365
1 INTRODUCTION
Shipping, as the most efficient transportation mode is
responsible for more than 80% of global trade by
volume [13]. So, increased activity of the global
market corresponds with higher shipping demand
that can result in large traffic volume inside port
areas. Even though seaports tend to attract more
shipping activity, traffic overcapacity can increase
overall risk of accident. In order to continue the
growth trend of demand for services while
maintaining a high standard of safety, port
communities developed navigation risk assessment
models based on different methodologies.
General guidelines for developing navigation risk
assessments are provided by global organisations like
International Maritime Organisation (IMO),
International Association of Marine Aids to
Navigation, Lighthouse Authorities (IALA), World
Association for Waterborne Transport Infrastructure
(PIANC), etc. While some researches used approaches
provided by mention organisation, individual risk
assessments combined them with or established own
location-specific methodologies for securing safety of
maritime traffic in ports.
Most of existing risk assessments are based on
either historical data of the accidents or assumptions
derived from expert knowledge. Main benefit form
constructing models based on historical accident
databases is that causal factors are more obvious and
learning from a mistake process can be easily applied.
Drawback of this approach is often manifested as
limitation of accident case numbers in targeted area or
when data is incomplete. Also, historical data of the
Methodology for the Development of Parameters for
the Navigational Safety Risk Assessment Model in Port
Approaches
F. Bojić, R. Bošnjak, Z. Lušić & A. Gudelj
Maritime University in Split, Split, Croatia
ABSTRACT: Risk of an accident is an ever-present component in the maritime transportation process, especially
in congested waters such as port areas. Since safety is of crucial importance in the maritime industry, different
models of risk assessments were developed to ensure minimal navigational danger. The aim of this paper is the
development of modular, dynamic sets of parameters, applicable for future risk assessment models on port
approaches by introducing top-down expert appraisal structure methodology organised in three steps. Firstly,
approaches and criteria from relevant international recommendations and scientific studies on maritime risk
assessment models were analysed and compared, in order to obtain general categories of navigational safety
parameters. Secondly, existing risk assessment parameters were structured and combined into new dynamic
sets. In the third step, these dynamic sets of parameters were selected, and numerical values were assigned to
them according to the specific context of the port. Finally, this top-down methodology aims to provide relevant
dynamic sets of criteria for navigational safety risk assessment development that are flexible and widely
applicable for the needs and characteristics of different ports.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 15
Number 2
June 2021
DOI: 10.12716/1001.15.02.13
366
accidents does not take into account near miss
situations that involve high level of uncertainty to the
potential threat, therefore it not possible to have full
perspective of number, weight and interactions
between different risk factors. To compensate for the
lack of data on navigational accidents, some
researchers used expert analyses in their approach to
determine relevant navigation risk factors. Use of
expert judgment in complex system such as port
approach operations, can improve relevancy of
different risk components and make up for the ever-
present lack of information important for developing
risk assessment. Expertly approach can contribute to
the proactive nature of the methodology and may
improve quality of the historical data. Further
historical data may be evaluated by the use of expert
judgment by which the quality of the historical data
may be improved [5].
Although navigational risk understanding in
variety of scenarios can deepen the insight of potential
accidents and can extend the relevance of the results,
it won’t erase the uncertainty of the evaluation in the
process of predicting the final outcome. Because of the
uncertainty of processes inside maritime traffic, some
researches applied fuzzy logic methodology that
allows development of risk predicting models based
on imprecise or incomplete data. It was found that
deployment of fuzzy logic should enable taking into
account the insufficient information and the evolution
of available knowledge [1]. But since fuzzy logic
tolerates some level of data deficiency and
uncertainty, process of criteria election and validation
must rely on expert’s judgement. To ensure additional
relevancy of expert’s appraisal of risk criteria in fuzzy
logic setup, quantification of and examination of
previous risk assessment methods and models can be
used.
In this paper, top-down expert approach with
fuzzy logic background was applied throughout three
steps in order to define dynamic sets of criteria for
navigational safety risk assessment development.
