International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 6
Number 1
March 2012
11
1 INTRODUCTION
Traffic conflict refers to the event of vessel interfer-
ence, which occurs in port waters due to the special
characteristics of port traffic in limited sea space,
high traffic density, and complex operational regula-
tions. As undesirable incidents, conflicts have a di-
rect effect on the safety of navigation. A conflict
without proper resolution may lead to a collision re-
sulting in a loss of life and property, and even
threaten the ocean environment.
In recent years marine traffic has been increasing
greatly due to the sustained growth of seaborne
trade. As a result, the port traffic network becomes
finely meshed and intensively used. The demand for
the use of sea space sometimes exceeds the available
capacity, such that even a small interaction (i.e. a
conflict) between vessels may have a large impact
on the entire network. The most common product of
a conflict is time delay, which results from the eva-
sive maneuvers of vessels to avoid a collision. With-
in a saturated network, these delays can slow the
speed of traffic stream, increase vessel-waiting time
and the length of waiting queue. Traffic congestion
would arise accordingly.
The world’s busiest ports are faced with potential
risk of traffic conflicts. However, maritime control
centers often can only play an advisory role, which
cannot satisfy the demand on traffic management
arising within port waters. There is no positive con-
trol as to conflict avoidance.
If conflict risk could be predicted in advance, we
could take appropriate measures to resolve or elimi-
nate conflicts so as to avoid incidents/accidents and
reduce the impact of conflict on network efficiency.
However, to the best of this researcher’s knowledge,
no systematic method has been developed for detect-
ing marine traffic conflicts. A review of past studies
related to marine traffic safety revealed that almost
all were focused on collision avoidance. Neverthe-
less, a conflict can be considered as a collision risk
with a low degree of danger. Hence, works in colli-
sion avoidance are worth reviewing, which could
provide valuable reference to this research.
A Simulation Model for Detecting Vessel
Conflicts Within a Seaport
Q. Li & H. S. L. Fan
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore
ABSTRACT: Conflicts represent near misses between two moving vessels, and often occur in port waters due
to limited sea space, high traffic movements, and complicated traffic regulations. Conflicts frequently result
in congestion and safety concerns. If conflict risk can be predicted, one could take appropriate measures to re-
solve conflicts so as to avoid incidents/accidents and reduce potential delays. To the best of this researcher’s
knowledge, no systematic study has been carried out on the issue of detecting marine traffic conflicts. In this
paper, we present an algorithm designed to determine a conflict using the criterion of vessel domain. The al-
gorithm aims to evaluate the relative positions of vessel domains to detect potential conflicts. To implement
the algorithm, a simulation model has been developed in Visual C++. The model at present provides a single
function for conflict detection but can be expanded to a multi-functional system for resolving conflicts in fu-
ture work.
12
Two criteria are used in past studies for determin-
ing a collision risk: the closest point of approach
(CPA) and ship domain.
The criterion of CPA is applied with two parame-
ters: distance of closest point of approach (D
CPA
) and
time of closest point of approach (T
CPA
).The value of
CPA parameters indicates the relative position be-
tween two vessels. For example, a smaller CPA in-
dicates a higher risk of collision. The CPA parame-
ters are applicable in a collision avoidance system,
which can guide vessel to execute proper anti-
collision maneuvers. An example is Lenart’s studies
(Lenart 1999, Lenart 2000) on what speed and/or
course maneuver should be undertaken to achieve
the required CPA time and distance.
The criterion of CPA is difficult to use in restrict-
ed waters such as narrow fairways. In view of this,
the concept of ship domain has been proposed as a
more comprehensive and accurate criterion. It can be
explained as “a water area around a vessel which is
needed to ensure the safety of navigation and to
avoid collision” (Zhao et al. 1993). Vessel domain
was first presented by Fujii et al. (1971). Based on
field observations, Fujii’s study established a do-
main model for a narrow channel. Later, Goodwin
(1975) developed a domain model in open sea. Be-
sides presenting a model, the study also analyzed
how traffic density and length of vessel affect the
size of vessel domain.
The shape and size of a vessel domain are affect-
ed by a number of factors (vessel’s speed and length,
sea area, traffic density etc.). As different factors are
taken into account, ship domains proposed by vari-
ous studies differ from one another. Many studies
have focused on improving the vessel domain model
(Davis et al. 1980, Coldwell 1983, Zhu et al. 2001,
Pietrzykowski 2008).
