135
1 INTRODUCTION
Following the revision of the Pilotage Act in April
2007inJapan,trainingofnewpilotshasstartedatthe
Tokyo University of Marine Science and Technology
andothereducationalinstitutions.Atthecoreofthis
training is the ship maneuvering simulator (SMS), a
useful training tool capable of simulating ba
sic ship
maneuvers and special techniques for new trainees.
There are basically two maneuvering methods used
for ship docking. The first method is to enter the
docking area from outside the port and berthing the
vesselatthetargetquaydirectly.Thesecondmethod
involves carrying out a turn in front of the ta
rget
quay, followedbyberthing.Inaprevious study, the
authors suggested an evaluation index that could be
used to determine the efficacy of the training
techniques for the first docking method (Inoue 2010,
2011). In this paper, the authors propose an
evaluationmethodforthecaseofcarryingoutturnin
frontoftheplanningpositionanddocking.
Turning the ship is t
ypically done in front of the
berthasaresultoftheplannedmaneuveringandcan
be done by the ship itself, in cooperation with
tugboats or by use of anchors. For the present
evaluation, the aut
hors prepared a docking scenario
involvingalargepurecarcarrier(PCC).Thedifficulty
in allowing for wind arises from the variable effect
that wind can have on the high‐side ships such as
largePCC’s,roll‐on/roll‐off (“Ro‐Ro”)cargoshipsand
containerships.
Theexperiencedpilotrecognizesthemeasurement
informat
ion of many inputs, including ship speed,
course deviation, and wind velocity, and steers the
vessel while operating thrust controls such as the
main engine and the rudder and monitoring the
assistance of tugboats, all at the same time. This is
possiblebecause theskilled pilothaslearnedhowto
Fundamental Study of Evaluation at Berthing
Training for Pilot Trainees Using a Ship Maneuvering
Simulator
K.Inoue&T.Okazaki
TokyoUniversityofMarineScienceandTechnologyFacultyofMarineTechnology,Tokyo,Japan
K.Murai&Y.Hayashi
KobeUniversity,Hyogo,Japan
ABSTRACT: Use of the ship maneuvering simulator (SMS) is at the core of pilot trainees education and
training, so it is desirable to have an evaluation method that can be completed shortly after each berthing
training session. There are basically two methods of docking maneuvering that pilot trainees learn: one in
whichtheshipentersfromoutsidetheportandisbertheddirectlyattheta
rgetquay,andasecondmethodin
which the vessel carries out a turn in front of the target quay before berthing. The authors suggested an
evaluation indexinapreviousstudy concerning the first docking method. In the present study, the aut
hors
proposeanevaluationmethodforthecaseofberthingthevesselusingtheturningmaneuver.Sincetheindex
obtainedbythismethodoffersasinglenumericalbenchmark,itisaneasy–to‐understandresultofthetraining
exercise.TheauthorscarriedoutexperimentsusingaSMSandconfirmedtha
ttheproposedevaluationmethod
iseffectiveandhelpfultoimprovetheeffectivenessofSMStraining.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 7
Number 1
March 2013
DOI:10.12716/1001.07.01.18
136
judge this large volume of information immediately,
recognizing it from experience, and uses all
informationtoproperlycontroltheshipmaneuvers.
When examining the process of training pilots
using SMS, the authors could not determine how
trainees learn to sense overall patterns that were
“normal”or“abnormal”aspartof
thestandardSMS
training. Therefore the authors examined the SMS
training process using the concept of Mahalanobis‐
TaguchiSystem(MTS)method.(TatebayashiK.2004)
(Elizabeth 2007). This approach utilized the data
generated during the training to establish a quality
index which may be used to determine training
efficacy.
Sincethe
indexobtainedbythis methodprovides
a simple numerical benchmark, it offers an easy‐to‐
understandmeasureoftrainingresults.Inthispaper
the authors carry out experiments using a SMS
confirming that the proposed evaluation method is
effectiveandhelpfultoimprovementofSMStraining.
2 TRAININIGSCENARIOS
2.1 Port
Information
TheportofChibaislargestseaportinJapan,located
in the interior of Tokyo bay. The port handles
166,964,000 tons of cargo annually from a total of
65,200 vessels. The training scenario involved the
berthinga largePCC atthis crowdedportefficiently
andsafely.Theberthingobjective
wasChibaFQuay;
withadistancetotheoppositequaysof720m.