Paper arrangement is compliant with mentioned
approach. So, in Section 2, after the presentation of
brief general methodological background, three
subsections are introduced. Initial subchapter contains
analysis and quantification of risk criteria from
relevant researches. In subsection 2.2. aggregated risk
criteria are selected and classified into fuzzy sets.
Also, causal connections between risk parameters are
clarified. In final subsection, method for validation of
risk criteria and its applicability are explained.
Finally, conclusion of proposed method is presented
and course of research development in the future is
pointed out.
2 METHODOLOGY
To accomplish the goal of this research, top-down
expert methodology was applied because accident
data is frequently qualitatively and spatially
restricted. Since the amount of casualties in ports is
limited, maritime traffic in ports cannot be assessed
based on single casualties. While it is impossible to
anticipate the risk for a nonextant situations based on
data-driven approach, this method also does not
allow the quantification of risk generated by near miss
situations, high traffic volumes or environmental
effects on navigation [3]. Various factors contributing
to the risk of a potential accident that were not
necessarily considered during the accident analysis
have to be taken into account in order to achieve
relevant results from risk assessment model. That is
why data important for conducting risk assessments
such as vessel information, weather influence or
traffic properties had to be gathered, quantified and
analysed by experts to determine their causal
relationships what can serve as a foundation for
developing a navigation risk model.
But to successfully apply different risk criteria on
risk assessment model, that were previously selected
through expert appraisal, it is necessary to have
methodological background that is able to produce
valid result in systems with incomplete data and level
of uncertainty. Therefore, the application of non-
binary fuzzy logic for creating connections and
assigning values to different parameters was found
suitable for predicting risk in uncertain, or in other
words, unprecedented environments such as port
approach operations and navigation in port basins.
The fuzzy logic is an efficient approach for design a
decision-making system in maritime domain. This
technique allows solving a lot of problems related to
dealing the imprecise and uncertain data [1].
So, focus of this paper was not on a historical
casualties nor risk aspects relevant to single location
but on providing modular fuzzy sets of risk criteria
that, when connected inside a risk model, can give
flexibility of defining realistic navigation risk scenario
of different port approaches.
2.1 First step – top-down criteria quantification and
analysis
To begin with the development of navigational risk
criteria sets, quantification and analysis of accident
factors from different navigation risk assessments was
conducted. The first step in risk quantification is to
define the boundaries and the objectives of the system
to be analysed [11]. With top-down approach firstly
general guidelines and recommended risk factors
from three different international organisations where
evaluated.
IMO presented methodology for risk control in
“Formal Safety Assessment” (FSA) document.
Through its five-step approach, guidelines regarding
hazard identification, risk analysis and control, cost-
benefit and decision-making recommendations are
provided with the aim of enhancing maritime safety
by developing and using risk analysis and cost-benefit
assessment [5]. Although FSA is publication that
offers detailed suggestions on data gathering and its
evaluation, application of expert judgment, use of
qualitative and quantitative methods or influence of
human error, its scope is wide, thus often not
completely applicable for the needs of different ports.
Inside this research expert appraisal and
quantification of risk data were considered, along
with suggested navigational safety aspects that are
generally presented in Table 1, while human error
was avoided due to its complexity that requires
different and thorough research.