In a port traffic system, vessels traveling along
fairways are required to keep various safety clear-
ances in accordance with the port’s regulation. The
domain of a vessel can thereby be referred to as the
clearance area around it. This paper would attempt
to design an algorithm to detect conflicts using the
criterion of ship domain. That is, the relative posi-
tions of the domains of two vessels will be evaluated
before they actually encounter each other. If the do-
main of a vessel will interfere with the domain of the
other, a potential conflict is indicated.
A simulation model is developed to implement
the algorithm, using Visual C++ 6.0. In the simula-
tion model, conflicts can be detected for a given de-
mand schedule of marine traffic within a seaport.
The first and most important goal of conflict detec-
tion is to enable safe navigation and avoid collision
between vessels. For system optimization, attention
should also be paid to reduce the impact of conflicts
on network efficiency so as to improve traffic condi-
tions within the seaport.
This paper is structured as follows: Section 1 in-
troduces the issues addressed; Section 2 presents an
overview of the simulation model; Section 3 de-
scribes the algorithm for conflict detection; Section
4 focuses on simulation model implementation; and
Section 5 summarizes findings and proposes future
work.
2 OVERVIEW OF SIMULATION MODEL
2.1 The seaport traffic system
A seaport traffic system can be treated as a network
of nodes and links. Within the network each link in-
dicates a section of a fairway, and a node can be a
berthing/anchorage area, a boarding point for port
pilots, an intersection area of fairways, or a separa-
tion point dividing a fairway into two sections due to
differences in widths and/or traffic regulations. The
route of a vessel can be represented by a path in the
network consisting of a series of nodes and links.
Figure 1 shows a seaport traffic system we use in
the simulation model, where black dots represent the
nodes and a rectangle between two nodes indicates a
link. The width of a rectangle indicates the width of
the link. A vessel is only allowed to travel within the
link.
Figure 1. A seaport traffic system for Singapore.
2.2 Flowchart for conflict prediction
A seaport traffic system usually involves a large
number of vessels. We need to detect a potential
conflict between any pair of vessels. For any pair of
vessels, the system will check whether they will con-
flict or not in a time interval (t
0
, t
3
).
There are two situations in conflict detection:
− Node conflict prediction: two vessels traveling
toward the same node are on different links.
− Link conflict prediction: two vessels traveling
toward the same node are on the same link.
13
In the first situation, the two vessels may have a
conflict when they are passing the node. Thus, be-
fore the two vessels reach the node, the system
needs to predict whether the two vessels will have a
conflict.
In the second situation, the two vessels may en-
counter a conflict on the link. However, if the fair-
way is sufficiently wide so that a vessel can overtake
the other safely, the conflict will not occur. Thus, the
factor of the link width should be considered into
conflict detection on a link. These are described in
the next section.
Note that, the relative position between two ves-
sels varies as vessels are moving. The conflict situa-
tion would change accordingly. Suppose that the
vessels have a risk of conflict during a certain time
period. According to the changes in vessel trajecto-
ries, this time period is divided into several time in-
tervals. The system needs to separately evaluate the
conflict situation during different time interval.
Figure 2 shows the flowchart for conflict detec-
tion (the notations t
0
, t
1
, t
2
, t
3
are defined in Section
3).
Figure 2. Flowchart for conflict prediction.
3 DETERMINE A CONFLICT BETWEEN TWO
VESSELS
3.1 Preliminaries and assumptions
Denote a vessel as V (O, d, Φ, Ψ, Φ
ˉ
, Ψ
ˉ
1
, Ψ
ˉ
2
) as
shown in Figure 3. For the purpose of simplifying
analysis, a vessel is regarded as a rectangle V, whose
dimensions are Φ (width) and Ψ (length). Suppose
O(x, y) is the center of the vessel. At present, it is
travelling along the direction d.
The clearance area of a vessel is defined as a zone
within which the vessel can keep enough distance to
avoid conflict with each other. The clearance area
varies according to differences in a vessel’s outline,
dimension, technical parameters and fairway charac-
teristics. In this research, the shape of a vessel’s
clearance area is assumed as a rectangle R. The pa-
rameter Φ
ˉ
refers to the vessel’s lateral clearance.
