115
1 INTRODUCTION
Duetothegrowthoftheworldshippingfleetandthe
increased spatial occupation of sea areas, maritime
traffic becomes denser in many areas. Denser traffic
could mean more complex traffic situations and
vessels traveling closer to each other. One way to
analyse the traffic and the consequences of denser
t
raffic are through collision investigations, or by
analyseofalltrafficsituationstoidentifyincidentsor
potentialaccidents,notleadingtoanaccident.
Today,itispossibletoanalysevessels’actualtrack
with information from the Automatic Information
System (AIS), used on board almost all merchant
vesselandmanyleisurecrafts.MessagesfromtheAIS
systemaresentoutwith shortti
meperiodsof 2‐360
seconds, depending on the vessel’s status (moored,
anchored,underwayetc.),speedandcoursechange.
The accuracy and integrity of positions from AIS is
discussed(Felski&Jaskólski,2013),butusedinmany
studies as the best possible way of mea
suring the
trackforvesselswithlowcost.
The aim of this study has been to identify the
traffic situations that occurred in Bornholmsgat, a
passageintheBalticSeasoutheastofSweden,during
24 hours and how they relate to the rules of the
Convention on the International Regulat
ions for
Preventing Collisions at Sea (COLREGs) to get a
pictureof theactual maritime trafficsituation in the
area. The result could be used as it is or as a
benchmarkfor measurementsin otherareas, butthe
primary purpose for this study is to be used as
calibrat
ionforadevelopedsoftwaretoautomatically
find and measure maritime traffic situations for
furtherresearch.
Maritime Traffic Situations in Bornholmsgat
F.Olindersson,C.E.Janson&J.Dahlman
DepartmentofShippingandMarineTechnology,ChalmersUniversityofTechnology,Sweden
ABSTRACT: Maritime traffic situations is regulated in the Convention on the International Regulations for
PreventingCollisionsatSea(COLREGs),buthowwellaretheserulesfollowedbyofficersonboardvessels?
Whentheworldshippingfleetgrowandthetrafficbecomesmoreint
ensive,theriskofcollisionincrease.
ByanalysingAISdatafromvesselsinthetrafficseparationschemeBornholmsgatduring24hoursinDecember
2013, 421 traffic situations were found where the passing distance between the vessels were less than 1.5
nauticalmiles.
The compliance with the Convention on the International Regulat
ions for Preventing Collisions at Sea
(COLREGs)seemstobegood,buttheaverageavoidingactionislessthantherecommendedmanoeuver.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 9
Number 1
March 2015
DOI:10.12716/1001.09.01.14
116
2 BACKGROUND
Themaintasksforanofficerofthewatch(OOW)on
boardamerchantvesselaretoavoidgroundingand
avoidcollisionwhenconningthevesselbetweentwo
locations. When the spatial occupation of sea areas
increase, i.e. by new wind mill parks, the space for
navigation of
ships decrease, resulting in denser
trafficandincreasedriskofcollision.
The marine traffic to and from the Baltic Sea is
increasing and the Bornholmsgat is one of the areas
withmosttraffic(Gucmaetal.,2007).In2006anew
trafficseparationscheme(TSS)inBornholmsgatcame
intoforce
(IMO,2006). A trafficanalyse for the area
hasbeenperformedbyGucma andPuszcz (2010)to
find the distribution of vessels position within the
traffic lanes. According to Hajduk and Montewka
(2008), there are still dangerous traffic situations in
the area, due to the fact that traffic to and from
the
PolishcoastcrossesthetrafficflowoutfromtheTSS
innorth‐easterlydirection.
The behaviour of maritime officers in traffic
situations have been studied in different situations.
Zhao, Price, Wilson, and Tan (1996) used cadets on
boardtogatherinformationfrom1053differenttraffic
situations.Theirresultwas
anaverageof0.2nautical
miles (M) in original distance to the closest point of
approach (CPA), an average actual passing distance
of 1 M, an average course alteration of avoiding
manoeuverof20°atanaveragetimetoCPA(TCPA)
of 10 minutes. Kobayashi (2006) put together data
from simulator
exercises from different training
centres around the world, and found that the
mariner’s behaviour is a function of the mariner’s
competence,suchaslicence,experience, physicaland
mental condition, together with the navigational
condition, such as own ship characteristics, water
area, maritime traffic, sea state, weather and rule of
road.
They also found that when the crossing angle
decreased (the target vessel appear more from dead
ahead),theactiontoavoidcollisionstartsearlier.
