Professional Documents
Culture Documents
CONTENTS
Page No.
1. INTRODUCTION
1.1
1.1 BACKGROUND TO THE FSA APPROACH
1.1
1.2 PROJECT OBJECTIVES 1.3
1.3 SCOPE OF WORK
1.4
Page No.
10. REFERENCES
10.1
GLOSSARY OF TERMS
Accident Categories
Collision is defined as the ship striking or being struck by another vessel, regardless of
whether under way, anchored or moored. Collision does not include striking underwater
wrecks. Contact is defined as the ship striking, or being struck, by an external object (eg.
navigation buoy or berth), but not another ship or the sea bottom.
Loss of Hull Integrity is the consequence of certain initiating events and is used to denote
any damage to the external hull, or to the internal structure and sub-division, such that any
compartment or space within the hull is opened to the sea or to another compartment or
space.
Abbreviations
1. INTRODUCTION
The project team drawn from the BOMEL consortium was supplemented by a large
number of experts. Many made contributions throughout the trial application by
attendance at structured review meetings or by individual input. In addition BOMEL
undertook extensive checking and quality assurance activities in accordance with
its ISO 9001 certified Quality Management System. In fulfilment of the Agency's
requirement for an independent review of the study, BOMEL chose to assemble an
Independent Project Review Panel (IPRP) consisting of acknowledged experts in
HSC operations, risk and safety assessment and international maritime regulation.
This report summarises the work done for Agency Research Project 404 starting
with a brief introduction to the development of FSA for maritime application. The
report then describes further methodology development carried out under this
project and summarises the trial application to HSC. It concludes with a
commentary on the practicability of the methodology and by making
recommendations as to how the methodology and its application might be further
improved and how the results of the trial application might be extended and used in
the improvement of regulations.
* As from 1 April 1998 the Marine Safety Agency merged with the Coastguard Agency to
form the Maritime and Coastguard Agency
Previous Agency funded projects considered the benefits of applying FSA to the
shipping industry and then went on to develop a detailed methodology for
application of FSA to the IMO rule making process. The objective of this approach
is to concentrate resources on high risk areas and to ensure that new regulations
were developed with a clear understanding of the risks being addressed and the
effect of the regulations in reducing such risks. The resulting methodology
consisted of five principal steps as follows:
Step 1
Identification of hazards in which different types of accident scenario are
identified. These scenarios fall within one of ten generic Accident Categories (eg.
Collision and Contact, Grounding, Fire, Loss of Hull Integrity). Factors contributing
to the causes of accident initiation, escalation and the level of loss are identified.
Step 2
Assessment of risks associated with the hazards identified in Step 1 in which
the events and factors which influence the likelihood of an accident occurring and
its outcome in terms of losses or consequences are structured and then quantified
by reference to historic data or with the assistance of expert judgement.
Step 3
Consideration of alternative ways of managing the risks assessed in Step 2
in which risk control methods (RCMs) are identified, grouped into comprehensive
risk control options (RCOs) and assessed for their effectiveness in reducing risk.
Step 4
Cost benefit assessment of alternative risk management options in which the
RCOs from Step 3 are assessed in terms of their cost of implementation, the
monetary benefits which may result and therefore the resulting net cost per unit
reduction in risk (CURR).
Step 5
Recommendation of options for decision making in which the costs and
benefits for each stakeholder (or interested entity) affected by an RCO are
considered to ensure, that as far as possible, the principle of ‘the risk imposer
pays’ is satisfied and the most cost effective and equitable regulations result.
The UK papers discussing the FSA methodology developed through the Agency’s
Phase 1 research programme were presented to the IMO at MSC 66. Additionally,
The Working Group met again at MSC 67 and MSC 68 and finalised IMO Guidelines
which were subsequently adopted by the Committee at MSC 68 (Reference 3).
The trial application was reported in two detailed documents prepared by the
BOMEL project team. The first (Reference 4) describes the work and results of the
trial application itself, whereas the second (Reference 5) describes methodology
improvements developed during the project as described in Section 2 of this
summary report.
The scope of work for the trial application was defined by the Agency Project
Specification (Reference 6) and was as follows:
• To report the results in a form suitable for submission to the IMO's Maritime
Safety Committee, so as to promote understanding and acceptance of the
methodology amongst all IMO members.