Vessel’s longitudinal clearance is represented by pa-
rameter Ψ
ˉ
1
in the direction of the bow and Ψ
ˉ
2
in the
direction of the stern. The values of these parameters
(Φ
ˉ
, Ψ
ˉ
1
, Ψ
ˉ
2
) are specified by regulation. These pa-
rameters can be set up in a simulation system as in-
put data.
Figure 3. A vessel and its domain.
3.2 Node conflict prediction
Two vessels, V
1
and V
2
, on different links travel to-
ward the same node. Table 1 lists the navigation in-
formation, where t
1
< t
2
, i.e. V
1
will reach the node
before V
2
.
Table 1. Two vessels on different links
______________________________________________
V
1
V
2
______________________________________________
Position A E
Velocity before the node v
1
v
2
Velocity after the node vˉ
1
vˉ
2
Time to the node t
1
t
2
Time to the next node
ˉ
t
1
ˉ
t
2
______________________________________________
Suppose t
0
= 0, t
3
= min (
ˉ
t
1
,
ˉ
t
2
). We aim to check
whether there is any conflict during the time interval
(0, t
3
). The movements of V
1
with respect to V
2
are
different in three different time intervals
− In the time interval (t
0
, t
1
), the velocity of V
1
with
respect to V
2
is w
1
= v
1
- v
2
.
− In the time interval [t
1
, t
2
], the velocity of V
1
with
respect to V
2
is w
2
= vˉ
1
- v
2
.
− In the time interval (t
2
, t
3
), the velocity of V
1
with
respect to V
2
is w
3
= vˉ
1
- vˉ
2
.
Figure 4 shows the movement of the center of V
1
with respect to the center of V
2
. With respect to V
2
,
starting at A, V
1
passes B at t
1
, moves from B to C
during [t
1
, t
2
], and reaches D at t
3
. Thus,
( )
( )
( )( )
( )
12 1 2 2
212 1 212
33 1 2 3
,
,
.
AB t t
BC tt tt
CD t t
= =-
= -=- -
= =-
w vv
w vv
w vv
14
At location A, the domain of V
1
follows its mov-
ing direction v
1
(Fig. 5(a)). Similarly, the domains of
the vessels at different locations can be obtained
(Table 2). Suppose q
5
ij
= q
1
ij
, i = 0, 1, j = 0, 1, 2, k = 1,
2, 3, 4. Table 2 tells that
− Q
ij
is a domain of the vessel V
i
at t = t
i
,
− q
k
ij
is the k-th corner of the domain Q
ij
,
− q
k
ij
q
ij
k+1
is the k-th edge of the domain Q
ij
.
The movement of the domain of V
1
with respect
to the domain of V
2
is denoted as the relative move-
ment of V
1
to V
2
. For example, referring to Figure 4,
Figure 5 shows the relative movements of V
1
to V
2
,
in the three different time intervals.
Figure 4. The movements of the center of V
1
with respect to the
center of V
2
: (a) In time interval (0, t
1
); (b) In time interval [t
1
,
t
2
]; (c) In time interval (t
2
, t
3
).
Table 2. Domain of vessels at different locations
______________________________________________
Location Domain of V
1
Domain of V
2
______________________________________________
t =t
0
Q
10
(q
1
10
,q
2
10
,q
3
10
,q
4
10
) Q
20
(q
1
20
,q
2
20
,q
3
20
,q
4
20
)
t =t
1
Q
11
(q
1
11
,q
2
11
,q
3
11
,q
4
11
) Q
21
(q
1
21
,q
2
21
,q
3
21
,q
4
21
)
t =t
2
Q
12
(q
1
12
,q
2
12
,q
3
12
,q
4
12
) Q
22
(q
1
22
,q
2
22
,q
3
22
,q
4
22
)
______________________________________________
For any j=0, 1, 2, in the time interval (t
j
, t
j+1
), the
velocity of V
1
with respect to V
2
is w
j+1
. The move-
ment of the corner q
k
1j
with respect to V
2
is a line
segment q
k
1
p
k
1j
where
p
k
1j
= q
k
1j
+ (t
j+1
- t
j
) w
j + 1.
