Abstract
This paper describes a methodology for accident assessment in the frame of a judicial procedure and provide the formal mathematical relations for the fault tree application to the case.
The predictive use of Fault Tree (FT) technique is well known: the best estimate of an undesirable event (the Top Event, TE) frequency is computed from all the basic frequencies of component failures, human errors, and external events.
The FT use is less frequent in “post-dictive” mode: in situations for which the TE is verified, as in the case of a judicial procedure.
In such circumstances, several experts are appointed by the judge (JE) and by the parties, the plaintiffs and defendants (PEs), to reconstruct the event by following a top-down procedure analogous to that necessary for FT construction. They start from the event and try to find out the causes of the event, using all the circumstantial evidences. PEs often form different hypotheses, reaching contrasting conclusions, strongly influenced by the part represented, with JE in a more balanced position.
It is so crucial to adopt a methodology that helps the experts, in particular the JE, to delimit, as much as possible, the area of the uncertainties. FT may be useful for this purpose. In fact, the FT construction allows to incorporate the evidences acquired from the investigations at the level of basic events: the basic events, now, are qualified as “true”, “false” or “unknown” (in the predictive contest, they are quoted through frequencies). In this way, it is possible to select the sequences accountable for the TE.
In this paper a formal development for FT application to judicial procedure is given, together with a real case study. The method can be applied to all kind of accidents, provided that the situation can be described by a fault tree.
Highlights
► Fault Tree (FT) adaptation to forensic and investigation applications is presented. ► We use FT for a real event, in a reverse mode respect to the forecast case. ► This is the case of a judicial procedure, following an incident or an accident. ► Investigation evidences are incorporated in the FT, which is greatly simplified. ► The remaining FT branches are the possible causes of the incident investigated.
Keywords
Fault tree; Risk analysis; Judicial procedure; Accident investigation; Forensic investigation; Forensic engineering
1. Introduction
Let us start from this comparison: “Assign the blame” – vs – “Learn how to improve the system”.
The scope of a judicial procedure is to assess the responsibilities of the accident: to assign the blame. On the other side, it is well known that major improvements of human development come from learning by the errors done. In this case, we try to understand what happened, in order to prevent future similar situations. The scopes are different, although both cases are examples of accident investigations. Actually, according to Levenson (2004), the two previous scopes are “the basic reasons for conducting an accident investigation, etc. When the goal is to assign the blame, the backward chain of events considered often stops when someone or something appropriate to blame is found”. In other words, in a judicial procedure the main goal is not to learn as much as possible, but to assign the blame. This implies that different accident models will be usually adopted, as far as the pursued scopes are different (Svenson, 1999).
However, different accident models may use the same accident investigation methods, i.e. those practical tools “designed with the purpose of helping a specific user (person) to accomplish a specific task (investigation and analysis) in a specific settings. A certain accident investigation method is not necessarily linked to a specific accident model: it can be a tool of its own.” (Katsakiori et al., 2009). A number of accident investigation models and accident investigation methods have been developed in the last decades and the analysis of these is out of the scope of the present paper. Interested reader can refer, for example, to the already quoted work of Levenson (2004), Katsakiori et al. (2009), and to DOE (1999), Kjellén (2000), Johnson (2003), Sklet (2004), and Levenson (2011).
In the case of a judicial procedure, although the general scope is well defined (i.e. the blame assessment), different experts (the expert appointed by the judge (JE) and by the parties, the plaintiffs and defendants, PEs) very often have contrasting scopes; they are not members of a team investigating the facts, but rather competitors. Looking to the list of biases provided by Johnson (2003), it is apparent that the “Sponsor bias”, i.e. the influence due to the risk of damaging the reputation of the expert’s part, is really dominant for the PEs. The experts have definitively different accident investigation models, they have different interests in collecting data and in analyzing them, they (experts & models) “may either act as a filter and bias toward considering only certain events and conditions or they may expand activities by forcing consideration of factors that are often omitted” (Levenson, 2004). Several authors (most of the above quoted, in particular Johnson, 2003; Levenson, 2004) analyze the problem of biasing in accident investigation, not only in the case of legal disputes, and most of the bias comes from the different models. For this reason, as far as in judicial procedure is practically impossible to adopt a common model, the choice of a common, good and simple method may be very effective in reducing bias (Lundberg et al., 2010).
