Version classiqueVersion mobile
OpenEdition Books

Proceedings of the fourth Resilience Engineering Symposium

Erik Hollnagel
Éric Rigaud
Denis Besnard

A Simulation-based Analysis of “Resilience” in Enroute Air Traffic Control Tasks

Daisuke Karikawa, Makoto Takahashi, Hisae Aoyama, Masaharu Kitamura et Kazuo Furuta


Abstract. Air Traffic Control Officers (ATCOs) routinely maintain safety and efficiency of air traffic operations by their resilient responses to unanticipated variations of a situation. Although resilience can be very essence of ATCO’s skills, it is hard to analyze their resilience for improvement of training program of ATCO trainees because their resilience may be derived from their tacit practical knowledge. The present research has proposes an analysis method of ATCOs’ resilience by using a cognitive system simulation. Through simulation-based experiments, a couple of features of resilience involved in their control strategy of air traffic have been successfully identified. The basic effectiveness of the proposed method for analyses of resilience at the sharp-end of the air traffic control domain has been demonstrated through this attempt.

Note de l’auteur

Note portant sur l’auteur1

Note portant sur l’auteur2

Note portant sur l’auteur3

Note portant sur l’auteur4

Note portant sur l’auteur5

Texte intégral

1 Introduction

1Air Traffic Control (ATC) may be one of typical work domains requiring resilience which involves an ability to respond to regular and irregular disruptions and disturbance (Hollnagel, 2010). Even in normal conditions, fluctuating factors such as weather, wind condition, communication issues with pilots, etc. sometimes produce unanticipated variations of a situation to which Air Traffic Control Officers (ATCOs) must adjust their prescribed plans and performance dynamically. And, in almost all cases, ATCOs have successfully managed such unforeseeable situations and achieved both safety and efficiency of air traffic operations. This fact strongly implies that resilience is very essence of ATCO’s skills. Hence, analyses of ATC tasks from the resilience engineering perspective are definitely meaningful for developing more effective training programme for ATCO trainees and for actualizing more sophisticated human-machine systems in the ATC domain. However, in actuality, it is hard to perform such analyses because ACTOs sometimes cannot give coherent explanations for their decisions based on tacit practical knowledge.

2The present research, therefore, proposes an analysis method of resilience in enroute ATC tasks with using a cognitive system simulation. In addition, some features of ATCO’s resilience are discussed based on the result of the preliminary analysis. The extent of the analysis in this paper remains within individual ATCO. However, the authors believe that skills of individual ATCO have a close connection with organization issues because their expert skills have been cultivated and inherited in the ATCOs’ community.

2 Method

3Our research group has developed a cognitive system simulation of an ATCO based on cognitive task analysis researches of enroute ATC tasks (Inoue et al, 2005; Inoue et al, 2006). The cognitive systems simulation named “COgnitive system Model for simulating Projection-based behaviors of Air traffic controller in dynamic Situations (COMPAS)” has originally aimed at simulating controller’s response behaviors (i.e., adjustment of monitoring and control strategies) to an ongoing situaiton involving uncertannty (Karikawa et al., 2010). Using the installed cognitive model of an ATCO, the COMPAS can detect existing ATC tasks in a given traffic situation (e.g., conflict resolution or in-trail spacing between aircraft, altitude change of individual aircraft) and visualize them as color-coded aircraft symbols on the simulated radar screen and as a time-series graph shown in Figure 1 and 2. The transition of remaining ATC tasks in a certain time frame visualized by the COMPAS can reflect the effect of ATCO’s response, such as workload reduction and proactive risk management, to the situation.

2.1 ATC Task Index

4The COMPAS has equipped a performance index named “Task Level (TL)” (Aoyama et al., 2010) in order to analyze ATC task performance for a purpose of educational support of ATCO trainees. As shown in Table 1, the TL consisting of 4 levels is a performance index based on the task demand of each aircraft. The TL can also be utilized as a task index because each level of the TL corresponds to specific ATC tasks. For example, the Lv.2 of the TL indicates the task demand of altitude change (i.e., issue of climb/descent clearance to a cruise/assigned altitude of the target aircraft). The Lv.3 represents the demand of separation assurance (i.e., conflict resolution or assurance of in-trail separation between multiple aircraft) in addition to one of altitude change. The Lv.4 indicates a risky situation, that is time pressure situation in terms of conflict resolution or collision avoidance of multiple aircraft. Aircraft coming from upstream sectors have various task levels from Lv.1 to Lv.3. However, by completing necessary ATC tasks, the TLs of all aircraft are decreased to Lv.1 until the time they are handed off to downstream sectors. In short, we can recognize the number of and the kinds of remaining ATC tasks in the specific situation from the TL.