367
Table 1. Overview and quantification of general risk criteria used in examined literature
__________________________________________________________________________________________________
Main category General criteria Specific criteria Source
__________________________________________________________________________________________________
Ship data Size Length [5] [16] [6] [3] [11] [6]
Breadth [16] [6] [6]
Gross tonnage [5] [6] [1] [3] [6]
Draught [16] [6]
Dynamics Speed [16] [6] [3] [6]
Manoeuvrability [16] [3] [11]
Characteristics Type of ship [5] [16] [6] [1] [11] [2]
Year of construction [5] [1] [11]
Number of companies [1]
Duration of detention [1]
Type of hull [5] [1]
Crew [16] [11]
Flag [1]
Pilotage requirements [11]
Escorting requirements
Propulsion [5] [11]
Steering [11]
Electrical power [11]
Structural integrity [11]
Environmental Wind effect Wind speed [16] [6] [1] [3] [11] [6] [2]
influence Wind direction [16] [3] [11] [2]
Sea effect Current [16] [6] [3] [11] [6] [2]
Sea State [16] [1] [6] [2]
Tides [6] [11]
Water density [6]
Ice [16] [6]
Visibility effect Visibility [16] [6] [1] [3] [11] [2]
Time of the day [1] [3] [2]
Traffic influence Traffic size Traffic volume [16] [3] [6] [2]
Time of year [16] [2]
Traffic diversity Traffic mix [16] [3]
Port organisation Port organisation Rules and regulations [3]
and assistance Navigational equipment
Port assistance Pilotage [16] [3]
VTS assistance [16] [3]
Tug assistance [2]
Port configuration Port design Water depth [6] [3]
Width [6] [3]
Location [6] [3] [11] [2]
Type of infrastructure [6] [3] [11]
Navigational aid Traffic separation [16] [6] [11]
__________________________________________________________________________________________________
IALA developed “Waterway Risk Assessment
Program” (IWRAP), quantitative modelling tool
useful for providing a standardized method of
assessing the risks within most waterways [8, 16].
IWRAP can estimate the frequency of collisions and
groundings in a given waterway based on information
about traffic volume/composition, route geometry and
bathymetry [16]. This model is convenient for
acquiring relevant estimation of the annual number of
collisions for specific area but it is difficult to calculate
the level of risk for individual navigation scenarios,
especially because causation of other risk factors
inside model, like environmental influence or port
organisation are quite uncertain. Also, it is difficult to
apply in areas with complicated traffic tracks [8]. Risk
factor used in this model are generally described in
Table 1.
PIANC introduced “Harbour approach channels
design guidelines”, a report that provides
recommendations for the design of harbour approach
channels, the manoeuvring and anchorage areas
within harbours, along with defining restrictions to
operations within a channel [9]. Even though focus of
this document is primarily orientated towards the
navigational design of ports and their approaches, its
theoretical explanations and relationships between
criteria important for safe navigation in enclosed
waters are well-defined and applicable for
development of risk assessment for port approaches.
General criteria presented in this document are
illustrated in Table1.
In addition to the general recommendations from
IMO, IALA and PIANC, five different relevant studies
that tackled the problem of navigation safety with
various methodologies where analysed and their
navigational risk factors were examined and
displayed in Table 1.
In research “Utilizing the fuzzy IoT to reduce
Green Harbour emissions” conducted by S.L. Kao, J.L.
Lin and M.R. Tu, fuzzy logic was applied inside risk
model in order to determine safe manoeuvring speed
which will contribute to better safety and air quality
standards of Keelung Port [6]. For defining the
Nautical Port Risk Index, in paper “Risk Assessment
Methodology for Vessel Traffic in Ports by Defining
the Nautical Port Risk Index”, X. B. Olba, W. Daamen,
T. Vellinga and S. P. Hoogendoorn used expert
validation and quantification method on several risk
assessment researches to create risk factors from
which risk assessment model was derived for Port of
Rotterdam [3]. Paper “A decision-making system to
maritime risk assessment” of J.F. Balmat, F. Lafonta,
368
R. Maifret and N. Pessel has also demonstrated the
use of fuzzy approach for development of decision-
making system for mitigating risk of accidents and
pollution based on expert appraisal of factors from
different sources [1]. In research “Risk based
methodology for safety improvements in ports”, V. M.
Trbojevic and B. J. Carr offered methodology and risk
criteria for development of navigational risk
assessment relevant for seaports [11]. In paper
“Simulation Method - Based Oil Spill Pollution Risk
Analysis for the Port of Šibenik” conducted by G.
Belamarić, Ž. Kurtela and R. Bošnjak risk factors from
different relevant sources were applied inside risk
matrix to determine the level of chance for accident
occurrence in different scenarios relevant for
navigation in Šibenik port approach channel [2].
Finally, the objective of first phase was to achieve
transparent analysis of relevant studies and offer
overview of their risk parameters illustrated in Table
1. that served as a foundation for developing valid
dynamic sets of risk criteria based on expert
understanding in next step of this research.