Figure1.Aerial‐viewofturningoperation.
2.2 LargePCC
ManeuveringthelargePCCwasverydifficultforthe
pilot trainees. The large PCC was a high‐side ship
withhighfreeboard.Whenberthingsuchavessel,the
pilots pay extra attention to wind conditions and
hydrodynamiceffects.
Figure2.Side‐viewofturningoperation.(distanceofberth
totheshipis50m)
2.3 Passageplan
Before entering port, a ship’s master normally
prepares a passage plan. It is the pilot’s task to
carefully steer the ship from the pilot station to the
target berth, taking into account information on the
localenvironment, localguidelines andrules,andso
on.Thepilotshouldbe
briefedontheship’spassage
plan, and should make any necessary corrections.
After taking port information into account and
comparing the pilot’s suggested pla n with that
initially developed on board, the pilot and master
shouldagreeanoverallfinalplanearlyinthepassage
before the ship is committed. All
parties should be
awarethat elementsofthe planmaychange. During
the berthing approach, the pilot follows the plan
faithfully and performs the steerage. Thus during
training, pilot trainees should imitate all such
maneuveringandsteerageusingtheSMS.
Duringthesimulation,pilottraineessteer theship
parallelofthe
berth,performa200m,turnoftheship
infrontofFquayandcompletetheshipberthing.The
trainer uses the main engine, the rudder, two
tugboats,andabowthrusterforsteerage.(Ref.Figure
1,2and3)
Figure3.Typicalpassageplan.(berthing)
137
3 EVALUATIONMETHOD
3.1 Patternrealization
This scenario is intended to train a pilot’s judgment
based on information from many inputs. However,
whentraineesmadeamistaketheinstructionsoffered
fortherequiredcorrectionswerevague.
Theexperiencedpilotmonitorsalargeamountof
informationontheship’spositionandspeed,aswell
as the effect
s of wind and tugboats thrust, while
maneuvering the vessel into berthing position. The
pilot is able to control the ship by recognizing this
enormous amount of information as a pattern. The
experienced pilot has learned methods of pattern
recognition and processing from his experience. In
thi
ssenseaharborpilotissimilartoanaircraftpilot.
The air pilot judges the state of the aircraft through
constant monitoring of dozens of instruments,
operates the plane’s controls, and confirms that the
craft is responding to his instructions. An aircraft
pilot does not need to take the ti
me to confirm
individual information inputs; he instead steers by
patternrecognition.
3.2 MTSmethod
TheMTSmethodisatechniquefromthefieldof the
quality engineering. It defines a person’s action as a
state and sets a normal area (NA) based on the
expected normal range of that action. The index
describing the standard state is in the normal area,
whileanindexva
lueoutsidethenormalareaisinan
abnormal state. The degree to which the state is
abnormal can be expressed as the distance from
standardspace.Withthisinmind,theauthorsdefine
thisdistanceastheDiscrete Distance(DD),asshown
inFigure4.
The first step in performing MTS analysis is to
define a ʺnormalʺ group as the normal area. The
normal group is selected with discretion to define a
referencepointonameasurementscale.Definingthe
normal group is a crit
ical step in thi
s method, since
the NA is the reference point and basis of the
measurement scale. An abnormal state or condition
will lie outside of the normal group. The degree of
abnormality is measured in reference to the normal
group.
Athreshold valueofthe indexcanbe established
to split the group of all states int
o two groups, with
thosevaluesinsidethethresholdgroupbeingnormal
andthoseoutsidedefinedasabnormal.Thethreshold
maybesetempirically,basedonstatisticalanalysisof
pastdata, and/ormaybe decidedbasedon technical
judgment.
Figure4.Patternrealization.
3.3 Turningexperiment
Asaspecificcaseforstudy,theauthorsexaminedthe
simulation training exercise of turning a vessel 180
degrees in place using one tugboat. The maneuver
was examined to establish the threshold of the SMS
normal area. The ship’s depth and position were
measured during the ship’s movement, and the
position data are shown in Figure 6. The aut
hors
carriedoutthisexperimentatsimulatedinitialspeeds
from zero to 3 knots; higher speeds were not
investigated because of the difficulty in applying
tugboatthrustatthesespeeds.