  • * The definition of the TL shown in Table 1. is based on Aoyama et. al (2010).

Table 1. Definition of Task Level*

Table 1. Definition of Task Level*

Note ** Each level is dispalyed with this color code on the simulated radar screen and time -siries TL graph of the COMPAS.

2.2 Cognitive Model of ATCO

5The cognitive model of an ATCO installed in the COMPAS is utilized to detect existing ATC tasks in a given traffic situation. The Figure 1 describes basic structure of the COMPAS. As shown in Figure 1, the COMPAS consists of three separated models; External World Model (EWM), Human Interface Model (HIM) and simulated ATCO model (sATCO). The EWM represents situation of the target airspace which includes sector boundaries, airways, aircraft and wind condition. The sATCO is a model of a human controller consisting of Situation Awareness Model (SAM) and cognitive agents. The SAM involves not only raw data of the traffic situation (e.g., aircraft’s speed, altitude, route) but also future projection by the sATCO based on obtained external information and stored knowledge. The SAM has to be updated by sATCO’s information acquisition from the HIM which represents a radar screen of a controller’s workstation because the sATCO cannot access information directly from the EWM in reality.

6In order to simulate quasi-parallel and adaptive task execution by an ATCO, Multi-Agent System (MAS) architecture has been adopted to implement ATCO’s cognitive model in the COMPAS. The MAS is a system which consists of multiple interactive agents. An important characteristics of the MAS is decentralization which means any central command agent does not exist in the system. Although one agent has simple function and local information, intelligence as a whole system such as adaptive behaviors or ability to solve complex problems can emerge through interaction among agents. Based on this MAS architecture, ATCO’s cognitive functions are implemented as an assembly of various agents. Each agent has a specific cognitive function such as information acquisition from a radar screen, execution of communication with a pilot, store of schematic knowledge. Those agents activate each other, and the activation levels of agents determine the overall behaviors of the sATCO. At the same time, cognitive activities of the sATCO (i.e., operations to the external world and internal cognitive processes by agents) are limited by multiple cognitive resources; visual, auditory, cognitive and motor resources.

Figure 1. Basic Structure of the COMPAS

3 Analysis & Results

7The authors have conducted a preliminary analysis of ATCO’s resilience through a simulation-based experiment using the COMPAS. The working hypothesis of the analysis is that ATCOs can intuitively select a resilient control strategy of air traffic for safe and effective traffic management based on their tacit practical knowledge.

8In order to validate this hypothesis, a traffic scenario and two different control strategies for the given traffic situation have been prepared. ATCOs have evaluated one of the control strategies as the better solution than the other one in the given situation. The purpose of the simulation-based experiment described here is to validate the objective rationality of the evaluation indicated by the ATCOs using the COMPAS.

3.1 Traffic Scenario

9Figure 2 shows the traffic scenario used in the simulation-based experiment. The traffic scenario has been developed based on a couple of typical traffic patterns in the Kanto-North sector in Japan. According to the regulatory requirements, the target altitude of BBB542 which is 13 000 feet has to be achieved by the BBB542 pass through the TLE point. The other two flights, AAA573 and AAA736, have to be controlled so that they can reach their cruise altitudes within this sector.

10The original flight planed route of AAA736 is indicated by dashed-dotted line in Figure 2. However, in this case, it is inefficient to follow the original planed route because it can lead to conflict between descending BBB542 and climbing AAA736 near the GOC point. ATCOs often reroute aircraft in order to resolve a conflict effectively in such a situation.

Figure 2. Traffic Scenario

11For the given traffic situation, two possible control strategies were prepared. The strategy 1 makes AAA736 shortcut to the prior fix (CHINO) directly. The strategy 2 leads AAA736 to west by radar vector and, after the conflict with BBB542 is cleared, directs AAA736 to the CHINO point. Although both strategies can resolve the conflict between BBB542 and AAA736 in advance, ATCOs has evaluated the strategy 2 as the better solution in the given traffic scenario.

3.2 Results

12The simulation results by the COMPAS are shown in Figure 3 and 4. In the Figure 3(i) (ii), the TL is overlaid on the horizontal trajectory of each aircraft with color code. The Figure 4(i)(ii) are time-series graphs of the TL of three flights.

13The results have demonstrated that the strategy 1 has led to continuous Lv.3 by posing another conflict between AAA736 and AAA573 around point (a). On the other hands, the strategy 2 could successfully resolve not only a conflict between AAA736 and BBB542 but also another one between AAA736 and AAA573 by moving crossing point to northern point (b) where AAA573 is certainly expected to reach high enough altitude to maintain vertical separation with AAA736. In addition, the conflict between AAA736 and BBB542 has been cleared in the earlier time frame when the strategy 2 was adopted.