• To validate the FSA approach and demonstrate its practicability, thus facilitating
its future use by the IMO.
• To extend the scope and contents of the HSC database produced by Project
384, to include data relevant to Steps 3, 4 and 5 of FSA.
• To generate a hierarchical risk profile for HSC covering the accident categories
considered.
The scope covers all five Steps of the FSA approach. The project was faced with
tight time constraints in order that the findings of the study could be reported to
MSC 68. Therefore the scope of the project was limited to the following three major
categories of accident.
2.1 INTRODUCTION
The methodology specified for the trial application was that which had been
developed under Phase 1 of the Agency’s FSA research programme referred to in
Section 1 (Reference 1). However additional methodology development was
undertaken in three areas as discussed below. A major requirement of Phase 1
had been to incorporate methods for accounting for the contribution of human error
to marine incidents and for evaluating the effectiveness of risk control approaches
to reducing human error. An approach, based on a hierarchical network of
influences (Influence Diagrams) from the regulator, through the implementation of
regulations within an organisation, to the direct influence on performance of
personnel and equipment, had been identified as the most promising approach.
Within the constraints of Phase 1, this had not been developed into a generic and
user friendly tool. Therefore additional work to make improvements in both the
structuring and quantification of influence diagrams was included in this project.
Another methodology development activity carried out in this project concerned the
elicitation and use of expert judgement. This becomes particularly important for
topics such as HSC where much of the technology is new and rapidly developing
and where, due to limited operational experience, historic incident data are not
statistically reliable. In such cases the judgement of experts is vital in estimating
risk levels and the likely effectiveness of identified risk control approaches.
Therefore alternative methods to incorporate judgement were investigated. The
Delphi methodology appeared to be an appropriate and widely recommended
approach and therefore this was investigated and subsequently used in the trial
application.
During Phase 1 both the Tavistock Institute and HRA provided input on the
influences which are thought to affect human performance. In parallel an IMO
correspondence group on the human element had also been collecting useful
thinking in this field. Therefore the first task was to collate and review this
information to ascertain whether a generic set of influences covering all aspects of
both human and hardware performance and the effect of external events on
accident causation could be identified.
The second stage in the work was to improve the quantification process and make
it transparent and easily understandable to the marine practitioners and experts
who would be required to make judgements on which the quantification would be
based.
The definitions for each of the direct level influences are given below:
Influence Definition
Competence The capacity or ability to perform a task.
Influence Definition
and proximity to land.
Influence Definition
The definitions for each of the organisational influences are given below:
Influence Definition
making.
Inspection / Maintenance The system that ensures equipment is maintained in
good working order.
Design Quality The system that ensures designs are fit for purpose
and meet user requirements.
It is also important to define clearly the range of each influence from best to worst
conceivable practice. Within the trial application it was found useful to define best,
worst and medium scalar positions for each influence. Examples at both the direct
and organisational level are provided below.
The scalar positions should be defined for each influence in order that there is a
clear and precise understanding prior to embarking on quantification.
occurrence of the top event. These relationships between influences are identified
and linked in a network which forms the quantification model as described below.
The contributory influences on the influences at a higher level are weighted. For
any particular circumstance each influence can be rated as described below. In
principle, the weighting factors are generic, ie. applicable to all circumstances;
whereas the ratings are assigned for a specific situation, eg. a particular level of
operating procedure. The weighting factors and ratings are then used to evaluate
the overall influence on the likelihood of the top event occurring.
It is proposed that the diagram is constructed from the top level using a top-down
approach. Thus, from the top level, the lower level influences affecting the event
are identified and weighted. This is repeated at the next level down, and so on
through the diagram. The weighting factors should be between 0 and 1 (or 0 and
100%) and at any particular node the weighting factors, W, from the immediate
lower level influences should sum to unity (or 100%). Thus, for n contributory
influences at a node j:
∑W ji =1 Equation 2.1
i =1
At the bottom level under consideration the influences are rated on their relation to
best possible and worst practice. The scale can be over any range, but each
influence should have the same range. A typical scale is 0 to 1, with 0 having the
worst or no effect, and 1 representing the best identifiable practice as defined in
Section 2.2.1.