2.2 Second step – criteria aggregation and their
classification into dynamic fuzzy sets
As can be seen in Table 1., five main categories of
criteria are developed, along with twelve general
criteria groups, all based on specific risk factors from
sources examined in first step of this research. In this
phase, risk factors that were previously quantified
and analysed are now aggregated, classified and
connected inside different fuzzy sets that interact with
each other in a dynamic manner. All risk factors
related to navigation should be assessed in a
structured way, and a selection of the important
factors should be made according to expert judgement
for development of valid risk assessment [5]. With
reliance on expert perception of the causal
connections between aggregated safety parameters,
dynamics sets of risk criteria are formed, and ones
relevant for Port of Split case study are presented in
Table 2.
Table 2. Dynamics fuzzy sets of risk criteria relevant for Port
of Split case study
_______________________________________________
1 Ship data
_______________________________________________
Ship size Gross tonnage
Ship dynamics Speed
_______________________________________________
2 Environmental influence
_______________________________________________
Ship characteristic Type of vessel
Wind effect Wind speed
_______________________________________________
3 Traffic influence
_______________________________________________
Traffic size Time of year
Traffic size Traffic volume
_______________________________________________
The aim of categorising selected criteria was to
connect them inside sets that could be applicable for
navigation scenarios in different ports. By adding or
removing preconfigured risk sets, flexible model
design is enabled. But since focus of this research is on
methodology for development of risk parameters,
process of aggregation, classification and connection
of criteria displayed in Table 2 is examined.
Since vessel length can be associated with gross
tonnage (GT) of individual vessel type, first set of
criteria demonstrates connection between vessels size
which is described by a function of its mass expressed
in metric tons and vessel speed expressed in knots
(KT). This way level of risk for individual ship motion
can be assessed [6]. In second set, windage area of
different ship kinds is represented by type of vessel
criteria and related with wind speed expressed
according to Beaufort scale (BF), so that
environmental influence can be measured as level of
wind effect on diverse freeboard designs and surface
sizes [10]. Traffic influence is manifested as traffic size
that is relevant to its dynamic throughout the year and
its anticipated volume inside port basin and port
approach in different day periods [7].
All three criteria sets, illustrated in Table 2. are
structured based on expert comprehension of the
causal relations between navigation risk criteria, and
to develop functional navigation risk assessment they
need to be connected in a same manner. The
interaction between parameters denotes changes in
the risk profile due to changes in port management,
ship characteristics, or other parameter sets [11]. But
before model designing and obtaining any results,
each component has to be transformed into numerical
data.
2.3 Third step – assigning data-based values to risk
criteria
Complex operations like port approaches, here
defined through individual dynamic sets of risk
criteria, are described as processes with degree of
uncertainty. The maritime risk evaluation can find an
interest in the fuzzy logic approach because much
data are linguistic variables [1]. That is why in this
research top-down expert appraisal with fuzzy logic
background was used to quantify, select, classify and
finally characterise risk parameters by assigning data-
based values to them. Each value is represented by
membership function that defines the degree of truth
as an extension of its valuation [14]. Although there is
no clear limitation for number of criteria and
membership functions in each fuzzy set, by increasing
their quantity connection between them will grow
exponentially, thus potential model will become too
complex and inadequate for intended application.
That is why usually there are two to five membership
functions per criterion that is represented by average
value.
Databases and publications of local relevancy are
examined in this phase to determine the membership
function of each criterion in fuzzy sets, presented in
Table 2 for Port of Split case study. The reason for
considering data from port-oriented sources is to
ensure the adaptability of the final risk model to the
specific needs of that area.
To establish three categories for average sizes of
ships that arrive in Port of Split, available database of
Split Port Authority (SPA) and Croatian Register of
Shipping (CRS) were analysed. For acquiring
information about average ship speeds, Automatic
Identification system (AIS) was used. Types of vessels,
determined also by examining SPA database, served
as a three membership functions of windage areas
369
relevant for each vessel type inside environmental
influence fuzzy set. Membership functions that
represent wind speed mean values were established
by relying on information from Maritime navigation
study, navigating area of Split and Dubrovnik,
Admiralty sailing direction NP 47, Mediterranean
Pilot Vol.3 and Croatian Hydrographic institute Pilot.