Figure 3. Simulation Results (Horizontal Trajectories)

14These simulation results have demonstrated a couple of important features of resilience, i.e., proactive risk reduction and workload management. By adopting the strategy 2 which ATCOs have evaluated as the better strategy in the given traffic scenario, the potential risk of loss of separation between AAA736 and AAA573 has been prevented in advance. The strategy 2 has also led to an earlier completion of ATC tasks for BBB542 by resolving the conflict of BBB542 and AAA736 in earlier time frame. These resilient responses can also lead an effective reserving of cognitive resources to manage unexpected events especially in high density and complex traffic situations. The COMPAS-based analysis has successfully demonstrated the rationality of the evaluation of control strategies indicated by the ATCOs.

Figure 4. Simulation Results (Time-Series Graphs of the TL)

4 Concluding Remarks

15The present research has proposed an analysis method of resilience in enroute ATC tasks with using the COMPAS, a cognitive systems simulation. Through the preliminary simulation experiment, a couple of features of resilience involved in a control strategy selected by ATCOs have been successfully identified. This result has indicted the basic effectiveness of the proposed method for analyses of characteristics of resilience at the sharp-end of the ATC domain.

16ATCOs have to perform rapid and intuitive decision making under time-limited conditions in the actual operation. However, for ATCOs, the adoption of fixed rules or patterns which describe how to handle a specific situation may be not effective to achieve quick and appropriate decisions because air traffic situations should be managed by ATCOs are too variant and complex. Nonetheless, they demonstrate sophisticated resilient responses to changing situations. The authors have considered that this fact implies the possibility that the tacit practical knowledge of ATCOs might work as a kind of flexible envelope leading ATCOs to resilient responses. Further detailed analyses of ATCOs’ practical knowledge from the resilience engineering perspective are planned for our future work.


17This research was supported by the Program for Promoting Fundamental Transport Technology Research of Japan Railway Construction, Transport and Technology Agency and Grant-in-Aid for Scientific Research (B) 21310103 of Japan Society for the Promotion of Science.



Aoyama, H., Iida, H. & Shiomi., K. (2010). An Expression of Air Traffic Controller’s Workload by Recognition-Primed Decision Model. 27th International Congress of the Aeronautical Sciences (ICAS 2010-11.10.2), September 19-24, 2010, Nice, France Hollnagel, E. (2010). The Scope of Resilience Engineering. In E.

Hollnagel (Ed.), Resilience Engineering in Practice (pp. xxiv-xxxix). Farnham, UK: Ashgate Publishing.

Inoue, S., Aoyama, H., Kageyama, K., & Furuta, K. (2005). Task Analysis for Safety Assessment in En-route Air Traffic Control. 13th International Symposium Aviation Psychology (pp. 253–258), April 18-21, 2005, Oklahoma City, USA

Inoue, K., Ando, H., Aoyama, H. & Yamato, H. (2006). A Research on Task Analysis System for Enroute Air Traffic Control. The Eighth International Conference on Probabilistic Safety Assessment and Management (PSAM-0254), May 14-18, 2006, New Orleans, USA

Karikawa, D., Takahashi, M. & Aoyama, H. (2010). Performance Visualization in Air Traffic Control using Cognitive Systems Simulation. 27th International Congress of the Aeronautical Sciences (ICAS2010-11.10ST1), September 19-24, 2010, Nice, France


1 Tohoku University, 6-6-11-808, Aramaki-Aza-Aoba, Aoba-ku, Sendai, Japan

2 Tohoku University, 6-6-11-808, Aramaki-Aza-Aoba, Aoba-ku, Sendai, Japan

3 Electronic Navigation Research Institute, 7-42-23 Jindaijihigashi-machi, Chofu, Tokyo, Japan

4 Tohoku University, 6-6-11-808, Aramaki-Aza-Aoba, Aoba-ku, Sendai, Japan

5 The University of Tokyo, 7-3-1, Hongo, Bunkyo- ku, Tokyo, Japan

Notes de fin

* The definition of the TL shown in Table 1. is based on Aoyama et. al (2010).

Table des illustrations

Titre Table 1. Definition of Task Level*
Légende Note ** Each level is dispalyed with this color code on the simulated radar screen and time -siries TL graph of the COMPAS.
Fichier image/, 116k
Légende Figure 1. Basic Structure of the COMPAS
Fichier image/, 108k
Légende Figure 2. Traffic Scenario
Fichier image/, 112k
Légende Figure 3. Simulation Results (Horizontal Trajectories)
Fichier image/, 328k
Légende Figure 4. Simulation Results (Time-Series Graphs of the TL)
Fichier image/, 99k

© Presses des Mines, 2011

Conditions d’utilisation :