Air traffic control complexity and safety: A framework for sector design based upon controller interviews of complexity factors.
ByDr. Arnab Majumdar, Lloyd's Register Educational Trust Research Fellow in Transport Risk Management Leader Air Traffic Management Team, Centre for Transport Studies,
Department of Civil and Environmental Engineering, ImperialCollegeLondon and
Washington Yotto Ochieng, Professor of Positioning and Navigation Systems, Centre for Transport studies, Dept. of Civil & Environmental Engineering, Imperial College London.
ABSTRACT
Air Traffic Control (ATC) complexity can be thought of as the interaction between air traffic and the sector characteristics through which this air traffic flies, and research indicates that this is the prime determinant of the air traffic controller’s workload. There has been considerable literature in the past that have attempted to define and classify air traffic complexity variables, which have in turn been used in to assess the capacity of en-route airspace in Europe and North America. In addition, such variables can be used to assist in ATC sector design by highlighting those complexity factors that can lead to high workload situations as well as potentially risky scenarios for the controller. By using a structured interview technique, this paper analyses these ATC complexity variables for their impact on controller workload and safety. Seventy nine air traffic controllers were interviewed face-to-face as to understand what factors affect their workload and how they do so. These controllers were based in 14 Area Control Centres in Europe, including those of high traffic complexity, e.g. Maastricht, Geneva, as well as in nine centres in Asia and one in Africa, e.g. Mumbai, a region that has seen remarkable traffic growth in recent years.Based upon these interviews a taxonomy of over fifty complexity variables sub-divided into 11 major groupings, e.g. traffic mix, entry and exit points and characteristics of neighbouring sectors have been developed. In addition to rating the impact of the individual complexity factors to assess their impact of airspace, interactions between complexity variables were also deduced from the interviews. By the use of such interactions, it is possible to assess what factors add to layers of complexity for air traffic controllers. Such a taxonomy can assist airspace planners in designing sectors, especially with regard to avoiding complex ATC situations.The paper concludes by outlining how the impact of such complexity factors can be investigated in sector design.
1.Introduction
Air traffic controllers play a vital role in the aviation system by controlling the air traffic under their jurisdiction in a safe and expedient manner. The airspace through which this traffic flies is divided into air traffic control (ATC) sectors, which are defined by both the horizontal and vertical dimensions. Sector design is a crucial aspect of ensuring the safety of the ATC system and the overall aim of this has been to ensure that controllers have a manageable workload, without it becoming too high or too focused on particular airspace or traffic characteristics. Indeed, various studies in Europe and the USA have highlighted that inappropriate sector design is associated with controller caused loss of separation incidents between aircraft1, e.g. (1).
Such studies have indicated what complexity factors can lead to high workload or potentially unsafe situations. Similar complexity factors have been obtained from fast–time simulation studies that analyse the factors that affect controller workload as they impact on en-route airspace capacity in high traffic density areas, e.g. (2).
In recent years there have been attempts to derive methods to analyse and manage the safety aspects of airspace and traffic complexity which may impact on controller performance and in extreme cases lead to losses of separation. In Europe, where traffic complexity is on the rise, this is seen as a risk area2.Methods thus far have focused on examination of those airspace design and traffic characteristics that appear to cause perceived complexity for controllers, and how airspace is designed to compensate for such factors. This work is culminating in a safety management approach for airspace design projects.
However, it has become increasingly apparent that a significant need is the prediction of complexity-safety impacts when using complexity modelling tools, whether fast-time or other modelling approaches. Although taxonomies of complexity factors exist, these are often not broadly-based enough, and it is often unclear how they might interface with fast-time simulation tools which predict workload based on sectorcharacteristics (3). There is therefore a need to develop a consolidated list of compexity shaping factors (CSF) that can impinge on safety, and investigate whether these can indeed be utilised in a predictive fashion for new airspace considerations. If such feasibility exists, then research is needed to integrate such factors into tools that are being developed to predict workload, to ensure that future sector designs remain safe.
One way of ascertaining these CSF that can impinge on safety, especially when related to sector design, is to interview controllers in a structured manner and based upon this obtain a taxonomy of such CSF. This paper outlines a method by which to do this, based upon the results of face-to-face interviews conducted with over 100 controllers not just in Europe, but also in high traffic growth areas of Asia, Africa and Australasia. This represents, to the authors’ knowledge, the first such attempt to determine CSF in the two continents of Europe and Asia.