[4, 12, 15]. Maritime navigation study, navigating area
of Split and Dubrovnik, SPA database, AIS data
served as basis for validating traffic size and the
navigation risk area. First, vessel arrivals throughout
the year were analysed to determine traffic volume in
different periods. Similar approach was applied on
three periods of the day where AIS was used to track
the number of vessels in motion around estimated
manoeuvring port area of about 1 square nautical mile
(NM2). This way, approximate number of active
vessels per 1 NM2 (V/NM2) could be anticipated
temporally, so year and day periods are used as
membership functions to express assumed traffic
quantity. Average values of all risk criteria relevant
for Port of Split case study are presented in Table 3.
Table 3. Classification of validated membership functions
for each risk criteria in fuzzy sets relevant for Port of Split
case study
_______________________________________________
1 Ship data
_______________________________________________
Classification Large Medium Small
Ship size (GT) 28411 6496 3156
Classification Fast Moderate Slow
Ship dynamics (KT) 15 10.5 6
_______________________________________________
2 Environmental influence
_______________________________________________
Classification Large Medium Small
Type of vessel Ro-Ro/ Bulk/ Bulk/
Passenger/ General/ General/
Yachts/ Tankers - Tankers -
Boats ballast loaded
Classification Gale Moderate Gentile
Wind speed (BF) 8 6 3
_______________________________________________
3 Traffic influence
_______________________________________________
Classification High Moderate Low
Time of year Season Pre/ Offseason
Postseason
Classification High Moderate Low
Traffic volume Daytime Dusk/Dawn Night-time
(V/NM2)
_______________________________________________
Implementation of membership functions in model
designing allows graphical expression of fuzzy set
elements. For this model, triangular membership
function was found as most convenient for graphical
description of values in each fuzzy set, since average
values represent the highest degree of membership in
each class. Classification of risk criteria, expressed
through triangular membership functions was
conducted in MATLAB software. Membership
functions of Ship dynamic criteria are presented in
Figure 1.
Figure 1. Triangular membership functions of ship’s
dynamic as part of ship data fuzzy set
Lastly, validation of risk criteria as a key element
for the development of flexible navigational risk
assessment model is ensured by completing the final
phase of top-down expert methodology. After
applying the final methodological step inside the
model composed in MATLAB software,
defuzzification process of realistic navigational
scenarios produced output that had a high level of
compatibility with the expert’s expectations. Surface
representation of defuzzification process is illustrated
in Figure 2.
Figure 2. Surface representation of defuzzification process
with three fuzzy sets from Table 3.
As can be seen in Table 3, both general relevancy
and spatial flexibility of chosen dynamic fuzzy set can
be established through the three-step process, where
general risk criteria are structured in fuzzy sets, then
selected and validated in relation to the port’s needs.
3 CONCLUSION
Port-related navigation is a complex operation where
the risk of an accident is an ever-present component.
Various factors can negatively affect navigational
safety, particularly in spatially limited port areas. To
mitigate the risks of port approach activity, this study
proposed the methodology for the development of
parameters for the navigational safety risk assessment.
Expert knowledge was used throughout three
methodological segments to define modular dynamic
sets of risk criteria that can serve as a foundation for
the development of navigational risk assessment
model based on fuzzy logic.
The model constructed according to the concept
provided in this research should be able to estimate
the level of navigational risk of realistic scenarios in
different ports. To meet the diverse needs of
particular ports while retaining universality of
selected risk criteria, a top-down approach was used.
With this method, the process of risk criteria
development went from quantification and selection
of general and global risk parameters to their
validation based on specific and local data. For more
accurate and prompt risk prediction of individual
navigation scenarios, this study proposed use of AIS
data for determining input values. But since not all
ships that participate in maritime traffic are required
to have an AIS device, limitation of partial
370
navigational safety surveillance could be bypassed by
introducing novel Internet of Things (IoT) platform.