Assigning the weighting factors and ratings to the various influences are the most
important tasks of the work. The weighting factors and ratings may be obtained by
expert judgement for example by:
Using the approach discussed above the overall influence can be evaluated very
simply. Starting at the bottom of the diagram, for any particular node j the influence
rating, r, is evaluated as:
n
r j = ∑ W ji r ji
i =1
Equation 2.2
This is repeated for all nodes at this level, and then for the next level up, until the top
level is reached.
Since all of the weighting factors sum to unity at each node the range of each
influence rating is the same as the range at the bottom level, eg. from 0 to 1.
At this stage of development of the methodology it is felt useful to have the ability to
compare the ratings calculated from Equation 2 with expert judgement and where
differences arise to modify the ratings to take this judgement into consideration.
One method for this would be simply to average the calculated value and the expert
value thus giving equal weight to both sources. This method was adopted in the
trial application.
Evaluation of Risk
The quantification process described above results in ratings for the direct causes
(human failure, hardware failure and external events). Weighting factors are
assigned to each and an index for the top event results. This index (the Influence
Diagram Index) can be calibrated to an established risk level for the top event in
order to calculate how the risk will be affected by changes in influence weighting
factors or ratings brought about by application of a risk control measure (RCM) or
package of measures (RCO). In order to do this it is necessary to establish a
transformation function between the Influence Diagram Index and risk magnitude.
Figure 2.2 relates the range of worst to best practice for all relevant influences (0 to
1 in the index) to a three orders of magnitude change in the risk. Three orders of
magnitude are selected on the basis that individual risks span 103 from the border
of tolerability to the level where society currently places no demand for further risk
reduction however low the cost (see Reference 7). The literature on occupational
risk across a range of industries indicates this as the range of risk from the most to
least dangerous occupations.
Figure 2.2 is presented in a log-linear format such that changes in the index
translate to a relative change in risk. If Ro is the baseline risk under present
practice and Rrco is the risk after implementation of the RCO with Io and Irco the
corresponding Influence Diagram Indices:
Note that in Figure 2.2, n is a number such that for a given value of the Influence
Diagram index (lo) and corresponding level of risk (Ro), the intercept on the risk axis
(ie. the x value) is such that log (Ro) = n + x. Thus the whole range of variation of
risk, corresponding to the range of variation of the Influence Diagram index from
zero to 1, comprises three orders of magnitude, or from zero to 3 on the scale of n
+ x.
Figure 2.2 can be applied to all Accident Categories to assess the effectiveness of
the respective RCOs in terms of a relative reduction in risk. For each Accident
Category, the absolute risk depends on the specification of n (shown in Figure 2.2)
which is calibrated against an established risk level.
The logarithmic representation of risk reduction in Figure 2.2 implies that the same
absolute improvement in index has a larger absolute reduction in risk when the
baseline index is low, and less effect when practice is already delivering a high
index value.
The Delphi technique is a forecasting technique which elicits and refines expert
judgement and has been extensively used to support decision making in a wide
variety of applications. It is a particularly appropriate method when there is a lack of
available historical data.
The classic Delphi technique was developed and first used by the US Air Force in
the early 1950's. The primary objective was to obtain `the most reliable consensus
of opinion from a group of experts by a series of interactive questionnaires
interspersed with controlled opinion feedback'.
The widespread use of the Delphi technique has resulted in the evolution of a
number of Delphi variations. The general view amongst Delphi users seems to be
that there is no hard and fast rule for applying the technique, although the classic
Delphi typically involves four rounds of questionnaires and feedback and appears
to take about 4 weeks. There also appears to be no firm consensus on the number
of iterations which are required, although most applications of the technique which
have been identified involve between two and six iterations. A study into the
optimum number of rounds (Reference 8) concluded that `.... Delphi groups
reached stability in their decision making after the fourth iteration, thereby providing
empirical support for the length of the classic Delphi'. Nevertheless, many of
studies have used as few as two iterations.
• The questionnaires are collated and average values calculated. Where there
are significant differences, the range of responses and justifications are
circulated back to the experts for reconsideration and further justification.
By way of example, a typical question set drawn from the hazard ranking exercise
in Step 1 is shown in Figure 2.3.
Contact
Minor Moderat Major Total Justification
e (This is essential)
Note: The estimates in this table would suggest that approximately 95% of all Contacts which occur whilst Berthing,
Berthed, or Unberthing would have minor consequence (ie. less than 10 minor injuries), 4% would have moderate
consequence, and 1% would have major consequence.