A study, measuring the rule following behaviour
in the Dover Straits, has been performed by Belcher
(2007).Inthe DoverStrait,mean passingdistance in
crossing situations
was 0.58 M and 0.44 M in
overtaking situations, when there were risk of
collision. Situations, with risk of collision, were
identified as situations where the passing distance
werewithin0.8M,resultingin175situations duringa
24hoursperiod.
The Convention on the International Regulations
forPreventing Collisions
atSea (COLREGs),1972 as
amended, regulates the behaviour of vessels in
marinetrafficsituations.TheCOLREGsstatesthatthe
risk of collision should be carefully assessed at an
earlystate,andifthereisariskofcollisionnormally
one vessel is the give‐way vessel and the other is
stand‐on vessel (except in head‐on situations and in
restricted visibility when both vessels are give‐way
vessels). This means that the officer of the stand‐on
vesselneedtotrusttheofficerofthegive‐wayvessel
andwaitforanavoidingmanoeuver.TheCOLREGs
do not provide
detailed information on when and
how an avoiding manoeuver should be performed,
leading to uncertainties sometimes (Belcher, 2007;
Zhao,Price,Wilson,&Tan,1995).
Habberley and Taylor (1989) found that officers
initiated the avoiding manoeuver by measuring the
distance to the target ship or the time to collision
(TCPA). The individual
behaviour was consistent in
different situations, but varied between different
persons.
Chauvin and Lardjane (2008) analysed 62 traffic
situationintheDoverStraitswithferriescrossingthe
trafficseparation scheme betweenDover andCalais.
Whenthecargovessel followingthetrafficlanewas
thegive‐wayvessel,theprobabilitythat
thevesseldid
an avoiding manoeuver increased with the speed of
the cargo vessel, a slow cargo vessel only changed
coursein19%ofthesituations,butafastcargovessel
changed course in 91% of the situations. When the
ferry was the give‐way vessel, there was almost
alwaysan
avoidingmanoeuvre.Themeanamplitude
of this avoiding manoeuver was 18° at an average
distanceof3.5Mtocrossasternofthecargovesselat
adistanceof0.7Moraheadofthecargovesselata
distanceof1M.
3 THEORY
TheConvention on the International
Regulationsfor
PreventingCollisionsatSea(COLREGs),regulatesthe
behaviour of vessels in marine traffic situations
throughrule1to19.Situationscouldbecategorized
ashead‐onsituation,crossingsituationorovertaking
situationdepending on the relative bea ring between
vessels (figure 1). Rule 13 regulates overtaking
situationsandstatesthat
theovertakingvesselalways
isgive‐way vessel.An overtaking situations occur if
theovertakingvesselapproachtheothervesselfrom
a direction more than 22.5° abaft the beam. Rule 14
regulates head‐on situations between power‐driven
vesselsandstatesthatbothvesselsshouldaltertheir
coursetostarboard
iftheyaremeetingonreciprocal
or nearly reciprocal courses. Rule 15 regulates
crossing situations and states that “the vessel which
hastheotheronherownstarboardsideshallkeepout
ofthewayandshall,ifthecircumstancesofthecase
admit,avoidcrossingaheadoftheothervessel”.
Figure1. Categorization of situations depending on the
relativebearing
Theexactdistanceortimeforanavoidingactionis
notstated.Rule8statesthatanavoidingmanoeuver
should be “positive, obvious and made in ample
time”, without given any specific time limits or
distances. One common interpretation is that an
Head‐on
situation
Crossing
situation
Overtakingsituation
Crossing
situation
117
avoidingmanoeuvershouldbeatleast30°(Cockcroft
&Lameijer,2012).
Rule 10 regulates the traffic flow in a traffic
separation scheme (TSS), for example the TSS
Bornholmsgat.Thebasicrulesarethatvesselsshould
followthetrafficlanes,preferablyentertheTSSatthe
end points and cross the
lane perpendicular (if
necessary). In case of a traffic situation with risk of
collision, the steering rules mentioned above should
befollowed.ThemainpurposeofaTSSistosplitthe
trafficin different lanes to avoidhead‐on situations.
As a consequence, the number of overtaking
situationsincreases.
Figure2. Definition of closest point of approach (CPA)
usingrelativemotionofvessels
Inthe dailywork forthe officer of thewatch the
risk of collision is measured as the distance of the
closestpointofapproach(DCPA)andthetimetothe
closestpointofapproach(TCPA),usingradar‐and/or
AIS‐information(figure2).
4 METHOD
Inthisstudy,AISinformationfrom
December1,2013
has been used. The AIS information is part of the
information in the HELCOM database of AIS
messages,andextractedforthisstudybytheSwedish
Maritime Administration (SMA). The area used is
latitudeN54°29’toN55°51’andlongitudeE013°35’to
E015°46’,andduringthestudied24
hourstherewere
totally 586,518 AIS messages sent from 309 different
vessels. The messages consist of the vessel’s MMSI
number, ship name, ship type, current position,
course over ground, speed over ground, heading,
navigationalstatus,destinationetc.