Possibility of integration of devices and decision
makers inside a unique risk assessment model can be
secured through IoT web, what would in the end lead
to better safety aspect of port approaches. Since the
initial model has shown an adequate level of output
validity, for now, the extension of this research will be
directed towards design improvement of the model’s
architecture by adding more fuzzy sets, its
programming inside the fuzzy logic software, and
additional proofing of output based on realistic
navigational scenarios.
REFERENCES
1. Balmat, J.-F., Lafont, F., Maifret, R., Pessel, N.: A
decision-making system to maritime risk assessment.
Ocean Engineering. 38, 1, 171–176 (2011).
https://doi.org/10.1016/j.oceaneng.2010.10.012.
2. Belamarić, G., Kurtela, Ž., Bošnjak, R.: Simulation
Method - Based Oil Spill Pollution Risk Analysis for the
Port of Šibenik. Trans. Marit. Sci. 5, 2, 141–154 (2016).
https://doi.org/10.7225/toms.v05.n02.005.
3. Bellsolà Olba, X., Daamen, W., Vellinga, T.,
Hoogendoorn, S.P.: Risk Assessment Methodology for
Vessel Traffic in Ports by Defining the Nautical Port Risk
Index. Journal of Marine Science and Engineering. 8, 1,
(2020). https://doi.org/10.3390/jmse8010010.
4. Croatian Hydrographic Institute: Peljar I. Jadransko
more- istočna obala. (1999).
5. International Maritime Organization: Revised
Guidelines for Formal Safety Assessment (FSA) for Use
in the IMO Rule-Making Process. , London, UK (2018).
6. Kao, S.-L., Lin, J.-L., Tu, M.-R.: Utilizing the fuzzy IoT to
reduce Green Harbor emissions. Journal of Ambient
Intelligence and Humanized Computing. (2020).
https://doi.org/10.1007/s12652-020-01844-z.
7. Li, M., Mou, J., Liu, R., Chen, P., Dong, Z., He, Y.:
Relational Model of Accidents and Vessel Traffic Using
AIS Data and GIS: A Case Study of the Western Port of
Shenzhen City. Journal of Marine Science and
Engineering. 7, 6, (2019).
https://doi.org/10.3390/jmse7060163.
8. Park, Y.-S., Kim, J.-S., Aydogdu, V.: A Study on the
Development the Maritime Safety Assessment Model in
Korea Waterway. Journal of Navigation and Port
Research. 37, 6, 567–574 (2013).
https://doi.org/10.5394/KINPR.2013.37.6.567.
9. Pianc: MarCom WG 121: Harbour Approach Channels -
Design Guidelines…,
https://www.pianc.org/publications/marcom/harbour-
approach-channels-design-guidelines, last accessed
2021/04/26.
10. Szymoński, M.: Some Effects of Wind on Ship’s
Manoeuvrability. TransNav, the International Journal on
Marine Navigation and Safety of Sea Transportation. 13,
3, 623–626 (2019). https://doi.org/10.12716/1001.13.03.19.
11. Trbojevic, V.M., Carr, B.J.: Risk based methodology for
safety improvements in ports. Journal of Hazardous
Materials. 71, 1, 467–480 (2000).
https://doi.org/10.1016/S0304-3894(99)00094-1.
12. United Kingdom Hydrographic Office: Admiralty
sailing direction Mediterranean Pilot Volume 3, NP 47.
(2014).
13. United Nations Conference on Trade and Development
(UNCTAD): The Review of Maritime Transport 2020. ,
United Nations Publications 300 East 42nd Street New
York, New York 10017 United States of America (2020).
14. Zadeh, L.A.: Fuzzy sets. Information and Control. 8, 3,
338–353 (1965). https://doi.org/10.1016/S0019-
9958(65)90241-X.
15. Zec, D., Frančić, V., Rudan, I., Maglić, L., Lušić, Z.,
Ðurđević-Tomaš, I., Brajović, M., Vukić, M.: Prometno-
plovidbena studija plovno područje Split, Ploče i
Dubrovnik. Faculty of Maritime Studies of the
University of Rijeka (2014).
16. IALA AISM, https://www.iala-aism.org/, last accessed
2021/03/19.