Advantages
The Delphi Technique has been applied to a broad range of problems throughout
the world, including Cost-Benefit Assessment and Risk Assessment. It is seen by
many researchers to be one of the most successful predictive tools available.
Some of the reasons for this are:
• The technique has some of the benefits of group decision making, whilst at the
same time reducing the negative effects of group interactions and any tendency
towards domination;
Disadvantages
An effective and thorough application of the Delphi Technique requires substantial
effort and a great deal more time than some other decision making techniques and,
consequently, there is a tendency to reduce the number of rounds to as few as
possible. This temptation should be resisted as the iterative process and
associated justifications are a major strength of the technique.
An HM Treasury report of 1996 stated "It is widely held among regulators, but much
less often among academic or other commentators, that people should as a
general rule be less well protected from the risk of incidents that effect only one or
a few people at once than from the risks of incidents which affect many people,
such as a major rail or air crash, although there is no consistent evidence that this
reflects the preferences of those at risk" (Reference 9).
Valuations of Regulators
There is evidence that safety regulators give more than directly proportionate
weight to the prevention of accidents which cause large numbers of deaths. This
would seem to reflect institutional concerns about, for example, adverse publicity
and political or commercial consequences, rather than the preferences of those at
risk or of the public at large. Evidence on the sizes of risk aversion factors by
regulators is scarce, while risk aversion factors are likely to vary from industry to
industry. Clearly what would be of particular value would be risk aversion studies in
the shipping industry but, so far as we are aware at present, no such studies exist.
In financial terms there are very few examples in transport where risk of death
(either in a small or a large accident) affects human behaviour. In shipping neither
the Herald of Free Enterprise or the Estonia adversely affected the growth of
passenger journeys in Ro-Ro ferries. The only examples identified that could be
considered to be aversion are the general reduction in air traffic following the Gulf
War and the reduction in French traffic accidents following the French School Bus
collision in the 1980s. However in both these cases there is no evidence to
determine if people were influenced by the severity of the accident or just by its
occurrence and attendant publicity.
It should therefore be possible to use the current priorities for action by the IMO as
an indication of the comparative level of risk to which, by implication, the IMO are
reacting. Three of the main priorities of the IMO at present are the general level of
worldwide shipping losses, Ro-Ro ferry losses and Bulk Carrier losses.
If the IMO were acting to reduce Potential Loss of Life then the relative focus of IMO
activities should be in line with the PLL, ie. they should be spending 27 times more
effort (771/29) on worldwide shipping losses than on Ro-Ro ferries.
Therefore it does appear that the regulators’ decisions have been affected by
aversion although, as stated above, there have not yet been comprehensive
studies or debate to determine reliable values. In the trial application it was found
that, for all accident categories considered, the range of outcomes included large
scale losses and therefore the inclusion of aversion would not affect the rankings
between accident categories. Furthermore many of the RCOs developed were
aimed at reducing accident frequency rather than being aimed at any specific size
of outcome. Therefore it was decided to present a unified picture of the risk profile
and to value costs and benefits across all accident categories and outcomes with
no weighting for aversion.
It is important to associate the existing risk profile with the risk control measures
required by relevant international and national regulations. Apart from the IMO HSC
Code other IMO Rules such as MARPOL, STCW, ISM, IMDG, IRPCS and SOLAS
together with Flag State interpretations and requirements, Port State requirements
and Classification Society Rules contain provisions affecting HSC.
Figure 3.1 Relationship Between Generic Shipping Descriptions and Accident Categories
Figure 3.2 Typical Layout of a High Speed Twin Hull Passenger Ferry
3.3 HSC STAKEHOLDERS (OR INTERESTED ENTITIES)
These stakeholders are considered further in later steps of the FSA process. The
Risk Control Options may lead to significant changes to risks and costs
disproportionately affecting one or more of these stakeholders and therefore these
differences may be significant and need to be addressed during Step 5 (Decision
Making).
Human expertise and decision making are fundamentally involved in all aspects of
the design, construction, operation and maintenance of HSC and human error is a
major contributor to most shipping incidents. To develop an understanding of how
human factors contribute to incidents in the shipping industry, it is first necessary to
understand the environment in which those responsible work. In coming to
understand this environment it is necessary to consider such factors as the man-
machine interface, the working environment (physical and psychological), working
relationships and motivation and the various factors influencing them. It is usual to
gather the information required using a combination of formal and informal
interviews and most importantly, by observation.