Thus, the movement of the edge q
i
k
q
i
k+1
with re-
spect to V
2
is P
k
j
= q
k
1
q
k+1
1j
p
k+1
1j
p
k
1j
(Fig. 6). If V
1
and V
2
conflict with each other, the movement of at least
one edge of V
1
will intersect with the domain of V
2
,
i.e.
P
k
j
∩ Q
2j
≠ Ø.
Figure 6 shows an example when there is no con-
flict between V
1
and V
2
. Figure 7 is another example
when there is a conflict between V
1
and V
2
.
In summary, V
1
and V
2
will conflict in the time
interval (t
j
, t
j+1
) if and only if
∪(P
k
j
∩ Q
2j
)
≠ Ø.
In this way, the conflict detection turns to check-
ing whether two parallelograms intersect with each
other or not.
3.3 Link conflict prediction
Suppose a vessel V
1
follows another vessel V
2
along
a link (see Fig. 8(a)). Table 3 lists the navigation in-
formation of these vessels. The velocity of V
1
with
respect to V
2
is
w
1
= v
1
- v
2
.
If v
1
is not larger than v
2
, V
1
and V
2
will conflict if
and only if
12
2
LL
AE
+
<
.
Suppose t
3
= min(t
1
, t
2
). We need to check wheth-
er the two vessels will conflict with each other dur-
ing (0, t
3
). After that, the two vessels will not be
conflicting on the link, because one vessel leaves
this link. If v
1
is larger than v
2
, during (0, t
3
), the
relative movement of V
1
with respect to V
2
is shown
in Figure 8(b), where
1 1 13
, 1, 2, 3, 4.
kk
p q wt k=+× =
Figure 5. The relative movement of V
1
to V
2
: (a) In time inter-
val (0, t
1
); (b) In time interval [t
1
, t
2
]; (c) In time interval (t
2
, t
3
).
15
Figure 6. P
k
j
∩ Q
2j
≠ Ø, V
1
and V
2
will not conflict with each
other in the time interval (t
j
, t
j+1
).
Figure 7. P
2
j
∩ Q
2j
≠ Ø, P
3
j
∩ Q
2j
≠ Ø, V
1
and V
2
will conflict
with each other in the time interval (t
j
, t
j+1
).
Obviously, V
1
and V
2
will have a conflict if and
only if q
1
1
q
4
1
p
2
1
p
3
1
intersects with Q
2
. In Figure 9(a), q
1
1
q
4
1
p
2
1
p
3
1
intersects with Q
2
, thus V
1
and V
2
are in con-
flict. In Figure 9(b), q
1
1
q
4
1
p
2
1
p
3
1
does not intersect with
Q
2
, thus V
1
and V
2
will not conflict with each other.
The conflict detection method described earlier is
merely based on an assessment of relative movement
of one vessel to another vessel. The system judges if
a conflict will occur by checking whether the two
parallelograms intersect. If link width is taken into
consideration, the result may be different. Suppose
the width of the link is W, and the width of the do-
mains of the two vessels are W
1
and W
2
. The two
vessels can travel in parallel without a conflict if the
width of the link is not smaller than the sum of W
1
and W
2
(Fig. 10).
Table 3. Two vessels on different links
______________________________________________
V
1
V
2
______________________________________________
Position A E
Velocity v
1
v
2
Domain Q
1
(
1234
111 1
,,,qqqq
) Q
2
(
1234
222 2
,,,qqqq
)
Domain length L
1
L
2
Domain width W
1
W
2
Time to leave t
1
t
2
______________________________________________
(a) (b)
Figure 8. Vessels on the same link.
(a) (b)
Figure 9. Vessel conflict on the same link depends on initial
vessel positions
Figure 10. Vessels travel in parallel on a link
4 EXAMPLES AND DISCUSSIONS
We have implemented the conflict detection algo-
rithm in the simulation model using Visual C++ 6.0
on a Windows XP operating system. In this section,
some examples are shown to illustrate the results of
our algorithm.
Figure 11 is the first example. Two vessels travel
toward the same node. A vessel is represented as a
rectangle with a solid line indicating the travelling
direction. The vessel on the left hand side is V
1
and
the other one is V
2
. The gray areas in Figure 11(a)
indicate the link areas. At any time, a vessel keeps
inside a link area. The gray areas in Figure 11(b),
enclosed by solid lines, indicate the relative move-
ment of V
1
to V
2
. A conflict is predicted since the
relative movement intersects with the domain of V
2
(Fig. 11(b)). Figure 11(e) shows that the two vessels
conflict with each other when they are passing the
node.
