International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 2
Number 4
December 2008
377
The Analysis of Possibilities How the Collision
Between m/v 'Gdynia' and m/v 'Fu Shan Hai'
Could Have Been Avoided
A. Banachowicz
Gdynia Maritime University, Gdynia, Poland
P. Wolejsza
Maritime University of Szczecin, Szczecin, Poland
ABSTRACT: The report presents the simulation results of collision between m/v “Gdynia” and m/v “Fu Shan
Hai”. The analysis was performed by means of decision support in collision situations. This system is based
on a structure of programme multiagents using AIS data (Automatic Identification System) with the
possibility of cooperation between agents or vessels. The multiagent system of supporting anticollision
decisions increases the reliability of navigational information and permits making right decisions, thereby
increasing safety at sea.
1 INTRODUCTION
The collision occurred on 31st May 2003 to the
north from Bornholm by day with visibility over 10
nm. The distance from the place of collision to the
nearest navigational danger in the form of a shoal is
equal to 3 nm. There were also a few fishing vessels
in the area, the traffic parameters of which did not
constitute immediate threat of collision with any of
the vessels. Figure 1 presents the location of the
vessels at the moment of collision.
Fig. 1. Location of vessels at the moment of collision
Source: Danish Maritime Administration – www.dma.dk
Figure 2 presents both vessels’ position from
1200 hrs up to the moment of collision reconstructed
on paper chart based on data from the report by
Danish Maritime Administration. Below the
photograph there is a table 1 that presents the traffic
parameters of these vessels.
Fig. 2. Reconstruction of vessels position on paper chart
378
Table 1. Tabular reconstruction of collision situation parame-
ters
Source: Danish Maritime Administration – www.dma.dk
2 MULTIAGENT DECISION SUPPORTING
SYSTEM IN COLLISION SITUATIONS
The structure of the system has been so prepared that
it permits a parallel performance of tasks bound with
supporting decision processes in the conduct of the
vessel. The main element of the advisory system on
every vessel is the managing module. This is a
cooperation agent with the following tasks:
1 detection of occurrences on the basis of AIS
information,
2 assigning a task to the module (modules)
responsible for a given kind of occurrence,
3 control of tasks assigned, time regimes included,
4 communication with cooperation agents on other
vessels,
5 communication with the operator.
The next system element is the information agent.
Its task is a continuous listening watch on AIS and
GPS (Global Positioning System) frequencies. After
receiving the message it transforms it into a format
readable for other agents, at the same time extracting
data and eliminating those which are unnecessary in
further process and would unduly lengthen
calculation time.
The module of occurrence identification is a
navigational agent. On the basis of data obtained
from the information agent, inter alia positions,
courses and speeds of own and foreign object, the
vessel’s navigational status, it determines the
vessels’ mutual location, encounter parameters like
CPA (Closest Point of Approach) and TCPA (Time
to Closest Point of Approach) and determines way
priority for a vessel. Information on vessels in
relation to which we have priority of way and to
which we must give way is transmitted back to the
cooperation agent.
At the same time the other navigational agent, i.e.
manoeuvring module, calculates new movement
parameters of our vessel, permitting the passing of
the set CPA. New courses and speeds are fed back to
the cooperation agent, who makes a decision
whether to keep present course and speed or to
perform an anticollision manoeuvre.
The last element of the multiagent system is the
execution module, the task of which is to control the
vessel’s propulsion and steering devices. This
module is operated by the navigator himself. In
future, these activities, too, could be performed
automatically.
The input data for the algorithm and the
programme is the following information from AIS
and GPS systems:
1 for vessel identification:
− MMSI number - Maritime Mobile Service
Identity,
2 for designating the vessel’s movement elements:
− geographic position,
− course over the ground,
− true course,
− speed over the ground,
− time of forming the information package,
3 for detecting the vessel’s manoeuvre:
− the vessel’s angular speed,
4 for determining the predicted trajectory of the
vessel’s movement:
− planning the vessel’s route with all way points
from the port of leaving to the port of
destination,
5 for accurately determining encounter parameters:
− locating the antenna in relation to the vessel’s
bow/stern,
6 for determining the degree of privilege between
vessels:
− the vessel’s navigational status.
− 0 – at berth (moored), at anchor, grounded,
− 1 – not under command,
− 2 – hampered,
− 3 – restricted by draft,
− 4 – catching fish,
− 5 – sailing boat underway,
− 6 – mechanically propelled, underway.
The output data are:
1 suggestions for course alterations which, with
maintained speed, lead to passing clear of all
objects on the set CPA,
2 suggestions for speed alterations which, with
maintained course, lead to passing clear of all
objects on the set CPA,
3 minimal-time suggestion for own vessel’s
movement parameter alteration (course or speed)
379
which leads to passing clear of all objects on the
set CPA and fulfils the task of optimisation.