Task Inventories identify those tasks which are carried out by HSC personnel that
are important to the successful operation of the vessel and were developed during
the course of interviews with HSC Masters and Officers on craft engaged in short
sea "international" voyages between the UK and France, UK and the Channel
Islands, and Scotland and Northern Ireland.
The Task Inventory identified fourteen principal activities, the first thirteen of which
follow the pattern of normal operation of an HSC, as follows:
The fourteenth item deals with emergency preparedness drills and real situations in
which HSC crews have been able to put their emergency preparedness training
into practice.
The factors which influence performance of the above tasks were identified as
being:
• Skill Levels – These are a direct function of and map onto Training, Experience
and Certification (the elements recognised as inherent parts of Competence).
• Service Resources – These are external to the ship and crews, and include
shore side berthing crews, booking and vehicle loading staff, maintenance
personnel etc
• Manning and Work Organisation - This factor includes the provision of sufficient
personnel to do the tasks required at the appropriate skill level, the organisation
of those tasks among different ship’s personnel and between ship and shore
crew and the scheduling and rostering of work to ensure that staff do not suffer
from fatigue and enjoy a socially acceptable life style.
• Equipment Design - This concerns the design and function of any single piece
of equipment.
• Time Pressure may be short term (eg. the requirement to process navigational
information and take collision avoidance decisions within seconds) or long term
(eg. too many tasks for the allotted resource or "not enough hours in the day").
The former problem tends to be a function of personal performance and the
latter, of Work Organisation.
• Stress in this context correlates with high levels of personal performance, ie. if
stress levels are high, the individual may give less than best performance. The
causes of stress may be straightforward, eg. a requirement to process
information and take decisions in too short a time span, or more subtle, eg.
social and family pressure brought about by working antisocial hours or shift
patterns.
The above factors were later used as input to the Influence Diagram definitions in
Step 2.
Experts within the project team on each of the accident categories developed a
description of the technical and operational issues known to be of current concern
to HSC regulators, designers, constructors and operators in order to establish the
state-of-the-art for the project team as a whole and to establish the current base
line benchmark for the risk assessment. A summary for each accident category is
given below.
Obviously collisions may occur in either confined waters or open sea. Confined
waters, by their very nature, tend to be busy as they are usually port approaches,
rivers, estuaries, canals etc. in which the traffic must come into closer proximity
than it would in the open sea. Contacts generally occur in confined waters.
One of the main uses of the Generic Model described above is to provide sufficient
information for the Accident Categories to be broken down into their relevant
Accident Sub-categories. In the case of collision it is likely that consideration of the
environment will be of primary importance, in particular the generic locations of an
HSC. Thus collisions in specific locations such as docks, berths, harbours,
coastal waters and open sea were considered. Other considerations include
operational phases which are covered under the Managerial heading in the Generic
Shipping Model. Thus collisions during such operations as starting, manoeuvring,
berthing, anchoring, shutting down, dry-docking and maintenance were also
considered.
The main differences between HSC and conventional ships are that HSC travel at
approximately twice the speed, are of lightweight construction using aluminium or
composite materials for main hull and superstructure construction and have
significantly higher power to weight ratio propulsion systems. Also HSC are
regulated using a different approach from that for conventional shipping in the
limitation of operational sea conditions and the need for more advanced evacuation
and rescue arrangements.
The following aspects of the generic HSC were considered in relation to likelihood
and consequences of collision and contact:
Fire
The general approach to design for fire is to compartmentalise the ship based on
the compartmentalisation occurring for other reasons (cargo hold bulkheads,
separation between machinery spaces and other areas, deck separation, damage
stability etc). The boundaries of each compartment are then designed to prevent
spread of fire from one compartment to the next. The approach to boundary design
has in the past been to select a boundary construction which gives a known
performance in a standard fire test. The fire test is based on the type of cellulosic
fires which typically occur in buildings which assume (relative to a hydrocarbons
fire) a gradual temperature rise with the temperature at 60 minutes being 945EC.
In contrast a hydrocarbon fire could reach this temperature almost instantaneously.
The advent of composite materials which give the lighter weight for HSC, together
with the promise of low maintenance, have moved technology away from the more
traditional approach of using "non burning materials" such as steel. HSC are now
constructed with products which are inherently combustible but are engineered to
give their fire restricting characteristics.