This paper is organised as follows. Section 2 provides an overview of the theory underlying ATC complexity and safety incidents, before considering the literature on these aspects in Section 3. A method to assess the various complexity variables, and their impact on controller workload and airspace safety, is by the use of a structured interview technique with air traffic controllers. Whilst by this method, it is possible to ascertain these variables, there is a need to carefully frame the interview questionnaire to elicit the maximum amount of appropriate information. This is discussed in Section4, where the methodology outlined for this research is given. Section 5 provides a taxonomy of these variables, together with an explanation of some of them. Given the large list of taxonomy variables, Section 5 also ranks these variables to assess their impact on controllers based upon ratings given by airspace planners. Section 6 analyses the generic differences in CSF between between Europe and Asia. It became apparent during the interviews that controllers rarely considered the impact of these variables in isolation and more likely in terms of combinations of variables. Section 7 provides a taxonomy of two-variable combinations based upon the interviews. Section 8 then discusses two approaches by which to incorporate the CSFs into sector design and the paper concludes with Section 9.
2. ATC complexity and safety occurrences: the relationship
The mental and physical work done by air traffic controllers to control traffic under their jurisdiction is known as air traffic controller workload. Controller workload though is a confusing term with a multitude of definitions, models and measures in the literature (4). For airspace design, focus is given to ensuring safety by the avoidance of high workload for the controllers, whilst ensuring an efficient traffic flow.
Therefore good sectorisation should (5):
1. divide traffic equally between sectors. 2. divide conflicts equally between sectors. 3. minimise the conflict complexity within the sectors. 4. minimise co-ordination workload. 5. avoid exit points for descending aircraft close to a sector boundary. 6. be flexible to cope with short term fluctuations in traffic demand. 7. have a “reasonable” average transit time. 8. provide an effective compromise between instantaneous workload and the airspace required to resolve conflicts. 9. specialise the ATC task wherever possible. 10. have horizontal splits where overflying traffic is dominant and vertical ones where climbing or descending traffic dominates.
Controller workload itself is primarily affected by the situation in the airspace as determined by (6):
• physical aspects of the sector, e.g. size or airway configuration; and • factors relating to the movement of air traffic through the airspace, e.g. the number of climbing and descending flights; and • a combination of the above factors, which cover both sector and traffic issues, e.g. required procedures and functions.
The interaction between sector and air traffic features is a complex process, which can be termed as ATC complexity (6). The nature of complexity is such that the variables involved interact to generate workload and these variables can be thought of as the primary or source factors affecting controller workload.
The impact of these primary factors on controller workload can be mediated by secondary factors that include:
• the cognitive strategies the controller uses to process air traffic information; • the quality of the equipment (including the computer-human interface); • individual differences (such as age and amount of experience).
It is reasonable to assume that as the ATC complexity increases, the workload of the controllers too increases. This rise in workload could have the consequence that controllers make errors in their task of enforcing separation requirements (7).
Figure 1, shows the relationship between ATC complexity, safety occurrences and workload. The following section highlights the major features of the literature on traffic profile and sector geometry as it affects controller caused loss of separation incidents.
3. An overview of the literature
Given the obvious risk posed to air traffic safety by controller caused loss of separations or operational errors, there has been considerable research into the complexity factors that affect loss of separation incidents. Table 1 provides a summary of certain appropriate research on two topics: one investigating airspace incidents with relation to complexity factors, the other on research that models air traffic conflicts. Both these sources provide insights into the air traffic and sector geometry conditions that many prevail during a loss of separation between two aircraft in the en-route environment and thereby assist in the improved sector design for safety. The evidence from these studies indicates that such occurrences, and hence the actual workload associated with it, depends on factors other than sheer traffic volume, e.g. these occurrences happen when the workload and traffic are usually described by controllers as moderate. One such factor could be inappropriate sector design, which could facilitate the co-ordination problems noted in a many studies. In addition, at low workload conditions, the concentration and vigilance of controllers may fall and lead to error occurrence.
Further analysis of the literature reveals the following air traffic related CSF:
• The flight profiles of the aircraft involved in loss of separation incidents, especially with one aircraft at least in vertical transition. • The conflict geometry or encounter angles between aircraft, especially the conflict paths involved, and the speed variances between the aircraft involved. • The number of aircraft in the environment surrounding a conflict pair, especially if there is a cluster of aircraft.
With regards to sector geometry, the following CSF variables appear relevant:
• Sector size, especially a small sector; • The number of crossing points for intersecting traffic flows; • Other sector features, such as edge distance.
In general such studies reveal strong detail about the air traffic CSF at the time of the occurrences, though they are weaker with regards to the sector geometry based CSF. In addition, the interaction between air traffic and sector geometry CSF is not well elucidated in much of the research.
4. Methodology
Sector design aims to ensure that controllers have a manageable workload for their sectors. This should avert unsafe complex situations that require too much overall workload or an overlong focus of workload on one area of airspace or a particular type of aircraft movement. In order to ascertain the complexity factors affecting sector design, it is possible to ask controllers to detail such workload situations. This involves the use of a technique of controller interviews based upon a structured questionnaire. The methodology used for this analysis of CSF variables affecting controller workload and airspace safety is outlined in Figure 2.
