Adapting to increased automation in the aviation industry through performance measurement and training
Barriers and potential
Time: Thu 2020-12-10 10.00
Location: https://kth-se.zoom.us/webinar/register/WN_CyX9VkPURWG07yQdONDZoA, Stockholm (English)
Subject area: Industrial work science
Doctoral student: Johan Rignér , Management & Teknologi
Opponent: Professor Don Harris, Coventry University
Supervisor: Professor (em.) Lena Mårtensson, Management & Teknologi; Docent Pernilla Ulfvengren, Management & Teknologi
The increased use of automation has affected the work on the flight deck. The Single European Sky ATM Research (SESAR), deployed with the purpose to increase the European ATM system performance, identifies automation as a key enabler to increase future system performance. The aviation system is a complex large socio-technical system. The system is affected by internal and external stressors at all system levels. At a work process level of this system, the flight deck represents a Joint Cognitive System. When accidents or incidents do occur, the importance to look beyond the label of flight crew error to understand what happened is widely recognized. As flight safety improves, there are fewer incidents and accidents to learn from, which increases the importance to look at normal operations data for improvement.
The flight crew training environment is increasingly relying on collected data about an individual airline’s flight operational environment and performance. Through airlines’ performance measurement system, a large amount of performance data is collected. However, this data is not in a format immediately useful for studies of neither complex socio-technical, nor joint cognitive systems. In addition, regulatory, financial, and other constraints limit airlines’ use of collected data as well as how they perform training.
The purpose of this research is to increase knowledge about how training content and learning opportunities for flight crew relates to airline performance monitoring and measurement processes, given a highly automated dynamic environment. Against this background, barriers and potential for improvements to support the flight crew for the operation of the highly automated aircraft are identified.
This research has been conducted using a mixed method approach for collecting and analyzing data. The overall research approach is conducted in an applied research tradition. The empirical data in this thesis are primarily based on two research projects, HILAS and Brantare, both with explicit goals of knowledge generation and learning among participating organizations. The results are based on the following methods: 1) System analysis using Rasmussen’s model for a socio-technical system involved in risk management as the framework, to describe the aviation system, primarily with a perspective from the flight crew and their automated work environment, 2) Interviews of pilots, 3) Workshops with groups of pilots and safety office staff, 4) Implementation attempt of a proposed method how to use data and 5) Collection of flight operational data.
Based on Rasmussen’s model of a dynamic socio-technical system, the aviation system of interest ranges from “A single European Sky” to regulators, national legislation to flight operations, training, and the work on flight deck as well as political and financial pressures on the airline. The conclusions drawn from this comprehensive scope is reliant on the author’s domain knowledge acquired from some 30 years of experience in the aviation industry.
Several barriers against the use of performance data for knowledge and learning improvements are identified. The airline monitoring systems are not ideal for specifically measuring automation related problems and flight crew – automation interactions. Due to the already high flight safety levels, new performance measurement processes and activities are neither prioritized, invested in nor explored. When a proposed data-use method was attempted to be implemented it showed difficulties in finding causalities and relationships between available airline parameters. With unclear causality between various parameters recorded and actual outcomes, it is difficult for airlines to use data available as a source for confident training design. This is also the case for the selection of Safety Performance Indicators, that often are outcome based at a high level. More cross-system integration may render the current measurement systems insufficient to understand difficulties and possibilities in the greater aviation system.
Potential for improvement related to the use of data, knowledge and learning are also identified. Flight crew show a high acceptability towards a proposed learning concept based on normal flight data. A greater emphasis of using indicators showing airline adaptability and flexibility is proposed. Also, moving from a scheduled training activity mindset to a wider learning and knowledge management and sharing concept is suggested as a cost-efficient way forward. Increased utilization of normal operational flight data should be used for this purpose and have potential to contribute to both efficiency and safety in aviation.
This thesis contributes to airline performance measurement and flight crew training knowledge. Results from this research is valuable in other highly automated safety critical domains with a high acceptance of performance being measured and analyzed.