The result also shows that the conflict can be pre-
dicted at any time before the conflict time. Figure
11(b) and Figure 11(d) show the relative movements
of V
1
to V
2
at different positions. As we can see from
the figure, the same conflict is predicted in both po-
sitions. In fact, the conflict prediction algorithm will
detect the conflict any time before either vessel
reaches the node. The result implies that we can in-
crease the simulation time interval, thus reduce the
calculation for conflict detection.
Figure 12 is an example with two vessels travel-
ling on the same link. The vessel V
1
tries to catch up
with V
2
.The gray area enclosed by black lines in
Figure 12(b) is the relative movement of V
1
to V
2
.
Combining Figure 12(a) and Figure 12(b), we know
that this relative movement intersects with the do-
main of V
2
. Thus V
1
will catch up with V
2
. If the link
with is not enough, V
1
and V
2
will come into conflict
(Fig. 12(c). On the other hand, if the link width is
enough for the two vessels to navigate in parallel,
there will be no conflict (Fig. 13).
16
In Figure 13(a), two vessels in parallel will not
conflict with each other, because the relative move-
ment of V
1
to V
2
does not intersect with the domain
of V
2
. Therefore, V
1
can catch up with V
2
and over-
take it.
Figure 11. Two vessels from different links conflicting each
other at a node
Figure 12. Two vessels on the same link in conflict
Figure 13. Two vessels on the same link travel in parallel.
5 SUMMARY AND CONCLUSIONS
A simulation model has been developed for predict-
ing potential vessel conflicts within a seaport. An al-
gorithm for conflict detection was designed with the
use of ship domain criterion: when the relative
movement of one vessel with respect to a second
vessel intersects with the domain of the second ves-
sel, the two vessels will have a conflict. The algo-
rithm simplifies the conflict detection problem by
checking whether two parallelograms intersect with
each other.
An application of the model was demonstrated
using the seaport of Singapore as an example. Inputs
to the model include the background map, data on
fairways, and information on vessel types and char-
acteristics. Vessel arrivals and vessel routes are gen-
erated by the model according to statistical distribu-
tions. Simulation results showed that conflicts can
be accurately predicted in time. The logic of conflict
detection is applicable to other traffic systems by
changing the input data. Thus, the simulation model
is a generic model which can be adapted to other
busy seaports that are faced with traffic congestion
and delays.
For future work, human factor could be taken into
account. Human error would affect vessel movement
as well as conflict situation. An example is the situa-
tion where one vessel follows another vessel along a
link. Even with sufficient width for overtaking, an
accident/ incident may occur as a result of human er-
ror. Another possible improvement will focus on the
determination of a reasonable time step in simula-
tion. In the current system, a single conflict may be
detected for multiple times. When the time step is
too small, a conflict may be predicted many times.
On the other hand, if it is too large, some conflicts
may not be detected. Thus, the determination of an
optimum simulation time step is also an important
issue.
17
REFERENCES
Coldwell, T.G. 1983. Marine Traffic Behaviour in Restricted
Waters. Journal of Navigation 36(3): 430-444.
Davis, P.V., et al. 1980. A Computer Simulation of Marine
Traffic Using Domains and Arenas. Journal of Navigation
33(2): 215-222.
Fujii, Y. and Tanaka K. 1971. Traffic Capacity. Journal of
Navigation 24(4): 543-552.
Goodwin, E.M. 1971. A statistical study of ship domains.
Journal of Navigation 28: 328-344.
Lenart, A.S. 1999. Manoeuvring to required approach parame-
ters - CPA distance and time. Annual of Navigation 1: 99–
108.
Lenart, A.S. 2000. Manoeuvring to Required Approach Param-
eters - Distance and Time Abeam. Annual of Navigation 2:
81–88.
Pietrzykowski, Z. 2008. Ship's Fuzzy Domain-a Criterion for
Navigational Safety in Narrow Fairways. Journal of Navi-
gation 61(03): 499-514.
Zhao, J., et al. 1993. Comments of Ship Domains. Journal of
Navigation 46: 422-436.
Zhu, X.L., et al. 2001. Domain and Its Model Based on Neural
Networks. Journal of Navigation 54(1): 97-103.
