A Human and System Error Approach to Aviation Accident Prevention

Creating Safety from Factors Theory How to prevent system and human errors to gain more safety in aviation? However much is accomplished in terms of safety and operational improvements in aviation, the work is never done. As any other form of transportation, the prospect of having a 100% safety record in aviation is virtually impossible. Nevertheless, getting as close to 100% as possible is the challenge the aviation industry has taken. To study the topic of safety in terms of human and system errors, this thesis will present a theoretical framework of the topic and then carry out primary research in the form of a questionnaire and a simulation. As the primary research is still in progress, a brief overview of the experimental designs can be given but results or conclusions cannot yet be presented. What is Factors Theory? Basically the term Factors is often used without defining what is understood by it, although bordering the topic is very important as there are different understandings. On the one hand Factors are understood as just factors referring to humans and on the other hand Factors are defined by the interaction of humans with socio-technical systems.[1] The latter point of view is similar to Ed Hutchins' theory of distributed cognition, that takes as its unit of analysis how information is propagated through a system in the form of representational states of mediating structures that include internal as well as external knowledge representations within a culturally constituted functional group rather than an individual mind. Within my research I also go along with the definition of the term Factors that is, widely spoken, based on the interaction between humans and systems and incorporates the methods and principles of the behavioural and social sciences, engineering, and physiology as shown in figure 1. It is a key of improving aviation safety to analyze accidents, incidents and other occurrences using human factors models. Several statistics show that the percentage of accidents in aviation attributable to human factors is a relatively constant 60% to 80% almost independent of pilot experience or types of aircraft flown. factors methods and tools are seldom used in the analysis of accident/incident databases although they can provide a useful background and/ or guidelines. Such methods and tools can help find the unobvious reason(s) behind accidents or incidents, which are most often caused by the co-occurrence of several events. Well known models in the field of Factors are, for example [2]: SHEL model by Edwards (1972) expresses the interrelationships and interdependencies of different systemic components on the one hand and the human component on the other hand under the influence of environmental conditions. Edwards' observations are that socio-technical systems are always shaped by the three components S(oftware) H(ardware) L(iveware) embedded in an E(nvironment). Any change of a component impacts the others. When introducing e.g. a certain technology into a working environment, this has, consequently, an implication to the people working in the environment and to the processes in this environment. The model was further developed by Hawkins (1984) and turned into the SHELL model adding a second L(iveware) component. Swiss Cheese model by Reason (1990) [3] states that an accident occurs when one chain link within the production process breaks down. If there is a breakdown of one component the whole system becomes vulnerable which can lead to instability in the system or in the worst case to a catastrophic failure. According to the model each system consists of several layers and if a failure (represented by a hole) in one level occurs it leads to errors in the following layers. Because of this logic, accident investigators should analyze all facets and levels of a system to grasp the total cause of an accident. A weakness of the model might be that it does not take into consideration how and because of what reason failures originate. To bring more clarity into the occurrence and definition of the holes another model has been developed. HFACS, The Factors Analysis and Classification System by Shappell (2000) [3] describes four layers of failure with its details based on analyzing hundreds of accident reports. As named in fig.3 the four interactive levels of failure are: unsafe acts, preconditions for unsafe acts, unsafe supervision, and organizational influences, which prevent an effect framework based on theory and practice for accident investigation. Originally developed for military purposes, the development of the HFACS framework has proven to be a valuable first step in aviation safety programs. To this experience from theory and practice I will add my research, mainly referring to two of the four levels of HFACS, which are unsafe acts and preconditions for unsafe acts. The main research questions within these studies are: Is the pilot (always) aware of what the system is doing? Does the pilot trust the system? How sufficient is the information in airplane cockpits as provided by the technical systems (alarms, computers, displays, sensors, software), the paper checklists and the manuals? What impacts the pilot's workload? Methodologically the research includes a questionnaire that has been specially designed to evaluate human factors in relation to cockpit systems as well as a simulation that will analyze situation awareness / anticipation during system failure(s). While most existing questionnaires aim to evaluate ergonomics, this one puts more of an emphasis on the pilot's, the human view and interaction with the system, where cognitive factors such as perception, information processing, decision making and action taking play a critical role. The questionnaire Human Machine Interaction in Airplane Cockpits therefore consists out of two general sections referring to biographical and experience data and six special chapters including checklists/ manuals/ emergency procedures, comprehension, design, training, trust and workload. This construction of the survey with its 105 questions will provide a wide overview over several factors (mentioned above) from pilots' subjective points of view. Before creating the content of the survey, wide-ranging literature research was carried out in order to gain knowledge and get a good overview of the topic. Concerning technical aspects, the supervision via the German Aerospace Center improved the survey substantially as did personal consultations with pilots, both of which allowed the survey to be revised and improved several times. The final version of the questionnaire was then created in Limesurvey and hosted on the German Aerospace Center's server where it will be available to pilots. The deadline for filling out the questionnaire is the end of July and afterwards SPSS will be used to evaluate the survey. The questionnaire aims to provide the basis for a human and system error approach to aviation accident prevention and can be seen as a basic module for further (also more detailed) investigations into this field. The second component of the primary research carried out for this thesis will be a simulation to analyze situation awareness (SA)/ anticipation during system failure(s). As the experimental design is still in progress, so far only very general information can be provided. SA is apparent and relevant in everyday life and creates the basis for information processing, decision making and action taking but it is very difficult to study outside a laboratory without risk. Simulations are therefore essential for SA research, especially when it comes to complex and critical tasks such as system failure(s) in the cockpit. The goal of the simulation will be to test several pilots in a flight simulator in a specially-developed scenario and then evaluate their SA. There are several applied SA frameworks such as SAGAT, SART, SARS, etc. which could be used to assess the data. As the simulation is still being designed, the most appropriate data evaluation framework has not been chosen yet. In sum, the evaluation of the questionnaire and the simulation, these analyses should add knowledge to the field of aviation safety and help the industry take a small step forward towards the elusive target of 100% safety. References: [1]Badke-Schaub, P.; Hofinger, G.; Lauche, K. [Ed.] (2008): Factors, Psychologie sicheren Handelns in Risikobranchen; Heidelberg: Springer; p.4; p.7 [2]Garland, D.J.; Wise, J.A.; Hopkin, V.D. (1999): Handbook of Aviation Factors; Mahwah: LEA; p.595 [3]Wiegmann, D.A.; Shappell, S.A. (2003): A Error Approach to Aviation Accident Analysis, The Factors Analysis and Classification System; Aldershot: Ashgate; p.47-70