Theauthorsselectedthelateraldistanceofthebow
movement when the ship turned at init
ial speed 2
knots to be the threshold value of the standard
turning(Ref.Table2).
Table1.Shipinformationfortheexperiment.
_______________________________________________
Shiptype PCC(PureCarCarrier)
InternationalGrosston58,000 ton
DeadWeight 22,000 ton
Displacement 30,030 ton
LOA 199.93 m
Breadth 32.26 m
DraftFore 8.50m
DraftAft 8.50m
BowThrusterone 17.8ton
Depth 12.0m
H/d1.4
_______________________________________________
Table2.Resultsofturningexperiment.
_______________________________________________
Init.Speed(kts) X(m) Y(m) Bow(Xm)
_______________________________________________
0.0 255.30 278.30 ‐43.80
1.0 128.20 344.20 ‐75.00
2.0 ‐9.70 377.70 ‐112.50
3.0‐138.30 386.90 ‐190.60
_______________________________________________
138
Figure5.Diagramoftheturningexperiment.
Figure6.Positionsduringturnwithinitialspeeds0ktsand
2kts.
3.4 Indexstandardbythisscenario
Thepurposeofthistrainingscenarioistoacquirethe
technique of turning the vessel, controlling its speed
andpositionasitturnsanti‐clockwiseandisdocked
alongside the target quay. Therefore the authors
measured the position (the position that had a bow
for90degreesturned)
,andcalculatedthestraight‐line
distancebetweenthebowandthetargetposition.
ThisdistancewasdefinedastheDiscreteDistance
(DD).
The threshold value of this distance, which
definedthenormalareaofthisparameter,assumedit
tobe112.5m.Inthispaperwetransformalllengthsto
thedimensionless unitLbynormalizingthemto the
ship’s lengt
h overall (LOA), making this value
0.56L.(Basedonanexperimenttomentionat3.3)
3.5 VerificationofThreshold
The selected threshold splits the group of distance
values into two parts, comprising those inside the
threshold group (“normal”) and outside
(“ab
normal”). The authors considered whether this
threshold was proper using the results of simulator
trainingwiththesamescenariofrom2009and2008.
Thesedatawere basedonfour demonstrationsof
modelsteeragebytheinstructor, and13attemptsby
traineeswhocarriedoutthisscenariousingtheSMS.
Table 3 gives the results from carrying out the
scenariousingSMS.For eachSMSrun,thepositionof
the simulated ship’s bow was determined in
longit
ude(X) and latitude (Y) coordinates, and the
bow’sdistancefromthetargetpositionwascalculated
alongeach coordinate axis. Thesedistances,given in
dimensionless units L, area shown in Table 3 and
pl
otted in Figure 7. The points labeled “Ins” are the
instructor’smodelsteerage.
Table3.DeviationsofbowpositionsandDiscreteDistances
fromTargetpositionfor17SMSsimulationsruns.
ValuesofDiscreteDistanceinthenormalareaareshownin
reversedshading(threshold0.56L).
_______________________________________________
Class No. Long.X(L) Lat.Y(L)DiscreteDistance(L)
_______________________________________________
Ins1‐0.30 0.05 0.30
Ins2 0.05‐0.01 0.05
Ins3 0.060.05 0.08
Ins4 0.030.02 0.03
08 5‐0.51 ‐0.11 0.52
08 6‐0.70 ‐0.13 0.71
08 7‐1.20 ‐0.18 1.21
09 8‐0.80 ‐0.36 0.88
09 9‐0.79 ‐0.19 0.81
09 10‐0.60 ‐0.02 0.60
09 11‐0.43 0.01 0.43
09 12‐0.54 0.15 0.56
09 13‐0.36 ‐0.04 0.36
09 14‐0.39 ‐0.23 0.45
09 15‐0.36 ‐0.25 0.44
09 16‐0.13 ‐0.13 0.18
09 17 0.14‐0.19 0.24
_______________________________________________
Ascanbeseenintheseresults,thepositionofthe
turning ship’s bow was widely dispersed. This
suggeststhatthecombinationofcontrollingthemain
engineandship’srudderwhileundertugboatsassist
aredifficultforpilottraineestomaster.However,five
trainees achieved DD values within the 0.56L
threshol
d value. Since five of 13 values were within
thenormalarea,aDDof0.56Lappearstobevalidas
athresholdvalue.
