
824
Based on the conditional independence of
variables and the chain rule, Bayesian belief network
allows to determine the joint probability distribution
of the variables.
The decision nodes represent the risk control
options and their costs. The utility nodes represent
risks of possible accidents, including costs related to
their consequences.
The events – nature nodes of Bayesian influence
diagram considered in the general model are
presented in Tables 1-3.
Table 1. Definition of the events – berthing, mooring,
moored ship and unberthing
________________________________________________
Node Description of probability States
________________________________________________
Berthing Probability of an accident Safe
during berthing ALARP
Unsafe
Mooring Probability of an accident Safe
during mooring operations ALARP
Unsafe
Moored Probability of an accident Safe
Ship related to the moored ship ALARP
Unsafe
Unberthing Probability of an accident Safe
during unberthing ALARP
Unsafe
________________________________________________
Table 2. Definition of events - ship and port technical
conditions
________________________________________________
Node Node description States
________________________________________________
Ship Probability of failure on board ship Safe
ALARP
Unsafe
Port Probability of failure in port Safe
ALARP
Unsafe
________________________________________________
Table 3. Definition of events – external conditions
________________________________________________
Node Node description States
________________________________________________
Wind Probability of a dangerous wind Yes
speed and direction No
Current Probability of a dangerous current Yes
speed and direction No
Passing Probability of a dangerous impact Yes
vessel of a passing vessel No
________________________________________________
Table 4. Definition of risk reduction options – decision
nodes
________________________________________________
Node Node description States Cost keuro
________________________________________________
Tug Decision of tug boat Yes 1-1.5
boat assistance No
Leaving Decision on emergency Yes 103–105
port leaving the port No
Mooring Decision on application of Yes
lines additional mooring lines No
________________________________________________
Risk reduction options are related with decisions
to employ tug boats, leave the port and use additional
mooring lines. The costs of tug assistance during
berthing and costs related to emergency leave can be
estimated for the particular port and vessel [5].
For example the cost of slight impact with jetty is
up to 200 keuro e.g. touching dolphin as a result of a
failure to shift controls from central console to bridge
wing. Cargo spill , pollution can cause severe financial
loss – millions of euro, loss of reputation of a
company, in severe case could face bankruptcy. Fire
can cause severe financial loss – millions of euro.
The casualties mainly relate to accidents during
mooring operations, such as broken mooring lines or
structural damage to mooring equipment. There is
also a risk of an accident in the event of a collision or
collision with another object or vessel.
The definition of decision nodes of Bayesian
influence diagram are presented in Table 4.
The utility nodes of Bayesian influence diagram
presenting the risk of possible accidents are presented
in Table 5.
Table 5. Risk of possible accidents – utility nodes
________________________________________________
Node Node description
________________________________________________
RP Risk of port facility damage / delay in port
operation
RI Risk of people injuries
RF Risk of fatalities
RE Risk of environment pollution
RS Risk of ship damage
________________________________________________
The most important for development of the risk
model is the design of the network structure and then
input of the dependent probability values for the
defined nodes which allows for calculations of the
joint probability distribution for the nodes. The model
presented in the paper was developed using a
commercial tool for Bayesian belief network
development - Hugin Researcher - the computer
program with graphical interface, compiler and
system for design and use of knowledge base.
Bayesian network allows to dynamically assess the
probability of accidents. The information about the
occurrence frequency of the top event propagates
backwards through the network, changing the
probability of the primary events [9].
In the presented network, in the events of ship
damage or port facility damage, risks are calculated
on the basis of the joint probability distribution of
accidents and their costs dependent on states of the
events which can be negligible, minor, moderate,
major or catastrophic. In case of further results of ship
damage and port facility damage accidents like fire,
explosion and toxic leakage the consequences can be
personnel injuries, fatalities and environment
pollution.
The Bayesian risk model of port manoeuvres of a
chemical tanker is presented in Figure 3.
4 RISK CONTROL OPTIONS OF POSSIBLE
ACCIDENTS DURING PORT MANOEUVRES
The risk control options include proper exchange of
information between the vessel and terminal before
berthing, proper weather information, tug assistance
during navigation in difficult to manoeuvre and
dangerous areas, tug assistance during berthing,
unberthing, emergency port leave and proper
prediction and control of mooring forces [3].