Initially, these variables were determined using a literature review, and a series of detailed questions were framed to ascertain the impacts in the following areas:
1. Flight Profiles 2. Speed 3. Entry/ Exit points 4. Routes 5. Flight Levels 6. Surrounding Sectors 7. Intersection points, navaids and reporting points 8. Sector Geometry 9. Restricted areas 10. Bad Weather 11. Any other factors
These questions were reviewed and modified by subject matter experts at EUROCONTROL prior to use. The interviews were conducted face-to-face with controllers in the appropriate area control centres (ACC3s), i.e. en-route centres, in Europe, Asia as well as Johannesburg ACC (the busiest in Africa ) Table 2. In total, 95 controllers were interviewed to ascertain these complexity variables.
The choice of these ACCs is of importance, and those chosen encompassed:
1. Areas of high complexity and traffic density, e.g. Maastricht in Europe and Tokyo in Asia, as well as areas of less complexity, e.g. Oslo ACC. 2. Areas where complexity has increased rapidly in the past decade, e.g. Vienna ACC in Europe and ACCs in India.
Such a selection of ACCs for controller interviews allows for the analysis of both high complexity and rapidly increasing complexity areas, as well as for a major geographical mix of controllers. Indeed this mix of ACCs should ensure that the relevant complexity factors are obtained for safety analysis and to the authors’ best knowledge represents the largest geographical mix of controller interviews recorded.
For each ACC a minimum of two controllers were interviewed to account for the bias that a sole controller interview would have provided. Each interview itself lasted approximately 45 minutes, and was conducted either during controller rest periods or at times when there was little traffic to control.
These interviews provided a taxonomy of over fifty complexity variables in the following categories.
• Traffic Measures • Traffic Mix Measures • Traffic Speed Mix • Entry and exit point Measures • Routes Measures • Intersection points, reporting points • Flight Levels • Neighbouring sectors • Restricted area, Military airspace, Special area • Sector Geometry • Weather • Others
Table 3 outlines the complexity variables, according to the category and as to whether such a complexity variable has previously been outlined in the literature on safety or airspace capacity.
5. Commentary on complexity variables
Most of the variables in Table 3 are self-evident, e.g. the mix of ascends and descends and the number of flight levels. These variables have been stated in other studies too,e.g. (3), and there is a preponderance of such variables relating to the aircraft movement and profiles.
Some of the other variables, especially those relating to sector geometry may not be readily apparent. For example, Figure 3a indicates a crossing point located close to the sector boundary with two entry points. This situation provides considerable difficulty for the controller in both coordination entering aircraft and resolving conflicts between converging traffic. Figure 3b shows the angle of intersection of a route with the sector boundary in both a non-ideal (black line) and ideal mode (red line). In the ideal state, the angle of intersection should be at right angles for ease of controller workload. Figure 3c outlines how sharp-edged sectors can pose a problem for controller workload with some controller difficulties in assessing the distances from the edges to the crossing point. Also, a differential vertical split in two adjacent sectors of airspace can create extra coordination workload for the controller. Coordination with neighbouring countries posed a major workload problem for the controllers especially those with different procedures and technologies. Even with similar procedures and technologies, linguistic differences between controllers when speaking English, added to their complexity.
Whilst weather poses a major problem for controllers globally, the topography of a country plays a major role in this. In countries where sectors overlie mountain ranges,e.g. the Vienna sectors over the Alps or the Johannesburg sectors over the Drakensburg Mountains, controllers find themselves more prone to changeable weather conditions and in summer time to the development of thunderstorms. In addition, it not solely the duration of weather conditions itself that affects controller workload. Controllers highlighted the fact that immediately following a period of bad weather, there tends to be severe workload as aircraft prevented from departing and landing in bad weather request to do so.
CSF unrelated to aircraft movements and sector geometry can also influence controller workload. Some of these factors can be dealt with by the ATC authority of a country, e.g. the quality and range of the radar coverage and radiotelephony (RT) communications within a country. However, there are other factors within a country that are beyond their control, e.g. pilot compliance with controller instructions. Furthermore, there are factors that a country or its ATC authority is powerless to affect. Vienna ACC controllers for example noted that the air navigation charge regime of their neighbouring countries led to an increase in the number of aircraft overflying and undertaking complex movements in Austrian airspace. Section 6 will highlight major differences in CSFs between Europe and Asia.
In order to ascertain the importance of the complexity variables, the taxonomy list in Table 3 was then presented to airspace planners in the following areas of Europe:
• United Kingdom • Ireland • Denmark • Portugal • Austria • EUROCONTROL
The planners were then asked to rate the impact of the complexity variables on their airspace based upon their experience of their airspace in the following marking scheme: 3 for maximum impact; 2 for moderate impact and 1 for minimal impact.
It is apparent from Table 4 that the first eight factors with the highest ratings are those that relate to the aircraft, in particular with the vertical movement of the aircraft having the highest impact. The most important route features relate to the intersection of routes, whether unidirectional or bi-directional.