One of the most notable HSC fires resulting in the total destruction of the vessel
originated from an electrical fire understood to have started in a void space.
Thankfully due to the immediate availability of numerous leisure craft, rescue of all
personnel was achieved without loss of life but, in a different location, a major
catastrophe could have occurred.
Fires which develop large amounts of smoke (as most do) in a contained area
make identification of the source and its extinction very difficult. Unmanned and
effectively closed compartments can be flooded with inert gas, presenting an
opportunity to control escalation, even for large initial inventories, which is not an
option for occupied areas.
The following aspects of the generic HSC were considered in relation to fire:
It appears that the consideration of fire in the design of HSC is largely driven by the
regulations and as yet few designs embody the HSC Code requirements. The
HSC requirements represent a half way house between design based on a
standard fire test and that based on fire engineering principles which would now
typically be used in hazardous industries and offshore platforms and are seeing
increasing use in building structures. Fire engineering comprises the estimation of
realistic heat flux-time-temperature relationships for the realistic fire load for the
space under consideration and the evaluation of the boundary structure subjected
to such heat flux-time-temperature relationships.
• The standard fire test curve may not represent the actual fire conservatively.
• Engine room fires will almost certainly involve hydrocarbons which will have a
much faster rate of temperature rise than the standard fire test.
• The HSC Code space classification is ambiguous and the classification of fuel
tank compartments as being of minor risk can be compromised by a fire in an
adjacent compartment.
For HSC to operate efficiently at high speed their hulls need to be of lightweight
construction. Thus there is commercial and technical pressure to use minimum
acceptable scantlings and, where possible, low density materials. The individual
properties and characteristics of different material are traded-off in the design
process to arrive at an acceptable compromise between cost and weight for a ?fit
for purpose? structure. In this context, fitness for purpose requires consideration of
adequate strength, stiffness, corrosion performance, brittleness and fatigue
properties of the structure in a marine environment.
The following aspects of HSC design, construction and operation were considered
in relation to loss of hull integrity:
The structural and loading aspects of HSC important to Loss of Hull Integrity can be
summarised as follows:
• Their optimised lightweight structures are less rugged and experience different
loads from traditional vessels
• Non traditional materials are used which require different practices in design
and construction and greater attention to detail
• High speeds lead to high specific impact energies in contacts or collisions.
The following specific aspects of the Generic HSC, which are relevant to Loss of
Hull Integrity were identified:
Historic HSC data were collected from a number of sources (commercial data
bases, regulators and individual operators) and from human factors interviews with
experienced HSC personnel. Although the trial application is concerned with high
speed catamaran ferries, the data have been supplemented with data for all high
speed craft reflecting the similarity in operational characteristics (human and
• Local Authority incidents (ie. those reported to regional bodies such as MAIB,
LMIS, Hong Kong Government, NMD, etc) which are notable incidents and
generally the identity of the vessels involved have been withheld
Where an incident is reported in two or more sources it is retained only under one
category. Category 2 is considered to be the most comprehensive and balanced
dataset. Category 2 therefore takes precedence followed by the Category 1 public
domain dataset. These data have then been related to the corresponding fleet
sizes as described below.
Category 1 incidents are the most widely reported. In the HSC database it is
considered that a significant proportion of such incidents have been sourced, given
the worldwide scatter of incidents under this category which cover most regions of
operation except for USA and Japan. The data in this class have been related to
the world fleet for all HSC as illustrated in the Fast Ferry International Database
(Reference 10). From this it can be calculated that there have been a total of 3888
vessel operational years from 1981 to 1995 corresponding to all but one of the
incident dates within the Category 1 data.
Category 2 incidents are well reported in a limited number of regions. In the HSC
database it is estimated that in excess of 95% of such incidents have been
obtained from UK, Norway, Finland and Hong Kong/China. Therefore, the data in
this class have been related to this limited fleet covering UK, Norway, Finland, Hong
Kong and China. In Category 2, data are drawn from the last 15 years for Norway
and Finland and the last 5 years for the UK and Hong Kong/China. These Category
2 data form the basis of the frequency numbers used in this study.