139
Figure 7. Scatter plot of the X‐ and Y‐deviations of bow
position from the target position for 17 SMS simulation
runs.
4 EVALUATIONS
Figure 8 presents the results of simulatortraining of
51traineesandtheirinstructors.Thedataareaplotted
in the same fashion as Figure 7. The full data set is
showninTable4,whichincludesvaluesforthegroup
mean and standard deviation for the X and Y
deviations and the Discrete Distance. Among 51
simulator t
rials, 19(37.3%) produced final values of
DDwhichwereinthenormalarea.
ThedatainTable8indicatethattheSMSberthing
simulation is very challenging, even for those who
areaexperiencedmastersofmerchantships.Defining
the DD as the quality index for these training
exercisesexplainsthi
ssituationconcretely.Thisindex
may be effective not only for assisting the instructor
to provide concrete training assistance, but also in
supportingpeerassessment.
Figure8.Resultsofevaluation.
Table4.DeviationsofbowpositionsandDiscreteDistances
fromtargetpositionfor51SMSsim
ulationrunsperformed
toevaluateDDasthequalityindex
_______________________________________________
No. Wind X(L) Y(L) DiscreteDistance
(L)
_______________________________________________
1 ‐0.30 0.05 0.30
2 0.05‐0.01 0.05
3 0.06 0.05 0.08
4North 0.03 0.02 0.03
5 ‐0.52 0.250.58
6 ‐0.52 0.540.75
7 ‐0.93 0.711.17
8 ‐0.35 0.760.83
9 ‐0.39 0.39 0.55
10South‐0.51‐0.11 0.52
11South‐0.70 ‐0.13 0.71
12South‐1.20 ‐0.18 1.21
13South‐0.80 ‐0.36 0.88
14 ‐0.79 ‐0.19 0.81
15 ‐0.60 ‐0.02 0.60
16North ‐0.43 0.01 0.43
17North ‐0.54 0.15 0.56
18North ‐0.36 ‐0.04 0.36
19South‐0.39 ‐0.23 0.45
20 ‐0.36 ‐0.25 0.44
21 ‐0.13 ‐0.13 0.18
22North 0.14‐0.19 0.24
23 ‐0.12 0.630.64
24 ‐0.40 0.450.61
25 ‐0.03 0.34 0.34
26 ‐0.52 0.600.79
27 ‐1.26 0.431.33
28 ‐0.17 0.22 0.28
29 ‐0.37 0.540.66
30 ‐0.98 0.311.03
31 ‐1.24 ‐0.24 1.26
32South‐0.03 0.45 0.46
33North ‐0.06 0.25 0.26
34North 0.40 0.37 0.55
35North ‐0.12 0.11 0.16
36North ‐0.80 0.340.87
37North ‐0.80 ‐0.08 0.81
38 ‐0.63 0.280.69
39 ‐1.11 0.331.16
40North ‐0.88 0.120.89
41South‐1.20 0.191.22
42South‐0.49 0.610.78
43North ‐0.56 0.240.60
44 ‐0.90 0.330.96
45North
‐1.02 0.011.02
46 ‐0.53 0.680.86
47North ‐0.88 0.120.89
48 ‐1.64 0.441.70
49 ‐0.86 0.310.91
50 ‐0.73 0.791.08
51 ‐0.14 0.610.62
_______________________________________________
Average‐0.54 0.210.69
StandardDeviation 0.42 0.300.36
_______________________________________________
5 STATISTICSEVALUATION
5.1 Lateraldistanceandapproachspeed
Figure9and 10 show the lateraldistanceandvessel
speedsateach of threeevaluationspots, located. 1L,
2Land3L fromthetargetquay.Themean resultsof
four simulation runs by an instructor were obtained
andcomparedwiththesteerageofthetrainees.
This comparison forms an evaluation index,
followingthetechniquedescribedinpreviousstudies
140
(Inoue 2010, 2011). The authors’ statistical study
revealed that the pilot trainees 1) took ample lateral
distance in comparison with model steerage, and 2)
tendedtofavorslowspeed.
Wheninstructorsmakeascenarioforyoungpilot
trainees without the captain’s experience, they are
abletomakeuseofthese
statisticalprocessing.