6. The differences between Europe and Asia for complexity
The controller interviews revealed a number of differences between the European and Asian CSFs. Whilst some were of a localised nature, a number of generic differences were apparent, as outlined below.
a) Flow management
The Central Flow Management Unit (CFMU), operated by EUROCONTROL, is responsible for Air Traffic Flow and Capacity Management (ATFCM) over most of European airspace. As such TFCM aims to ensure the best utilisation of available capacity, the smoothing of traffic flows and protection against overloads. Consequently, flows within and between neighbouring states in most of Europe are well planned out.This is a crucial difference with Asia. The absence of a flow management centre within India, for example, leads to surges in air traffic and other flow related problems for the controllers. And whilst Japan has a flow management centre for its internal traffic flows, a major problem is the lack of flow management for air traffic in east Asia. Local flow restrictions in one nation, lead to further restrictions down stream with ever greater impacts. This fact was noted by controllers at Fukuoka ACC who face the impact of initial flow restrictions placed by Shanghai ACC, added to by Incheon ACC (in South Korea) before being passed downstream to Fukuoka ACC.
b) Coordination
Unlike Europe, the amount and variety of coordination tasks required by controllers in Asia tends to be much greater and such coordination is usually by voice with little automation. Controllers in Mumbai ACC pointed out that they must coordinate with a large number of neighbouring ACCs possessing differing levels of technology and procedures. Thus whilst automatic coordination was possible with one neighbouring ACC, with another there was a need for voice coordination. This differential type of coordination led controllers to perceive added complexity to their work.
Often compounding this coordination complexity in Asia is the need for controllers to change aircraft from RVSM flight levels in their airspace to CVSM levels in an adjoining airspace with a different level of technology.
Coordination is often tinged with a political aspect. A prime example of this is seen with the Incheon corridor, Figure 4. Due to a dispute about airspace boundaries, aircraft flying from China to Japan over Korean airspace face the problem that Shanghai ACC will not communicate directly with Incheon ACC. This leads to the need for controllers at Fukuoka ACC in Japan to conduct the coordination from aircraft between Shanghai and Incheon and thence onto Fukuoka, undoubtedly adding to their workload.
c) Pilot factors.
With regards to pilot interactions with controllers, two major factors were noted. In India, controllers indicated that domestic airlines pilots added to their workload by non-compliance of instructions at certain times. This they felt was related to the requirements of the airlines to achieve reduced delays, and obtain better routings. Controllers feared that such non-compliance would increase given further deregulation of Indian domestic air services.
Japan’s controllers noted that they faced major complexity due to language issues with two major groups of Japanese airspace users, namely Chinese and American pilots. The nature of the complexity though was of different for each group.
With Chinese pilots, problems occurred for Japanese controllers in understanding the English used by the Chinese pilots. This complexity increased if the Chinese pilots were unfamiliar with Japanese airspace regulations.
Similar difficulties in understanding English occurred with American pilots. Here, extra complexity for Japanese controllers was perceived by the use of non-standard phraseology by American pilots.
d) Military airspace
A major feature of the interviews with controllers in Asia was their perceived complexity due to military usage of airspace. Controllers in India and in Japan face a situation where their countries have two nuclear nations in their proximity, and in India’s case is itself a nuclear armed nation. Unlike the situation in parts of Europe, the flexible usage of airspace concept is not prevalent in Asia, leaving controllers with a number of difficulties, including:
i) large volume of military exercises; ii) little warning over the commencement of military exercises; iii) loss of airspace due to military activities; iv) military exercises in areas close to civil aviation routes; iii) coordination difficulties with military controllers.
e) Weather
Whilst weather constitutes a major complexity factor to controllers globally, in Asia the nature of weather faced is different from that in Europe. More extremes of whether were apparent, e.g. typhoons, cyclones and sandstorms. Such weather tended to be of a shorter duration than the fog and snow problems faced in Europe, but of greater impact, e.g. airport closures and heavy damage to aircraft.
7. Combinations of factors
During the controller interviews it became apparent that rather than single CSFs, controllers tend to consider combinations of variables that increase their workload. Table 5 outlines possible combinations of two factors affecting workload based upon the interviews. To all the combinations outlined can be added bad weather situations.The way these combinations interact shows the impact of a restricted area on a narrow sector. This basically amounts to a loss of airspace available for control making the sector even smaller and narrower, with less room for manoeuvre for the controllers. The jutting edge on the left hand of the sector is not ideal since two coordinations are required for aircraft that enter and exit the sector during a short period.Table 6 shows how it is possible to combine more than two factors that affect workload. Consider a sector where there is a long route in the sector, identified previous as a complexity variable. If on this route there is a mix of ascending and descending traffic, a situation of a combination of two CSFs occurs, as outlined in Table 5. Added levels of complexity are given by the route being close to the sector boundary and the presence of intersection points, with a narrow angle of intersection.