Figure3.OverviewoftheTSSBornholmsgat
IntheareabetweentheSwedishmainlandandthe
Danishisland Bornholm aTSS has been established.
TheTSS includessixtrafficlanes,one precautionary
area and two inshore traffic zones (figure 3). The
trafficlanesareapproximately2.7Mwide,separated
bya0.8Mwideseparationzone.(IMO,2006)
To focus on the situations in or close to the TSS
Bornholmsgat, situations south of latitude N54°45’
hasbeenremovedfromtheanalyse.
Theoveralltrafficsituationhasbeenidentifiedand
analysed for the area every third minute to identify
threedifferenttypesofsituations(seetable1),head‐
on
situations, crossing situations and overtaking
situations, fulfilling criteria for minimum passing
distance, CPA, TCPA, target ship aspect, relative
bearing and course difference measured at a range
calledactionrange.
Table1.Typesofsituationstofind
_______________________________________________
123
TypeofsituationHeadonCrossingOvertaking
_______________________________________________
CriteriaMinimumdistance <1.5M <1.5M <1.5M
DCPA<2M <3M <2M
TCPA<1hour <1hour <2hours
Targetshipaspect 0‐10°or 10‐112,5° >112,5°
Relativebearing <10° 10‐112,5° <90°
Coursedifference >160°‐‐
Assessmentrange12M 12M 12
M
Actionrange7M 5M 3M
_______________________________________________
The identification and measurement of traffic
situationsweredoneinfoursteps.
Step1:Identificationofpossiblesituations
Fromtheanalysestarttime,thedistancebetween
all combinations of vessels are calculated every
thirdminute.Ifthedistancebetweenthevesselsis
lower than a specified assessment range (in this
study 12 M), the situation is stored as a possible
situationfor
furtheranalyse.The situationis also
categorizedasheadonsituation,crossingsituation
or overtaking situation depending on target ship
aspect,relativebearingandcoursedifference.
Step2:Checkbasiccriteria
When all possible situations are found, each
situation is analysed in detail, checking if the
situation fulfils the criteria of minimum passing
distanceand, atactionrange,thecriteriaofDCPA
andTCPA.
Step 3: Find all manoeuvers (change of course
and/orspeed)
Thethirdstepistofindchangesincourseand/or
speed for the involved vessels. A manoeuver is
identifiedwhenaconstantcourse/speedhasbeen
hold for at least five minutes, then a change of
speedof minimumtwo knots
ora course change
of minimum three degrees, following of a new
period of at least five minutes with constant
course/speed. The type of manoeuver (starboard
turn, port turn or speed change), order of
manoeuver (change in course or speed) and the
situationmeasures(range,CPAandTCPA)atthe
time
ofmanoeuverwerestored.
Step4:Identifyavoidingmanoeuver
The fourth step is to identify which of the
manoeuversthataretheavoidingmanoeuver.This
is first done automatically by selecting the first
manoeuver occurred when the distance between
C
118
thevesselsarewithinactionrange.Therealworld
is not so simple, some avoiding manoeuvers are
performed in very good time and some changes
areverysmall,soallsituationshavebeenchecked
andupdatedmanually.
Data for all identified traffic situations were
exportedtoMicrosoftExcelandanalysed
statistically
withtheAdd‐in“AnalysisToolPak”.
5 RESULT
In the studied area, there were 421 traffic situations
fulfillingthecriteriaatDecember1,2013.48ofthese
were head‐on situations, 84 were crossing situations
and289 were overtaking situations. There were also
48 situations discarded due to lack
of information
(situationoccurredveryclosetothearealimitorvery
early in the time period) or by other reason (i.e.
situation occurred in a port area or pilot boat
approaching a larger vessel). The average passing
distanceinhead‐onsituationswas1.01M,incrossing
situations1.03M,
andinovertakingsituations0.79M.
5.1 Head‐onsituations
Itwasonlyinoneofthe48head‐onsituationswere
both vessels did an avoiding manoeuver. In 12
situationsoneofthevesselschangedcourseandin35
situationstherewerenoavoidingmanoeuver.
The situation when both
vessels did manoeuvers
werebetweentwotankerswithaDCPAof0.32M.at
adistance of7 M(TCPA19 minutes). Bothchanged
theircoursestostarboard,onewith4°atadistanceof
4.8Mandtheotherwith8°atadistanceof4.5M.The
passingdistance
was0.82M.