3 ASSUMPTIONS FOR SIMULATION
All information used for simulating the encounter of
two vessels is derived from the report by Danish
Maritime Administration placed at web page
www.dma.dk. The simulation was implemented at
ECDIS laboratory of the Maritime University at
Szczecin on a navigation manoeuvring simulator
Navi Trainer by firm Transas Marine, designed for
carrying out training tasks resulting from the
requirements of STCW 78/95 Convention. Navi
Trainer Professional programme works in an
integrated network environment based on Windows
NT operational system. Devices for simulating radar
work, ARPA, ECDIS, gyrocompass, the log, GPS
receiver and other navigational systems and devices
meet all applied functioning standards accepted by
IMO and international conventions.
On the basis of mutual location and vessel
movement parameters at 1205 hrs, the system
qualified the encounter as intersecting courses and
pointed out “Gdynia” as the give way vessel, in
accordance with rule 15 of COLREG. For this
reason, a manoeuvring suggestion was prepared for
vessel “Gdynia”, assuming that the other vessel
maintains her course and speed (COLGREG rule
17). The procedure determining way priority was
described in detail in (Wołejsza P. 2005b).
The following parameters for performing
simulation were assumed:
1 distance of passing clear CPA = 1852 m (1 nm),
2 good visibility,
3 maximum speed – 15 knots,
4 minimum speed – 0 knots,
5 minimum course alteration – 20°,
6 maximum course alteration – 90°,
7 LOA “Gdynia” – 101 m,
8 LOA “Fu Shan Hai” – 225 m,
9 location of “Gdynia” ‘s radar antenna - 85 m from
the bow,
10 “Fu Shan Hai” ‘s superstructure situated 200 m
from the bow is the vessel’s echo visible on the
radar screen, in relation to which ARPA
calculates encounter parameters (in theory, the
superstructure of a loaded bulk carrier should give
the strongest echo).
The speed interval was determined on the basis of
“Gdynia” ‘s manoeuvring data. The service speed of
this vessel is 15 knots. The minimum speed excludes
movement of the vessel astern. The interval of
recommended course alteration was determined so
as to be clear and visible (20°) on the one hand, and
on the other hand (90°) that it should not make it
necessary to turn back from the course chosen.
On the basis of data contained in the table 1
simulation was carried out in order to determine
encounter parameters and possible collision-
prevention manoeuvres at particular moments of
time. Manoeuvring elements (kinematic equations)
were not taken account of in the solution.
4 SIMULATION RESULTS
On the basis of assumptions the system worked out
solutions of the collision situation, presented in
Table 2.
Table 2. Encounter parameters and anticollision manoeuvring
suggestions worked out by the system.
Remarks:
1 Results marked in boldface in the table for CPA
value = 0.5 nm.
ARPA
obtained
encounter
parameters
System
calculated
encounter
parameters
Solutions suggested by
decision support system
Local time
CPA
[nm]
TCPA
[min]
CPA
[nm]
TC-
PA
[min]
Course
alteration
to star-
board [°]
Course
altera-
tion to
New
speed
[knot
s]
1205
0.40
0.29
17.5
340.6
8.4
1206
0.50
15.4
0.33
15.3
349.4
7.5
1207
0.50
14.8
0.30
14.7
351.3
7.4
1208
0.50
13.9
0.32
13.6
358.3
6.9
1209
0.40
13.3
0.25
11.2
359.0
7.1
1210
0.40
12.4
0.21
10.1
008.5
6.2
1211
0.40
10.8
0.38
10.4
020.5
4.6
1212
0.70
7.2
0.59
7.2
039.7
1.1
1213
0.50
6.3
0.36
6.4
054.2
lack
1214
0.40
5.5
0.19
5.4
lack
lack
1215
0.30
4.5
0.18
4.8
030.1
3.3
1216
0.20
3.0
0.10
3.0
lack
lack
1216.5 0.20 2.2 0.06 2.3
048.7
1.2
1217
0.10
1.8
0.04
1.9
lack
lack
1217.5
0.10
1.3
0.03
1.4
lack
lack
1218
0.10
0.7
0
1.0
lack
lack
1218.5
collision
380
2 Results marked in boldface in the table and
justified were obtained for CPA value = 0.25 nm.
Comparing encounter parameters CPA and TCPA
obtained from ARPA and decision-support system it
can be stated that:
1 CPA presented by ARPA is always larger than the
CPA calculated by the system,
2 TCPA values approximate each other.
Differences in CPA values are due to the vessel’s
length being taken account of when calculated by the
system. Information on the vessel’s size and antenna
location is taken from AIS system. Both in ARPA
and in the formulae presented in (Lenart 1999), on
the other hand, the vessels are treated as points;
hence the overstatement of results which can
translate into erroneous estimation of situation by the
navigator, particularly in the encounters of large
vessels, among which the “Fu Shan Hai” was
counted.
The largest CPA difference in the table equals
0.21 nm, which with CPA value of 0.4 nm consti-
tutes an error of over 50 per cent. On smaller vessels
like “Gdynia” the smallest passing distance on the
level of 0.5 nm is often considered as safe. Such
value was obtained from ARPA by “Gdynia” ‘s
Second Officer in the third, fourth and fifth minute
of tracking and which was probably why he did not
undertake any action, considering the situation as
safe.