8. Approaches to predictive sector design using CSFs
For a particular region airspace, an ATC authority can follow the framework shown in Figure 2 and outlined in Section 4 to determine the complexity factors (i.e. CSFs) for its airspace, both individual and in combination, that impact on controller workload and safety.
The need is there to investigate the impact of the CSF that can impinge on safety in a predictive fashion for new airspace considerations to ensure that future sector designs remain safe. This requires quantitative values for the CSFs, individually, in combination and at different levels, and two possible approaches exist.
One approach is to use fast-time simulation (FTS) and determine the functional links between the CSFs and simulated workload to enable the prediction of high workload unsafe situations. FTS models are discrete-event simulation models that model controller workload given the flow of air traffic through airspace, and is used to estimate airspace capacity in en-route airspace in Europe (7). Research using FTS models has indicated a number of complexity factors that impact workload. An example of this can be seen in the use of panel data techniques to analyse these CSFs (7, 8) and these studies indicate the following variables are statistically significant (20):
• Number of aircraft in continuous cruise; • Number of aircraft in continuous climb; • Number of aircraft in continuous descent; • Number of aircraft in cruise-climb profile; • Number of aircraft in cruise-descent profile; • Number of aircraft in climb-descent profile; • Total flight time; • Aircraft speed mix measure; • Number of flight levels; • Number of exit points; • Coordination measures
As can be seen, most of these CSFs relate to air traffic variables with fewer sector variables. Nonetheless the FTS approach provides a predictive impact of CSFs on controller workload.
An alternative approach is to undertake human-in-the loop simulations, in which the impact of various levels of these CSFs on controller’s performance is investigated. By careful experimental design (9), it is possible to test the impact on controller workload of for example:
• three differing proportions of ascending and descending air traffic flow with two levels of speed mix in the flow; • on a sector of two alternative designs where in the first case the two intersection points are far apart near sector boundaries, in the other they are near the middle of the sector.
Controller workload here can be measured by various methods, e.g. controller subjective methods. By testing various such experimental scenarios, it is possible to ascertain the predictive impact of alternative sector design approaches on controller workload.
Either approach offers a method of incorporating such CSFs into safe airspace design in a situation in which traffic complexity and density are increasing globally.
9. Conclusions
Researchers, especially when analysing the factors that cause loss of separation incidents in controlled airspace, including conflict geometry, have noted the importance of good sector design. Here poor design can lead to a situation where one or more ATC complexity factors impact upon safety.
This paper has outlined a framework to develop a consolidated list of compexity shaping factors that can impinge on safety. This was done by interviewing controllers using a structured face-to-face interview technique to ascertain the complexity factors that affect their workload. It became apparent during the interviews that great value lay in consider not just single complexity factors, but combinations of factors. This leads to consideration of how levels of complexity can impinge upon safety, and how can feedback to better sector design. Based upon this two alternative approaches to how to incorporate such CSFs in sector design were outlined, one by FTS the other using human-in-the-loop simulation.
References
(1) Kinney, G.C., Spahn, J. and R.A. Amato (1977). The human element in air traffic control: Observations and analyses of the performance of controllers and supervisors in providing ATC separation services. Report No. MTR-7655, McLean, Virginia : METREK Division of the MITRE Corporation. (2) Majumdar, A, Ochieng, W.Y. McAuley G, Lenzi, J.M. and Lepadatu C. ' The Factors Affecting Airspace Capacity in Europe: A Cross-Sectional Time- Series Analysis Using Simulated Controller Workload” Journal of Navigation, 57(3), (2004), 385-405. (3) Hilburn, B. (2004) Cognitive complexity in air traffic control – a literature review, EEC Note 04/04, Project COCA, EEC Network Capacity and Demand, EUROCONTROL, Bretigny-sur-Orge, France.
(4) Jorna P.G.A.M (1991) Operator workload as a limiting factor in complex systems, In Automation and Systems Issues (J. Wise and D. Hopkin, eds.), NATO Advanced Study Institute series, Kluwer Academic Publishers, Dordrecht, The Netherlands, 1991. (5) Philips, M. (1999) Testing Operational Scenarios for Concepts in ATM (Phase II) WP2: Airspace Sectorisation Optimisation, Reference: TOSCA/NAT/WPR/2/01 (6) Mogford, R.H., Murphy, E.D., Yastrop, G., Guttman, J.A., & Roske-Hofstrand, R.J (1993). The application of research techniques for documenting cognitive processes in air traffic control. (Report No. DOT/ FAA/CT-TN93/39). Atlantic City, NJ: Federal Aviation Administration. (7) Majumdar, A, Ochieng, W.Y. McAuley G, Lenzi, J.M. and Lepadatu C. (2006) ' The use of panel data analysis techniques in airspace capacity estimation”, Journal of Air Traffic Control Quarterly. (8) Majumdar, A, Ochieng, W.Y. McAuley G, Lenzi, J.M. and Lepadatu C. (2004) ' The Factors Affecting Airspace Capacity in Europe: A Cross-Sectional Time-Series Analysis Using Simulated Controller Workload” Journal of Navigation, 57(3), 385-405. (9) Cox, and Reid (2000) The theory of the Design of Experiments, Chapman & Hall, London, U.K. (10)Grossberg, M. (1989). Relation of sector complexity to operational errors. In Quarterly report of the FAA Office of Air Traffic Evaluation and Analysis, Washington, DC : Federal Aviation Administration.