In the 12 situations were one of the vessels
changed course, the average passing distance was
0.90M.Threevesselschangedtheircoursetoportand
nine vessels changed their course to starboard. The
averageorderofmanoeuverwas8°performedatan
averagedistanceof7.2
M.Thedistributionofpassing
distancescouldbefoundinfigure4.
In the remaining 35 situations were no avoiding
manoeuver identified, the average passing distance
was1.05M,withaminimumpassingdistanceof0.54
M. The distribution of passing distances could be
foundinfigure5.
Figure4. Passage distance between vessels in head‐on
situationswhenavoidingmaneuverhasbeenperformed
Figure5. Passage distances between vessels in head‐on
situationswhennoavoidingmanoeuvercouldbeidentified
5.2 Crossingsituations
In three of the 84 crossing situations, both vessels
performed an avoiding manoeuver. In 31 situations
oneofthevesselschangedcourseandin50situations
therewerenoavoidingmanoeuver.
Thesituationswhenbothvesselsdidmanoeuvers
were in two cases between two tankers and in the
third case between two cargo vessels. In the cases
between tankers, both vessels turned to starboard
withanaveragechangeofcourseof5°atanaverage
distanceof5.8M.Inthethirdcaseoneofthevessels
turned to port 49° at distance 2.1 M entering the
northeast‐
boundtrafficlane,butnotcrossingaheadof
theothervesselwhichchangedcourse5°tostarboard
atadistanceof3.1Mtogivemorespace.Theaverage
passingdistancewas0.74M.
In the 31 situations were one of the vessels
changed course, the average actual passing distance
was
1.02 M. Eleven vessels changed their course to
portand20vesselschangedtheircoursetostarboard.
Theaverageorderofmanoeuver was16°performed
at an average distance of 5.7 M. The distribution of
passingdistancescouldbefoundinfigure6.
In the remaining 50 situations were no
avoiding
manoeuver identified, the average passing distance
was1.06M,withaminimumpassingdistanceof0.43
M. The distribution of passing distances could be
foundinfigure7.
119
Figure6. Passage distance between vessels in crossing
situationswhenavoidingmaneuverhasbeenperformed
Figure7. Passage distances between vessels in crossing
situationswhennoavoidingmanoeuvercouldbeidentified
5.3 Overtakingsituations
Therewere98overtakingsituationswithanavoiding
manoeuver and 191 situations without avoiding
manoeuver.
Inthesituationswithanavoidingmanoeuver,the
average passing distance was 0.72 M. The average
orderofmanoeuverwas7°atadistanceof3.22M.14
vessels changed course by a
turn to port (average
course change of 11°) and 84 vessels turned to
starboard (average course change of 6°). The
distribution of passing distances could be found in
figure8.
Intheremaining191 situationswereno avoiding
manoeuver identified, the average passing distance
was0.83M,withaminimum
passingdistanceof0.31
M. The distribution of passing distances could be
foundinfigure9.
Figure8. Passage distance between vessels in overtaking
situationswhenavoidingmaneuverhasbeenperformed
Figure9. Passage distances between vessels in overtaking
situationswhennoavoidingmanoeuvercouldbeidentified
6 DISCUSSION
The traffic in the area in and around the traffic
separationschemeBornholmsgatis veryintenseand
therearea lotoftrafficsituation ona singleday. In
this study a day in December 2013 were selected
randomly,primarilytominimizetheimpactofleisure
boatswithoutAIS
transponderscausingmanoeuvers
intrafficsituationswhichcouldnotbeseenbyuseof
AISdataonly.
Theaimofatrafficseparationschemeistoavoid
head‐on situations, and it is clear that there are few
head‐onsituationscomparedtoovertakingsituations
in the area. Many of
the head‐on situations are
between vessels travelling between Sweden and
PolandornorthoftheislandBornholmwellclearof
theTSS.
The passing distance between vessels is a key
parameterfor safemaritime traffic.In thisstudy the
average passing distance in head‐on situations was
1.01 M, in
crossing situations 1.03 M, and in
overtakingsituations0.79M,thiscouldbecompared
toZhaoetal.(1996)withanaveragepassingdistance
of 1 M, Belcher (2007) measured 0.58 M in crossing
situationsand0.44 Minovertakingsituationsinthe
Dover Strait. The reason Belcher (2007) measured
a
smaller value could be that he only looked at
situationswithapassingdistanceoflessthan0.8M.