Figure9. Histogramsofmodeled shipspeed (m/sec)atthe
threeevaluationpositions.
Figure10.Histogramsoflateraldistancefromtheberth
Table5.Modelapproachlateraldistancesandshipspeeds.
(Average)
_______________________________________________
Evaluationspot Lateraldistance(L) Shipspeed
(m/sec)
_______________________________________________
1.0L1.031.50
2.0L1.082.10
3.0L1.122.60
_______________________________________________
Table6.Modelapproachlateraldistancesandshipspeeds.
(Standarddeviation)
_______________________________________________
EvaluationspotLateraldistance(L)Shipspeed(m/sec)
_______________________________________________
1.0L0.070.20
2.0L0.090.30
3.0L0.090.30
_______________________________________________
5.2 Figureoftrail
The “figure of trail” is a record of the moment to
momentpositionofthevesselduringthesimulation.
Because it is output promptly after each practice
session, the figure of trail is an effective debriefing
tool,helpingtodeepenthetrainee’sunderstandingof
theinstructor’s
evaluation.
Figure11.FigureoftrailforTraineeNo.48.
Figure12.FigureoftrailforTraineeNo.47.
Figure 11 shows the trainee No.48. His DD was
1.7.Becausetheapproachspeedwasslow,controlof
the ship was somewhat diminished, resulting in the
turn being completed beyond the target turning
position. The Instructor recommended additional
trainingontheapproach.TraineeNo.47maneuvered
theshipunderstrongNorthwind.
HisDDwas0.89.
When the speed of the ship declined as it closely
approached the turni ng target, his ship had large
leeway. The instructor taught him how to use the
vertical thrust control of the ship. Because the
instructorcanunderstand theDDindeximmediately
afterpractice,hecan
usetheDDandthetrailplotto
explainanyerrorsinatrainee’sapproachtothequay.
TheDDindexwasfoundtobeeffectiveasoneofthe
parameters that can explain ship pilot’s pattern
recognitionandtheirabilitytocontroltheshipwhile
monitoringmanyinputs.
6 CONCLUSIONS
The authors propose a new method of evaluating
maneuvering training harbor maneuvering using
SMS. The method is based on defining an indicator,
the discrete distance (DD), obtained from the ship’s
motionanalysisafterSMStraining.Becauseitiseasily
measured from a figure of trail output, which is
availablepromptly
aftertheconclusionofthetraining
exercise, the DD was found to effective for trainee
evaluation.Thestudyfoundthat
141
The instructor can review any mistakes made by
thetraineeduring practiceintermsoftheconcrete
DDindex;
Other trainees easily understand DD in the
evaluation between trainees, and are able to
recognize those areas that require additional
practicebythetrainee;
The instructor can evaluate
the trainee’s
performancemorepracticallyusing theDDindex
along with the speed and latitude distance at the
approachevaluationspots;
Training data for this scenario generated by
trainees who are already experienced ship masters
canbestatisticallycompiledtoproducescenariosthat
will be effective for young trainees who
do not yet
have captain experience. These scenarios could help
newtraineeslearnthefollowing:
The importance of maintaining approach speed
andampledistancetotheberth;
ThePCCismorevulnerabletowindatslowvessel
speeds. As speed reduces, hydrodynamic forces
reduce, and the effect of wind
on heading and
leewayincreases.
REFERENCES
Inoue k. & et al. 2010. On the Evaluation Method for
BerthingTrainingusingaShipManeuveringSimulator.
JournalofJapaninstituteofnavigation.vol..24:63‐69(in
Japanese)
InoueK.&YabukiH.&OkazakiT.2011.OntheEvaluation
of Berthing Training for Pilot Trainees Using a Ship
Maneuvering
Simulator. The 12th Annual General
AssemblyofIAMU:319‐326
Kornacki J. 2011. Ship’s Turning in the Navigational
Practice. TRANSNAV International Journal on Marine
NavigationandSafetyofSeaTransportation.vol.5No.
2:261‐246
Cudney E. & Paryani K. & Ragsdell K. 2007.Applying
Mahalanobis‐TaguchiSystemtoVehicle Ride.
Jounalof
IndustrialandSystemsEngineering.vol.1,No.3:251‐259
Kazuo Tatebayashi 2004. Introduction to Taguchi Method.
PublicationNikkagiren.(inJapanese)