(11) Rodgers, M.D., and L.G. Nye (1993). “Factors Associated with the Severity of Operational Errors at Air Route Traffic Control Centers,” in Mark D. Rodgers (editor), An Examination of the Operational Error Database for Air Route Traffic Control Centers, Report No. DOT/FAA/AM-93/22, Office of Aviation Medicine, Federal Aviation Administration, Washington, D.C., December 1993. (12) Magill, S.A.N. (1997). Trajectory predictability and frequency of conflict-avoiding action, Paper presented at the CEAS 10th International Aerospace Conference, Amsterdam, the Netherlands, 1997.
(13) Rodgers, M.D., Mogford, R.H. and Mogford, L.S. (1998). The relationship of sector characteristics to operational errors. US department of Transportation, FAA. (14) EUROCONTROL (2001) Investigating Air Traffic Controller Conflict Resolution Strategies (15) Gosling, G. (2002) Analysis of Factors Affecting the Occurrence and Severity of Air Traffic Control Operational Errors, Paper presented to the Annual Meeting of the Transportation Research Board, WashingtonD.C.,USA. (16) Majumdar, A and Ochieng, W.Y. (2003) 'A Trend Analysis Of Air Traffic Occurrences In The UK Airspace' Journal of Navigation, 56(2), 211-229. (17) Ehrmanntraut R. and R. Christien (2005) Analysis of aircraft conflict geometries in Europe, Note, EUROCONTROL Experimental Centre (EEC), Bretigny-sur-Orge, France. (18) Bailey, L.L.,. Schroeder D.J and J. Pounds (2005) The Air Traffic Control Operational Errors Severity Index: An Initial Evaluation, Final Report, DOT/FAA/AM-05/5, Civil Aerospace Medical Institute, Federal Aviation Administration, Oklahoma City, OK 73125 (19) Majumdar, A., Dupuy, M.D., Ochieng, W.Y. and Nalder, P A (2006) Framework for Development of Safety Indicators for New Zealand Airspace: Categorical Analysis of Factors Affecting Loss-of-Separation Incidents, Forthcoming, Transportation Research Records, TRR (20) Oshima, S. (2006) Airspace capacity analysis: study of Nordic airspace using controller workload simulation modelling, MSc. Dissertation, imperial College London, U.K.
Figure 1: The relationship between ATC complexity, operational errors and workload.
Modified from Mogford et al. (1993)
Figure 2: The methodology for determining complexity variables
Step-wise procedure
Literature
Possible Factors
affecting workload
Develop first questionnaire
EUROCONTROL
Approval
Final Questionnaire
Selection of
Control centres and
Interview technique
Interview Controllers
In en-route Area Control
Centres
Taxonomy of
Complexity FactorsInteraction of variables
Affecting workload
Airspace
Planners
Rate TaxonomyInteraction of Variables
Set of rules based
Upon impact on workload
Sector design
Figure 3: Complexity variables relating to sector geometry
Figure 4: Geopolitical coordination in the Incheon corridor, shown by thick red circle
Source: Japan Civil Aviation Bureau (private communications)
Figure 7: Possible methods of incorporating complexity shaping factors into sector design
Step 1: Controller Interviews
Step 2: Identify Complexity factors
Distinguish factors for capacity and safety impacts
(Intermediate step between Step 2 and Step 3)
Step 3: Chosen set of safety complexity factors
Step 4: Human FactorsFast - Time
ExperimentsSimulations
CognitiveFunctional link between
formulation ofcomplexity
link betweenfactors and task-time
complexity factorsworkload
and subjective
workload
Final output: Sector Design and Airspace Planning
Table 1: Summary of the literature on the relationships between OEs and workload
Author and Conclusions
Author:Kinney et al. (1)
Conclusions:OEs occur under low to moderate workload and moderate complexity. In en route centres, 95% of errors are attributed to attention, judgment or communications. Most errors occur in level flight.
Author:Grossberg (10)
Conclusions: Sector complexity factors include control adjustments to merge and space aircraft, climbing and descending aircraft flight paths, mixture of aircraft types, frequent coordination and heavy traffic.