The fleet size corresponding to the Category 2 data has therefore been calculated
on the basis of new vessel ownership between 1991 and 1995 for the UK, January
1991 and August 1996 for Hong Kong/China and between 1981-1990 and 1991-
1995 in Norway. The vessel operating years total 1093, broken down as follows:
UK/Finland 1991-1995 66
Hong Kong/China 1991-Aug 1996 591
Norway 1981-1990 202
Norway 1991-1995 234
Category 3 incidents were collected in detail for a very limited number of HSC
based on some 15 vessel operating years of data in the 1990s.
It can be seen in Table 3.1 that the two highest Accident Categories in terms of
numbers of incidents are Collision and Contact. However, there is a substantial
difference between these categories in that there are significantly more injuries
from collisions, while contacts tend to be with the berth/pier which are relatively
minor occurrences. Groundings are significantly less frequent, although
groundings on submerged or visible rocks, rather than sandbanks, tend to be the
highest consequence incidents affecting HSC. The other Accident Categories are
less significant in terms of both frequency and severity, however within the
“other/unknown/near miss” category there are a few incidents with a significant
number of serious injuries and one fatality.
The contact data show the predominance of incidents occurring near the berth with
90% of incidents occurring during berthing or unberthing.
No human casualties have been recorded for the Fire and Loss of Hull Integrity
Accident Categories. The data show that 93% of fires (14 incidents) have occurred
in the engine room and that 78% of fires have occurred during high speed passage.
Loss of Hull Integrity incidents have affected most principal components and
appurtenances and almost 80% of these incidents have occurred during high
speed passage.
Considerable effort was expended in ensuring that the Category 2 data were
checked and a number of anomalies were identified and resolved, in some cases
by screening out data where they were found to be unsatisfactory (eg. ownership
outside the operational fleet being considered).
The screened Category 2 data show that the frequency of contacts and collisions
is greater in Hong Kong/China than in Europe. This may be considered to reflect
the congestion in the region. Nevertheless reliance on the Hong Kong data alone
would bias the results, where exclusion may give figures which are
unrepresentative of a worldwide HSC fleet.
There is further contrast between the UK and Norway, there being a lower incident
rate across all accident categories in the latter case if the vessel population is
To assess the sensitivity, incident frequencies are calculated for each Accident
Category as shown in Table 3.3. Although Table 3.3 indicates some variability in
the frequencies, it is important to note that the values generally remain of a similar
order within each category despite significant physical and reporting differences in
the regions. These figures are considered to provide justification for the use of the
screened Category 2 data set with frequency and severity given in the final row of
Table 3.3.
1 2 3
Incident Fata Major Minor Incidents Fata Major Minor Incident Fata Major Minor
s l- l- s l-
ities ities ities
3 Contact 82 27.2% 4 0 2 0 49 0 4 7 29 0 0 0
5 Fire 16 5.3% 2 0 0 0 9 0 0 0 5 0 0 0
6 Flooding 6 2% 5 0 0 0 1 0 0 0 0 0 0 0
8 Machinery / 15 5% 4 0 0 0 6 0 0 0 5 0 0 0
Electrical Failure
9 Payload Related 0 0% 0 0 0 0 0 0 0 0 0 0 0 0
10 Procedural 0 0% 0 0 0 0 0 0 0 0 0 0 0 0
11 Explosion 0 0% 0 0 0 0 0 0 0 0 0 0 0 0
20 Other/Unknown/Near- 24 8% 4 1 22 10 13 0 0 0 7 0 0 0
Miss
Note: It should be noted that in some instances a detailed examination of the accident reports may result in a different categorisation of accidents from that shown
above especially regarding machinery failures
C705\27\001R Rev A August 1998 Page 3.16 of 3.17
MCA P404 – D9 SUMMARY REPORT
Collision Contact
Location of
Incident
No. of % known Fatalities Major Minor No. of % known Fatalities Major Minor
Incidents Incidents Injuries Injuries Incidents Incidents Injuries Injuries
Table 3.2 HSC Incident Data for Collision/Contact Sub-categories (all data sources)
Table 3.3 Calculations of Incident Frequencies and Incident Severity based on Screened Category 2 Data
10. REFERENCES
8. Erffmeyer R.C., Erffmeyer E.S., & Lane I.M. The Delphi Technique: An
Empirical Evaluation of the Optimal Number of Rounds. Groups &
Organization Studies, Vol 11, 1986.
10. The Fast Ferry Internal Database Version 2.0, Fast Ferry International,
Tenterden, Kent, 1996.