Techniques and Methods Used in the Study of Maritime Accidents
After a maritime accident, a successful investigation, is the one that will find the causes roots of the accident which is accompanied by valuable teaching for maritime safety and lessons to be learned. For these reasons, the team of investigators must be methodical and insightful, and must use all necessary and possible means; efficient use of data. There is also the mastery of data analysis techniques and its importance in the interpretation of collected data and the realization of plausible sequences of events. In this context, several investigations carried out superficially did not lead to the root causes, which is explained by the redundancy of the same consequences under different scenarios. However, the responsible factors remain unidentified. In this article we will present some techniques and methods used for the study of accidents.
Text Mining Method
This is a method of extracting important information from survey report texts to establish causal links to corresponding concepts (eg electrical fault -blackout); this method allows the concentration of the data and put them in relation with the major accident concepts.
This method can be combined with other text-related methods, such as the one used by the Marine Accident Investigation Branch (MSA) as Vector Space Model (VSM), which is a vector representation of text in matrices.
Analytical techniques differ from one investigating agency to another, as some offices develop their own technique: the case of the office "American Bureau of Shipping" and one type of accident to another depending on the factors to be analyzed (human, organizational and technical). One can even find the use of several techniques in the same accident. It depends on the size of the accident and the complexity of interaction between the factors involved.
Only the techniques and approaches used to deal with technical and socio-technical factors will be considered. In this case, the human factor is not addressed.
Technique of the model SwissCheese- Reason, 1990 & 1997
The model represents a metaphor for illustrating the systemic model proposed by Reason in which the mechanisms that lead to an accident are represented. According to Reason the genesis of an accident is very complicated, we very often discover its origin in the managerial layers and direction of a given structure. Accidents are therefore not the result of a single cause usually located at the operational level, but very often the consequence of a succession and combination of causes that have their origins both at the sharp-end and blunt-end levels that lead to an accident.
This model goes beyond the immediate causes of the accident that are found in the Sharp-end. It provides for the identification of "latent conditions" (holes in gruyere slices) at the blunt-end level and which may contribute to accidents when they combine with one or more failure (s) located at sharp end. Reason also introduced the concept of "In-depth defense barriers". Always from a metaphorical point of view, these defenses are represented by slices of Gruyere, are located at all levels of a given system (individual, local, managerial and organizational). These barriers have the function of preventing, seeing and limiting the propagation of the accidental sequence, and consequently the occurrence of an accident. For that, according to this model, the reinforcement of the barriers of defense (to fill the holes of gruyere) of a given system has a direct effect on the occurrence of an accident. Moreover, the combination of a set of latent conditions (erroneous decisions, poorly defined development strategy, unsuitable safety policy, etc.) and active failures (risky acts, technical failures, etc.) generates an accidental trajectory leading to an accident.
Finally, according to Reason's Swiss Cheeses model, risk and accident prevention consist in introducing and reinforcing the defense barriers at the Blunt-end and Sharp-end levels of socio technical systems.
Failure Modes and Effect Analysis (FMEA)
This method has been developed to evaluate the reliability of hardware systems. Objectively, the method consists in validating a design of a given piece of equipment by enumerating all the causes of possible failures and the behavior of the system for each case. The disadvantage of this method lies in the analysis of each cause separately. Therefore, a complex system must be divided into units with the impossibility of detecting the links between common causes.
Model STAMP from Leverson - 2004
The Systems Accident Model and Processes (STAMP) model is a technique that takes into account the technical part in its two (02) hardware and software components, combined with organizational and human factors. All forms a complex system. Leverson's approach is that the accident doesn’t arise from a failure of an isolated component but from the impossibility of rejection of this disturbance by the entire system through proper manipulation of safety parameters. In this vision, safety is considered a control problem and not a static one.
This technique is Simple, basic and applicable in all types of accidents, is to establish the sequence of events by answering the five (05) questions (When, Where, Why, What, Who). It is effective for technical failures and limited accidents. Used to detect root causes, but based on the investigator's experience with non-use of logic gates.
CFC Causal Factor Mapping Technique
It consists in combining the use of the chronological sequence and the logic of causality ,by attributing to each event its temporal framework. For the construction of cartography the technique proposes the reverse timing starting from the event of the major accident.
- Fault Tree Analysis (FTA)
A technique based on a logic diagram (use of AND and OR gates) to identify the causal relationships between system failure and hazards and undesirable or dangerous events. It is mainly used in the analysis of electromechanical systems.
- Technique relating to physical quantities
1. Model technique
It consists of the use of an equipment model and reproduce the operating conditions and the corresponding environment. Simulations are performed and the phenomenon is observed while recording the physical magnitudes of the system and comparing them to the circumstances of the accident. This technique is widely used in the case of capsizing, using the model of the ship in a wave generating basin.
2. Metro logical tests
These are the tests on the physical properties of the material and we find: the tensile test, bending, compression, hardness, shear, fatigue, friction and others. These tests are used to check the conformity of the building material resistances to the requirements of standards.