Author: Rodgers and Nye (11)
Conclusions: Most OEs occur with one aircraft in level flight and another descending or ascending. Most moderate errors are between aircraft in level flight. Horizontal, not vertical, separation varies with severity. Higher horizontal separation for SA OEs.
Author:Magill (12)
Conclusions: Most of conflicts involve climb or descent and the combination of a climbing flight with a descending flight accounts for over one third of all conflicts.
Author: Rodgers et al. (13)
Conclusions: Study of OEs in Atlanta ARTCC from 1992 to 1995 revealed that high error sectors had more aircraft, were more complex, smaller in size, had higher workload and were likely to have combined working positions than low error sectors.
Author: Phillips (5)
Conclusions: Developed a conflict complexity measure that incorporated not just flight profile,e.g. vertical mix, and conflict geometry measures, e.g. distance at closest point of approach, but also sector geometry features, such the edge distance to sector boundary where the conflict occurs. Based sector design on “good sectorisation” rules that aim to assure safety and efficiency by reducing controller workload.
Author: EUROCONTROL (14)
Conclusions: Investigated a standardised set of twelve conflict scenarios of varying complexity, which were presented to 45 controllers in seven countries in Europe. Factors influencing controller conflict resolution choice were primarily air traffic variables, e.g. type of aircraft, climb rates, non-nominal flights as well as weather.
Author: Gosling (15)
Conclusions: An analysis of data for OEs occurring in seven US ARTCCs indicated that most OEs occurred under moderate traffic complexity conditions. As the number of aircraft increased, a larger proportion of OEs occurred at higher levels of traffic complexity and the percentage of OEs where the number of aircraft contributed to the traffic complexity also increased. Most severe errors appear to occur irrespective of the level of workload and complexity.
Author:Majumdar and Ochieng (16)
Conclusions: En route and TMA areas in the UK have different traffic factors, workload conditions and phases of flight under which errors occur. Most loss of separation occurrences involving one aircraft in cruise and the other in descent. Those involving both aircraft in cruise are not the predominant category.
Author: Ehrmanntraut and Christien (17)
Conclusions: Considering flights in the Maastricht area of Europe, Only a small percentage of conflicts occur when both aircraft are cruising, 9% below and 18% above Flight Level 180. The 22 distribution of encounter angles and speeds differ depending on attitudes and flight levels: low flight levels have many in-trail conflicts with high speed variances, whereas conflicts at higher altitudes are more equally distributed over angle and converge to a speed-ratio of 1.1.
Author: Bailey, Schroeder and Pounds (18)
Conclusions: Evaluates Operational Error Severity Index formula in the USA. This is computed from data on vertical separation, horizontal separation, closure rate, flight paths, and the amount of control. Determination of whether the event was controlled or uncontrolled is the only variable that requires ATC expert judgment to determine a value.
Author: Majumdar, Dupuy Ochieng and Nalder (19)
Conclusions: Analysed loss of separation (LoS) incidents for the UK, New Zealand and Australia, focussing on the different aspects of the airspace structure and flight profiles involved in the LoS incidents. Authors found that 60% of LoS occurred with at least one aircraft in a descent profile in all three countries. With respect to conflict geometry, majority of LoS occurred when aircraft were following each other or were flying with similar or close headings, followed by the ‘crossing tracks’ configuration and then the ‘reciprocal tracks’ configuration.
Table 2. List of ACCs in which controllers were interviewed about the complexity factors, by Continent, and the number of controllers interviewed.
Asia Mumbai: 8 Delhi: 8 Kolkata: 8 Chenna: 6 Singapore: 2 Tokyo: 10 Fukuoka: 2 Kuala Lumpur: 2 Hong Kong: 2
Africa Johannesburg: 3
Table 3. Taxonomy of complexity variables
Traffic Measures Number of aircraft entering the sector: Yes Maximum number of aircraft instantaneously the sector: Yes Frequency congestion measure: Yes Clustering of aircraft in sector: Yes
Traffic Mix Measures Mix of descends and ascends: Yes Mix of slow and fast moving aircraft: Yes Mix of jet and turboprop aircraft: Yes Mix of Aircraft performance measures
(especially in climb and descend): No Mix of aircraft equipage No
Traffic Speed Mix Speed difference between slow lead aircraft
and trailing fast aircraft on same route: Yes Speed differential at same flight level entry point
between slowest and fastest aircraft in a specified
time period:Yes
Entry and exit point Measures Number of Entry Points (weighted by flights): Yes Ratio of entry/ exit points: Yes Geographical entry and exit points measure: No Clustering of entry and exit points: No
Routes Measures Number of routes in the sector: Yes Route miles flown: Yes Number of intersecting routes in the sector: No Number of routes which change direction: No Number of routes close to the sector boundary: No Proportion of unidirectional to bidirectional routes: Yes Number of intersecting bidirectional routes: No
Route length: Yes