In this paper, the well known Fault Tree (FT) method is reviewed for judicial procedure as a toll managed by the JE in order to collect and to integrate the information provided by the investigations and the reasons – often different, but not exclusionary – provided by the PEs. This use of the FT – as a common, good and simple method to force the reduction of bias due to the different scopes in judicial procedures – is quite new, in our opinion, and turns out to be very helpful in the practice (a case study is presented in Sections 4 and 5). A formal equation for the judicial application of the FT is derived in Section 3.
2. Over the fault tree and its extension to judicial procedures
The use of the Fault Tree (FT) technique has been previously combined with risk analysis and applied to a variety of industrial and non-industrial sectors, such as nuclear energy, aviation and aeronautics, and chemical industry (Fussel, 1976, McCormick, 1981, Vesely et al., 1981, Vose, 2008 and Zio, 2007). As it is well known, the FT can be regarded as the conjunction between qualitative and the quantitative analysis: it is preceded by the phases of the risk analysis aimed at identification of “all” the top events, and it is followed by the phases yielding the TE frequencies. Between these two risk analyses, the FT has to individualize, through the characteristic top-down procedure, the basic events that result in a particular TE occurrence. In essence, the whole procedure attempts to answer the following questions:
(a)
What are the TEs of interest for the system under examination?
(b)
Which events occur concurrently that result in a TE, and how do they occur?
(c)
What is the probability of each top?
(d)
What are the TE’s consequences?
(e)
What is the resultant risk?
(f)
What are the preventive and protective measures for risk reduction?
In any case, in the usual applications, all this is developed in a predictive way: the goal is to determine the probability of having damage and the cause of this potential damage to minimize the risk.
When we consider a judicial procedure following an accidental event, the scenario is quite different; nevertheless, the use of an FT turns out to be very effective.
The major difference is, obviously, that now the TE being considered has already occurred. The question (a) is not interesting, usually.
Following the judicial procedure, experts are appointed to explain what happened and why. These experts are (at least):
–
The technical advisor of the judge (JE).
–
The technical advisors of the parts (the plaintiffs and defendants) (PEs).
The judge’s questions most often attempt to ascertain why the accident occurred, i.e. she/he is mainly interested in point (b), i.e.: “Which events occur concurrently that result in a TE, and how do they occur?”. From this, it follows also “Why, what/who is responsible?”.
Very often, as pointed out in the introduction, contrary to what could be supposed, the experts (at least the PEs) produce opposite conclusions, following the process, described by Jean-Jacques Rousseau, that “marries the convenience with the conviction”, i.e. they tend to demonstrate what the PEs desire. They are strongly biased. Obviously, this results in an augmented confusion, especially high when the technical problem is complicated. The question now is: Is it possible to follow (and to share) a procedure that, even taking into account the differences due to the different PEs point of views, is developed on more rigorous technical basis, reducing the area of uncertainty? Or: is it possible to share a common method, given that the JE’s and the PEs’s models are different and with opposite goals (at least as far as the PEs is concerned) so that the biases are reduced and the different points of view integrated?
This is the field of forensic engineering investigation, which should provide results that will be acceptable under the rules of evidence in a court of law (Noon, 2001, NFPA, 2002 and EPRI, 2003). After the initial introduction of FT analysis in forensic engineering (Johnson, 1985), some important applications have been done. A review of these forensic and investigation applications is provided by Johnson et al. (2006). However, the diffusion of the forensic FT analysis has been quite limited up to now, due to the lack of a specific adaptation of