Kristiansen Svein. Maritime Transportation: Safety Management and Risk Analysis
Подождите немного. Документ загружается.
Another key point in Reason’s understanding is the focusing on so-called latent
failures that are difficult to detect before the accident is a fact. These factors are termed
general failure types (GFTs) and can be related to the organizational processes as
illustrated in Table 13.8. It is a primary objective of inspections and audits to detect the
GFTs before they cause harm.
13.10 OFFICIAL ACCIDENT INVESTIGA TIO N
All Flag States investigate, analyse and report maritime accidents. The degree of effort
and procedures may, however vary considerably. The responsibility is often given to
a body or agency separate from the maritime authority in order to secure independence
and legitimacy. The principles for the investigation are laid down in laws and regulations.
It is beyond the scope of the present discussion to elaborate on all the aspects
Figur e 13.10. Feedback loops (after Reason,1990).
Ta b l e 1 3 . 8 . The relationship between basic systemic processes and
general failure types
Processes GFTs
Statement of goals Incompatible goals
Organization Organizational deficiencies
Management Poor communications
Design Design failures
Poor defences
Build Hardware failures
Poor defences
Operate Poor training
Poor procedures
Poor housekeeping
Maintain Poor training
Poor procedures
Poor maintenance management
Source: after Reason (1997).
386 CHAPTER 13 AC CID ENT ANALYS IS
of an official investigation. A few common aspects will, however, be highlighted (Caridis
et al., 1999):
. In case of an accident the Master is obliged to report to the relevant authority.
. Which accidents to report is determined mainly by the degree of seriousness.
. In certain countries the first witness testimonies are given in a formal Sea Court
proceeding.
. Further investigation is undertaken by a single inspector, a team of inspectors or an
ad hoc appointed commission.
. The scope of the investigation is determined by the seriousness and potential of
learning.
. Both the investigation and the report may either be open to the public or confidential.
. Legal proceedings (criminal prosecution) may be a part of the investigation or it may
be performed separately by another body.
. All countries keep files on closed investigations and a few produce statistics and keep
database records.
13.10 .1 The Investigation Pro ce ss
Maritime accidents may be investigated by the authorities, the ship manager, and other
affected parties such as the insurance. The official investigation of accidents is undertaken
by the national maritime administration or an affiliated agency and the conduct is
regulated by law. Caridis et al. (1999) have documented a significant variation in practices
within the EU.
The criteria for undertaking an official investigation are pr imarily related to:
. Certain casualty types that are specified
. Loss of life
. Serious damage and loss of vessel
. Environmental consequences
. Expectation that the findings will contribute to improved understanding
These criteria also vary from one country to another. The investigation may be undertaken
by a single investigator or a team:
. A simplified or admini strative investigation mainly based on documentation
. An investigation by a professional investigator – either with nautical or engineering
background
. By a multidisciplinary team
. Or by an ad hoc appointed commission
The investigative body in the UK, the Marine Accident Investigation Branch, have
published their policy and main procedures (MAIB, 1998).
13.10 OFFIC IAL A CCIDEN T IN V ESTI G ATI O N 387
The emerging view that accidents are rooted in human and organizational
errors (HOE) has led to a greater focus on work psychology , ergonomics, social aspects
and company policies. It has been pointed out that our present knowledge of the
HOE factors is limited due to lack of focus in accident investigations. The
Transportation Safety Board of Canada (TSB) has been concerned about the effect of
fatigue and has recommended that the following factors should be determined in the
investigation:
. The time of day of the incident
. Whether the operator’s normal circadian rhythm was disrupted
. The number of hours the operator was awake
. Whether the sleep history 72 hours prior to the incident suggests a sleep debt
It has been raised as a problem that interference from police and prosecution may
hamper the search for the causes of an accident. Some nations have therefore separated the
two tasks, for instance the US, where both the US Coast Guard (USCG) and the Nation al
Transportation Safety Board (NTSB) are involved.