120
The average course alteration of avoiding
manoeuvershasearlierbeenfoundtobe20°(Zhaoet
al., 1996) and 18° (Chauvin & Lardjane, 2008) in
crossing situations, in this study the average course
alterationwas8°inhead‐onsituations,16°incrossing
situations and 7° in overtaking situations. In
all
studiestheaveragecourseofalterationiswellbelow
therecommendedalterationofatleast30°(Cockcroft
&Lameijer,2012).
Anavoidingmanoeuvershouldbemadeinample
time.Zhaoetal.(1996)measuredanactionpointatan
average time to CPA (TCPA) of 10 minutes, and
Chauvin
and Lardjane (2008) measured an average
action point at 3.5 M. In this study the average
distance between vessels when an avoiding
manoeuver was commenced was 7.2 M in head‐on
situations,5.7Mincrossingsituationsand3.22Min
overtakingsituations.
OneuncertaintybyusingonlyAISdatais
thatthe
local visibility in the area is hard to measure. If the
vesselsareoutofsightandonlydiscoveredbyradar,
bothvesselsaregive‐wayvessels.Inthisstudy,both
vesselsperformed avoidingmanoeuvers inless than
1%ofthesituations.
7 CONCLUSIONS
The maritime traffic in
the traffic separation scheme
Bornholmsgatisveryintenseand421trafficsituations
were found during one day. The average passing
distance was measured to be 1.01 M in head‐on
situations,1.03Mincrossingsituationsand0.79Min
overtaking situations. The average course alteration
was8°inhead‐on
situations,16°incrossingsituations
and7° inovertaking situationsand performed atan
averagedistanceof7.2Minhead‐onsituations, 5.7M
in crossing situations and 3.22 M in overtaking
situations.
The compliance with the Convention on the
InternationalRegulationsforPreventingCollisionsat
Sea (COLREGs) seems
to be good, but the average
avoiding action is less than the recommended
manoeuver,inlinewithstudiesinotherareas.
REFERENCES
Belcher, P. (2007). Rule Following Behaviour in Collision
Avoidance‐a study of navigational practices in the
DoverStrait.(PhD),CardiffUniversity.
Chauvin, C., & Lardjane, S. (2008). Decision making and
strategiesinaninteractionsituation:Collisionavoidance
at sea. Transportation Research Part F: Traffic
Psychology and Behaviour, 11(4), 259‐269. doi:
http://dx.doi.org/10.1016/j.trf.2008.01.001
Cockcroft,A.N.,&Lameijer,J.N.F.(2012).AGuidetothe
Collision Avoidance Rules (7th ed.). Oxford, UK:
Butterworth‐Heinemann.
Felski, A., & Jaskólski, K. (2013). The Integrity of
Information Received by Means of AIS During Anti‐
collision Manoeuvring. TransNav, the International
Journal on Marine Navigation and
Safety of Sea
Transportation,7(1),95‐100.
Gucma,L.,Antoniewicz,M.,Artyszuk,J.,Bak,A.,Gucma,
M., Gucma, S., . . . Zalewski, P. (2007). Baltic Master
Report‐General Assumptions of the Ship Safety on
SouthernandWesternBalticSea:MaritimeUniversityof
Szczecin.
Gucma, L., & Puszcz, A. (2010).
Ships’ traffic analysis in
north of Bornholm Traffic Separation Scheme (TSS
Bornholmsgat)bystatisticalmethods.ScientificJournals
MaritimeUniversityofSzczecin,21,46‐51.
Habberley,J.S.,&Taylor,D.H.(1989).SimulatedCollision
Avoidance Manoeuvres: a Parametric Study. THE
JOURNALOFNAVIGATION,42(02),248‐254.
Hajduk,J.,&Montewka,J.
(2008).SafetyofVessels’Traffic,
Bound for Polish Ports. TransNav, the International
Journal on Marine Navigation and Safety of Sea
Transportation,2(2),131‐134.
IMO.(2006).ResolutionA.977(24)‐ShipsʹRouteing.
Kobayashi, H. (2006). Standard Marinerʹs Behavior in
Avoiding Collision Maneuver. Paper presented at the
InternationalConferenceonMarineSimulation
andShip
Manoeuvrability (MARSIM)ʹ06, Terschelling, The
Netherlands.
Zhao,J.,Price,W.G.,Wilson,P.A.,&Tan,M.(1995).The
uncertaintyand uncoordination of marinersʹ behaviour
in collision avoidance at sea. Journal of Navigation,
48(3),425‐435.
Zhao, J., Price, W. G., Wilson, P. A., & Tan, M. (1996). A
statistical study of marinersʹ behaviour in collision
avoidance at sea. Paper presented at the International
Conference on Marine Simulation and Ship
Manoeuvrability(MARSIM)ʹ96Copenhagen,Denmark.