TCPA values obtained from both system are close
to each other, as the vessel’s size does not affect the
moment of contact, only its value.
5 ANALYSIS OF RESULTS
At 1205 hrs local time the vessels were in a distance
of 2.9 nm from each other and the CPA according to
ARPA was 0.4 nm. In result of the system’s work
the following results were obtained: in order to pass
each other at 1 nm distance, “Gdynia” ’s watch
officer could make a choice between altering course
to starboard by 61°, or to port by 27°; a speed
reduction manoeuvre was still viable at this distance,
but as it is practically rarely used in the open sea, it
will not be discussed in more detail. In the
subsequent (06-08) minutes CPA rose to 0.5 nm,
although it was actually on the level of 0.3 nm. Only
at 1209 hrs, when the CPA started decreasing, did
vessel “Gdynia” begin to alter course to starboard by
25°. According to the system’s calculations, in order
to pass “Fu Shan Hai” astern in a distance of 0.5 nm
or 1.0 nm the course should been immediately alter-
ed by respectively 44° or 80°. Thus, the action
undertaken was insufficient. At 1210 hrs vessel “Fu
Shan Hai” issued 5 short blasts; she must have not
noticed that “Gdynia” started altering course.
At 1213 hrs, when “Gdynia” had altered her course
by about 15° this fact went unnoticed on vessel
“Fu Shan Hai”, which is why the master decided to
stop engine. He did not notify other vessels about it;
the manoeuvre could be noticed neither visually nor
by radar. At 1215 hrs the vessels were at a distance
of 1.1 nm from each other. As the system did not
find a solution permitting the vessels to pass each
other in a distance of 1nm, he reduced the assumed
CPA by 50%. In this situation, altering course
immediately to starboard by 85° and to port by 25°
ensured respectively passing astern and ahead of the
vessel. Two minutes before the collision “Gdynia”
continued turning to starboard and was on a course
of 322°. “Fu Shan Hai” was decreasing her speed,
which is why, in order to pass her astern at a distance
of 0.25 nm, the course should have been altered by at
least 87°. An effective anti-collision manoeuvre by
altering course to port was sheerly theoretical, as the
rudder had been put to starboard. A minute before
the collision “Gdynia” continued altering course to
starboard (at the moment of collision she was on a
course of 350°), and “Fu Shan Hai” continued to
reduce her speed. From collision avoidance point of
view both manoeuvres were neutralizing each other
and eventuated in “Gdynia” striking the port of the
other vessel making it sink.
6 CONCLUSIONS
The case described proves that the application of
AIS for estimating the situation would have
permitted the avoidance of collision. “Fu Shan Hai”
would have noticed “Gdynia’s” altered course thanks
to the angular speed parameter, and “Gdynia” would
have noticed “Fu Shan Hai’s” speed reduction. Such
information is not provided by ARPA.
This does not change the fact that the vessel to
give way tarried with undertaking proper measures
according to the situation (non-compliance with
rules 8, 15, and 16 of COLREG). This may have
been due to erroneous estimation of the situation,
based mainly on ARPA information (breaking rules
5 and 7), which eventuated in undue nervousness of
the other party, resulting in ill-judged decisions
(action non-complying with rule 17) and leading to
collision.
Whereas AIS information would have helped to
estimate the situation properly, then the use of vessel
traffic parameters obtained from AIS for working
out the manoeuvre would have provided ready
solutions for the collision situation. Even inaccurate
ARPA data would have permitted the preparation of
effective solutions by a multiagent system of
381
decision support in collision situations. Course
alteration to starboard at 1209 hrs by at least 44°
instead of 25° had the following advantages:
1 the manoeuvre was definite and clearly visible for
the other vessel,
2 it permitted passing astern of “Fu Shan Hai” at a
distance of 0.5 nm (in the case of assumed CPA
being 1.0 nm, the course should have been altered
by 80°),
3 successive course alterations would have been
avoided (rule 8b),
4 seeing “Gdynia” ‘s definite manoeuvre, “Fu Shan
Hai” would not have started the manoeuvre of
speed reduction.
REFERENCES
Danish Maritime Administration, Casualty investigation
reports. www.dma.dk
Lenart A., 1999. Manoeuvring to required approach parameters
– CPA distance and time. Annual of Navigation No 1.
Magaj J. & Wołejsza P., 2006. Execution of anti-collision
manoeuvre. 4th International Conference ‘EXPLO-SHIP’,
Świnoujście, 18-19 May.
Wołejsza P., 2005a. A concept of navigation support system in
areas difficult to navigation. Polish Journal of Environ-
mental Studies Vol.14, Supplement I, pp. 126-129.
Wołejsza P., 2005b. Algorithm of anti-collision manoeuvre.
11th International Conference ‘Marine Traffic
Engineering’, Świnoujście 23-25 November.
Wołejsza P., 2006. Verification of the navigator’s decision
support system on an ECDIS simulator. Polish Journal of
Environmental Studies Vol.15, No 4b, pp. 196-199.