Considerable experience has now been accumulated on how such investigations should
be undertaken (Ferry, 1981; Bird and Germain, 1985). The key activities are:
. Planning and organization of team and activities
. Witnesses
. Investigation of vessel and systems involved
. Medical examination
. Background information and documentation
. Examine records for the operation
. Test of material and analysis of technical failure mechani sms
. Analysis and documentation of what happened
. Analysis of human error
. Analysis of basic or root causes
. Summary of lessons to be learned
. Reporting, comments and feedback
The US Coast Guard (USCG) has developed a comprehensive set of procedures and
techniques for accident investigation (USCG, 1998a, 1998b). Despite this, Hill and Byers
(1991) made a critical assessment of the accident reporting system of the USCG. They
found a number of weaknesses as to the reporting of so-called human errors, and the
following explanations were suggested:
. Inexperience of the investigators with human error concepts
. Tendency to carry mental models which leads to over-representation of favourite
causes
. Conflict with other investigation objectives
. Lack of adequate models
388 CHAPTER 13 AC CID ENT ANALYS IS
13.1 0.2 The Acci dent Report
The investigation process will usually be documented in a report. The typical sections or
chapters are indicated in Table 13.9. The report may also be supplemented by:
. Photos and drawings
. Source documentation, logs
. Special studies
. Transcri pts of testimonies, hearings, court proceedings, etc.
It is also common to have a standardized form for reporting key information about the
circumstances and immediate outcome of the accident in terms of accident type, degree of
seriousness and immediate consequences. The form will have an even broader scope given
that the accident will be entered into a computerized database. The matter of coding will
be discussed later in relation to accident models and taxonomies.
Official investigations have to some degree been under attack for not being up to date
in terms of organization, technical approach and understanding of causal mechanisms:
. Use of court proceedings
. Lack of teams with broad competence and specialization in modern investigation
techniques
. Narrow focus on technical and nautical aspects
. Lack of competence on human and organizational factors
. Conflict with objectives of criminal prosecution
. Bias: mental models lead to over-represent ation of favourite causes
13.1 0.3 Mar i ne Acci den t In vesti gatio n Branch (M AI B )
In the UK, the MAIB has been very active since the late 1980s in developing better
investigation routines and reporting formats. In 2001 they also revised their computerized
accident database. The investigation policy is outlined in a memorandum (MAIB, 1998).
Based on certain criteria, MAIB may undertake:
. Administrative inquiry for less serious cases.
. Inspector’s investigation of cases that need full and detailed investigation in the field.
. Inspector’s inquiry for major accidents. This will normally involve a team of inspectors
and the report will be published.
The coding of accidents is based on the record structure outlined in Figure 13.11
(MAIB, 2001). The coding starts with identifying data about the case vessel and systems
involved. The systems are further broken down (what), located (where) and injured
persons listed (casualty). The explanat ion of system failures is explained at two levels:
causal factor and subfactor. It may also be noticed that recommendations from the
investigators are coded.
13.10 OFFIC IAL A CCIDEN T IN V ESTI G ATI O N 389
Ta b l e 13 . 9. The accident report structure
Section Outline
Summary Short presentation of situation, key facts, events, causes and
lessons to be learned
Sources of information Persons and parties interviewed or given testimony, transcripts
Investigation technical reports
Accident Accident type, vessel damage, injuries and fatalities,
environmental pollution
Vessel Technical particulars, general arrangement, photos, etc.
Identification
Shipbuilder, year of delivery
Ownership, management, registration, classification
History of operation, incidents, deficiencies, detention
Crew Certificates, competence and experience, nationalities, language used
Watch system, hours on watch before accident
Physical location Day, hour Weather conditions
Geographical location, aids to navigation Visual conditions
Seaway type, docks, moorings Sea state
Port administration, tug service Pilot service, VTMS
Narrative events Description of what happened from initiation through escalation
to accidental outcome
Description of rescue and re-floating, salvage or damage
control operations
Is given in chronological order and supplemented with charts,
diagrams, etc.
Analysis Assessment of facts and evidence
Technical failures, environmental impact phenomena
Description of workplace, instruments, tools and ergonomic factors
Discussion of causal factors related to individual behaviour and
organization
Human factors problems: fatigue, cognitive problems,
human–machine interaction, etc.