Intersection points, reporting points Number of intersection points in a sector (weighted by flow): No Angle of crossing at intersection point: No Geographical location of intersection points in the sector: No Clustering of intersection points in a sector: No Number of reporting points that aircraft need to report prior to entry into the sector (from Oceanic sectors): No
Flight Levels Number of flight levels used in the sector (weighted): Yes
Neighbouring sectors Number of neighbouring sectors (weighted by flow): Yes Number of surrounding sectors with considerably difficult procedures compared to other neighbouring sectors: No Number of surrounding oceanic sectors:No Number of surrounding sectors with different
procedures required (weighted): No
Flight time through neighbouring sector
(indicates if sector has to undertake actions to prepare
for sector adjacent to neighbouring sector, see below): No Number of sectors adjacent to any neighbouring
sectors whose procedures require special controller
work in the sector in question, e.g. RVSM to CVSM:No
Number of the neighbouring country sectors at capacity: No Incompatible vertical split of sectors with neighbouring country: No
Restricted area, Military airspace, Special area
Location/s of restricted area within sector: No Volume of airspace restricted in the sector: Yes Military aircraft routes crossing civil aircraft routes: No Location of areas within sector in which slow moving aircraft fly: No Volume of areas within sector in which slow moving aircraft fly:No
Sector Geometry Sector volume: Yes Sector shape (difference from regular polygon): No Angles of routes intersection with sector boundaries: No Sharpness of sector boundary edges (angle)
in relation to routes: No Angle of parallel flow to sector boundary: No
Weather Topography of sector (can lead to bad weather situations):No Location of bad weather regions in sector: No Clustering of traffic in sector following end
of bad weather period: No
Others Pilot compliance with instructions: No Pilot experience/ type: No Human machine interface in control room: No Range and quality of radar: No Range and quality of RT communications: No Environment of control room: No Time of peak traffic period, e.g. circadian rhythms: No Policy of the neighbouring country: No
Table 4. ACC Taxonomy variables rated with moderate to high impact
Variable Average Rating Mix of descends and ascends 3.00 Pilot compliance with instructions 2.83 Range and quality of RT communications 2.83 Maximum number of aircraft instantaneously the sector 2.67 Number of aircraft entering the sector 2.50 Frequency congestion measure 2.50 Clustering of aircraft in sector2.50 Mix of slow and fast moving aircraft 2.50 Number of intersecting routes in the sector 2.50 Number of intersecting bidirectional routes 2.50 Location of bad weather regions in sector 2.50 Time of peak traffic period, e.g. circadian rhythms 2.50 Speed difference between slow lead aircraft and
trailing fast aircraft on same route 2.33 Number of flight levels used in the sector (weighted) 2.33 Location/s of restricted area within sector 2.33 Clustering of traffic in sector following end of bad weather period 2.33
Table 5. Two variable complexity combinations from controller interviews.
Variable One Variable Two Mix of descends and ascends Mix of slow and fast moving aircraft Number of intersection points in a sectorAngle of crossing at intersection point Number of intersection points in a sector Geographical location of intersection points in the sector Small Sector volume Volume of airspace restricted in the sector Crossing Bidirectional routes Mix of descends and ascends Large Sector volume Geographically disparate intersection points in the sector Number of Entry Points (weighted by flights) Geographical location of intersection points in the sector Small Sector volume Many entry and exit points Small Sector volume Mix of descends and ascends Long Route lengthMix of descends and ascends Long Route length Crossing points Long Route length Route close to the sector boundary Route close to the sector boundary Number of crossing points Mix of descends and ascends Directions of flow Mix of descends and ascends Number of intersection points in a sector Mix of descends and ascends Number of Entry Points (weighted by flights) Large Number of aircraft entering the sectorLong Route length Large Sector volume Geographical entry and exit points measure Slanted sector edges Crossing traffic Mix of slow and fast moving aircraft Pilot compliance with instructions Converging routes Geographical entry and exit points Mix of descends and ascends Maximum number of aircraft instantaneously the sector
Number of entry and exit points Number of surrounding sectors with different procedures required (weighted).
Table 6. Combination of five levels of complexity
Level
CF1
CF2
CF3
CF4
CF5
Level 1
Mix of descends and ascends
Level 2
Mix of descends and ascends
Long Route length
Level 3
Mix of descends and ascends
Long Route length
Route close to the Sector boundary
Level 4
Mix of descends and ascends
Long Route length
Route close to the Sector boundary
Crossing points
Level 5
Mix of descends and ascends
Long Route length
Route close to the Sector boundary
Crossing points
Angle of crossing
*CF: COMPLEXITY FACTOR
1In the USA, these are known as an operational errors (OE).
2 Based upon private communication with Dr. Barry Kirwan, Safety R&D Coordinator, EUROCONTROL Experimental Centre, Bretigny-sur-Orge, France
3An ACC is known as an Air Route Traffic Control Center (ARTCC) in the USA.
