented procedure outlines ITTC
be performed on the experimental data (after recommended ISO GUM approach
data reduction) and a polynomial equation fit tokjlnashakijslk uncertainty analysis in
to represent the data. The type B uncertainty seakeeping experiment measurements.
associated with the regression should be Intention of the procedure is to emphasize
included in the analysis. details unique for seakeeping experiment
measurements and data presentation.
A.2.3 Example Background information for ISO GUM
approach and assumptions are discussed in
Table A2 and Figure A1 present examples ITTC Specialists Committee on Uncertainty
of total standard and combine uncertainty Analysis procedure 7.5-02-01-01. The
calculation of model parameters and methodologies presented here are relevant to
responses from submarine model seakeeping uncertainties in measurements only. Subjects
surface experiments in irregular seas. of uncertainty in predictions
From the table one can conclude that the

uncertainty in the model main parameter is
07-02.1

.
Source of uncertainty
Type A Type B Standard
Units Description of accuracy
Nominal uncertainty uncertainty uncertainty
Value
Model Lpp 4.70 m +/- 3 mm, 90% confidence 0.0018 0.002
Model B 0.51 m +/- 2 mm, 90% confidence 0.0012 0.001
Model T 0.52 m +/- 2 mm, 90% confidence 0.0012 0.001
Model D 671.14 m3 resultant 0.0042 0.004
Model KB 0.29 m resultant 0.0012 0.001
Model BM 0.302 m resultant 0.0004 0.000
Model KG 0.264 m inclining experiment 0.0015 0.002
Model kxx 0.213 m swing frame 0.0024 0.002
Model GMt 0.038 m resultant 0.0008 0.001
Speed 1 3.4 knots Optical tracking Qualisys 0.057 0.012 0.059
Roll Angle 1 13.8 deg FOG 0.190 2 2.009
Pitch Angle 1 3.5 deg FOG 0.063 2 2.001
Heave Displ. 1 2.12 m Motion Pack 0.030 0.030
Vert. Accel. 1 0.16 g Honeywell, QA 1400 0.001 0.0031 0.003
Relative Mot. 1 1.33 m ULS, USS 635 0.018 0.0013 0.018
Wave Elev. 1 2.62 m Capacitance probe 0.004 0.004 0.005
Combined
Uncertainty
Froude no. 1 0.07 0.0022

Froude no. 2 0.12 0.0024
Froude no. 3 0.25 0.0041
Table A2. Estimate of uncertainty for responses during seakeeping experiments
07-02.1

20.00 3.00
Roll Angle
Pitch Angle
2.50
Rel. Mot.
15.00
Relative Motions [m]

2.00
Angles [deg]
10.00 1.50
1.00
5.00
0.50
0.00 0.00
0.00 0.10 0.20 0.30
Froude Number
Figure A1. Example of responses and range of uncertainty

ITTC - Recommended Procedures and Guidelines: Seakeeping Experiments

Uploaded by

Copyright:

Available Formats

ITTC - Recommended Procedures and Guidelines: Seakeeping Experiments

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ITTC - Recommended Procedures and Guidelines: Seakeeping Experiments

Uploaded by

Copyright:

Available Formats

ITTC – Recommended 7.

Effective Date Revision

1. PURPOSE OF PROCEDURE.............. 2 4. VALIDATION ....................................... 9

4.1 Uncertainty Analysis ......................... 9

Updated / Edited by Approved

26th ITTC Seakeeping Committee 26th ITTC

Date 04/2011 Date 09/2011

Effective Date Revision

1. PURPOSE OF PROCEDURE 0.6

state-of-the-art practice of model seakeeping 0.4

experiments for the evaluation of ship hull 0.3

performance in predefined operational and

The procedure describes requirements 0

Effective Date Revision

unified-slender ship theory were made by 9

are expected for a prolate spheroid of beam - 4 Fig. 1

length ratio 1/8. With K = ω2/g. The dotted

lines in Figure 2 show the results of Figure 1. 1

that the finite depth must be taken into

All appendages should be fitted, and the

2.3 Model Weight Distribution

If bending moments, shears, and torsion

If the longitudinal radii of gyration for

Effective Date Revision

Testing with a free running self-propelled 2.8 Measurement of Impact Loads

The autopilot parameters should be chosen

Effective Date Revision

2.9 Parameters to be Measured - Propeller revolutions. Whenever a self-

- Rudder angle. In cases where active 2.11 Regular Waves

Effective Date Revision

Effective Date Revision

Energy spectra of waves and relevant Dimensional presentations can sometimes

Effective Date Revision

Tabular presentation of results is involving wave characteristic period and

1978 Recommendation for open ocean

3.1 Parameters to be Considered (1)

3.2 Recommendations of ITTC for

1975 Performance in irregular waves

Effective Date Revision

4. VALIDATION 4-1) Comparative Tests of a Series 60 Ship

The detailed procedure of uncertainty 4-2) Experiments on Heaving and Pitching

Effective Date Revision

Effective Date Revision

A.1.1 Type A uncertainty quoted values of uncertainty, assumed

A.1.3 Standard uncertainty

The standard uncertainty, u(y), in a

Effective Date Revision

Effective Date Revision

All of these elemental type B uncertainties Before used in experiments, all

Effective Date Revision

α angle of misalignment in In the case of measuring instruments that

1 TotalVoltageRange Data Acquisition System.

Effective Date Revision

Data processing. n / 5000πD fπD

Model speed and heading. - Model speed 1

type B uncertainties. The type A component u C (v) =   u (f) +   u (D)

Effective Date Revision

Effective Date Revision