Bridge resource management
The role of managing company, port authorities or other
key parties involved
Conclusions Summary of critical events and causal factors
Submissions
Recommendations Point-by-point proposal of controls or safety measures
Appendix Vessel documentation, hull strength analysis, weather and sea state data
Background information, logs
Technical reports, analysis in detail of critical items or events
390 CHAPTER 13 A CCI DENT ANALY SIS
MAIB has a homepage where open publications are accessible:
http://www.dft.gov.uk/stellent/groups/dft_control/documents/contentservertemplate/
dft_index.hcst?n¼5481&l¼2
The following reports are published:
. Annual reports
. Safety digests: accident summaries and lessons learned
. Safety bulletins: safety preventive recommendati ons on the basis of a specific accident
. Safety studies: report on a specific safety problem based on experience gained from a
number of accidents
. Investigation reports
13.10.4 Official Homepages
There is now an increasing trend by the maritime authorities to publish accident
information on the Internet. The US Coast Guard Office of Investigation and Analysis has
the following homepage on accident investigation:
http://www.uscg.mil/hq/g-m/moa/casualty.htm
It covers a broad set of information such as accident statistics, lessons learned, safety
alerts, etc., quite similar to that of MAIB (Figure 13.12).
A similar homepage is published by the Australian Transport Safety Board (ATSB).
Figure 13.13 shows a window from the homepage. The accident reports are present ed
Figure 1 3.11 . MAIB accident database structure.
13.10 OFFIC IAL A CCIDEN T IN V ESTI G ATI O N 391
chronologically on the left side whereas the latest reports are acce ssible directly on the
right side of the window. In the lower section is shown how the search for specific reports
is facilitated by short text abstracts. The homepage also has basic information about
ATSB, reporting routines and safety bulletins.
It is clear that this policy makes a vast information potential available to the maritime
industry, research and other official agencies.
13.11 ACCIDENT ANAL YSIS SOFT WARE
The fact that the ISM Code specifies that companies shall undertake systematic analysis
and feedback of experience from non-conformities has opened a market for computer
software that can support this function. Here we will briefly comment on two systems that
are offered on a commercial basis.
13.11.1 SYNERGI
The SYNERGI system was originally developed for continental shelf operations but has
also won acceptance in the shipping sector. The coding structure is to a large degree based
on the Loss Causation Model (LCM) which has been outlined earlier.
Figure 13.14 shows the main window of the PC-based system. It contain s mainly
factual information and a short text summary (Description of Event). A separate
window has been designed for coding of immediate and basic causal factors (see
Figure 13.15).
Figur e 13.12. Homepage of USCG Office of Investigation and Analysis (http://www.uscg.mil/hq/g-m/moa/
casualty.ht m) .
392 CHAPTER 13 A CCI DENT ANALY SIS
Figure 13.13. Details from the homepage of the Austr alian Transport Safety Board (http://www.atsb.
go v.au / marine / indxf/i nde x.cfm ).
13.11 ACC ID E N T A N A LYS I S S OF T WAR E 393
13.12 HUMAN FACTORS ANAL YSIS
1 3. 12. 1 What is H uman Error?
The term human error is often heard as the explanation of accidents. On the other hand it
is often misused and, as already discussed, not a very clear concept. It might be useful first
Figure 1 3.1 5. SYNERGI: data in Causes w indow.
Figure 1 3.1 4. SYN ER GI main window.
394 CHAPTER 13 A CCI DENT ANALY SIS
to comment on some key findings of Jens Rasmussen cited by Reason (1997) on human
performance as outlined in Table 13.10:
. Skill-based: hi ghly trained tasks that are executed almost automatically and
unconsciously.
. Rule-based: certain pre-trained responses to a set of known situations (if situation then
response).
. Knowledge-based: basically what we understand by problem solving. The new
situation requires analysis or trial and error.
Reason (1990) has further shown that the basic forms of error can be understood
in terms of these performance levels. The basic error types are outlined in Figure 13.16.
First, one should make a distinction between unintentional and intentional error. The
Figure 1 3.16. Error ve rsus perfor mance leve l (adapted from Reason,1990) .
Ta b l e 13 . 1 0 . Performance le vels
Performance Situation Control mode
Skill-based Routine Automatic
Rule-based Trained for problems Mixed
Knowledge-based Novel problems Conscious
Source: after Reason (1997).
1 3. 1 2 HUMAN FACTORS ANAL YSIS 395