EUROPEAN ORGANISATION FOR THE SAFETY OF AIR NAVIGATION EUROCONTROL EUROCONTROL EXPERIMENTAL CENTRE

Transcription

1 EUROPEAN ORGANISATION FOR THE SAFETY OF AIR NAVIGATION EUROCONTROL EUROCONTROL EXPERIMENTAL CENTRE COSPACE CONTROLLER EXPERIMENT ASSESSING THE IMPACT OF SPACING INSTRUCTIONS IN E-TMA AND TMA EEC Report No. Volume I Project AGC-Z-FR Issued December The information contained in this document is the property of the EUROCONTROL Agency and no part should be reproduced in any form without the Agency s permission. The views expressed herein do not necessarily reflect the official views or policy of the Agency.

2

3 REPORT DOCUMENTATION PAGE Reference EEC Report No. Volume I Security Classification Unclassified Originator EEC SSP (Sector, Safety and Productivity) Sponsor EUROCONTROL TITLE Originator (Corporate Author) Name/Location EUROCONTROL Experimental Centre Centre de Bois des Bordes B.P. F 9 Brétigny-sur-Orge CEDEX FRANCE Telephone + () 9 7 Sponsor (Contract Authority) Name/Location EUROCONTROL Agency Rue de la Fusée, 9 B.P.B BRUXELLES Telephone COSPACE CONTROLLER EXPERIMENT ASSESSING THE IMPACT OF SPACING INSTRUCTIONS IN E-TMA AND TMA Volume I Authors Florence Aligne (Thales) Isabelle Grimaud (DGAC) Eric Hoffman Laurence Rognin (Steria) Karim Zeghal Date / Project AGC-Z-FR Pages xviii + Distribution Statement (a) Controlled by EUROCONTROL Project Manager (b) Special Limitations None (c) Copy to NTIS YES / NO Figures Tables 9 Task No. Sponsor AGC-Z-FR- Annexes (separate document) Period References Descriptors (keywords) ADS-B, airborne spacing, arrival flows, ASAS, controller activity, eye movement analysis, real-time experiment, sequencing and merging. Abstract This report presents the results and findings of the CoSpace controller experiment conducted in November. This experiment fitted in with a series of air and ground validation exercises aiming at investigating the use of spacing instructions (denoted airborne spacing) for sequencing of arrival flows. The previous ground experiment conducted in November focussed on E-TMA (from cruise to initial approach fix) using distance based spacing. The present experiment aims at going a step further by introducing time based spacing and integrating TMA (from initial to final approach fix). More precisely, the objective was for E-TMA, to compare the respective impact of distance and time based spacing on controller activity and quality of service provided; for TMA, to assess the usability of spacing instructions. Six E-TMA and four TMA controllers from different European countries participated during respectively and ½ weeks. The simulated airspace was derived from Paris Southeast area, and consisted of two E-TMA and two simplified TMA sectors. As today, the traffic had to be sequenced prior to be transferred to TMA. The level of traffic was high in TMA and very high in E-TMA. The distinct levels of maturity between TMA and E-TMA led to different levels of analysis. In TMA, still at an exploratory stage, the aim was to understand how airborne spacing could be used. This was structured along two dimensions design and feasibility, and initial insight on controller activity. In E-TMA, the aim was to compare the impact of variants (distance versus time) on user activity and on the quality of the service provided. This was structured along four dimensions human shaping factors, controller activity, effectiveness and safety.this experiment enabled to confirm trends obtained in, and to get an overall understanding of impact of airborne spacing on controller activity and on control effectiveness in E-TMA. When correctly used - i.e. fitting in with current sequencing practices - spacing instructions seem to be beneficial (increased controller availability and better stability of flows transferred to TMA). On the opposite, it was observed that incorrect use can lead to degraded situations (increased controller workload). Finally, the spacing instructions developed for E-TMA seem to be usable in TMA.

4

5 COSPACE EUROCONTROL EXECUTIVE SUMMARY This report presents the results and findings of the CoSpace controller experiment conducted in November. This experiment fitted in with a series of air and ground validation exercises aiming at investigating the use of spacing instructions (denoted airborne spacing) for sequencing of arrival flows. The previous ground experiment conducted in November focussed on E-TMA (from cruise to initial approach fix) using distance based spacing. The present experiment aims at going a step further by introducing time based spacing (expected to be more efficient) and integrating TMA (from initial to final approach fix). More precisely, the objective was for E-TMA, to compare the respective impact of distance and time based spacing on controller activity and quality of service provided; for TMA, to assess the usability of spacing instructions. Six E-TMA and four TMA controllers from different European countries participated during respectively and ½ weeks. The simulated airspace was derived from Paris Southeast area, and consisted of two E-TMA and two simplified TMA sectors. As today, the traffic had to be sequenced (Nm or 9s, not catching up) prior to be transferred to TMA. The traffic simulated was derived from a real traffic. The level of traffic was high in TMA and very high in E-TMA. Extended TMA The feedback from the controllers was positive the use of spacing instructions is perceived as useful and potentially leading to more anticipation and to an overall workload reduction. To assess the impact on controller activity and monitoring, the geographical based analysis of instructions and eye-fixations introduced in was reused. It consists in analysing the distribution of instructions and of eye-fixations as a function of their distance to the initial approach fix. As in, the results suggest that spacing instructions lead to anticipate the building of the sequences (earlier use of heading instructions), and to alleviate the controller of maintaining these sequences (fewer speed instructions). An overall reduction of manoeuvring instructions was observed. In addition, most of the fixations are concentrated upstream (before converging area), whereas without spacing instructions, most are concentrated downstream (after converging area). This suggests that, under very high traffic conditions, with airborne spacing, controllers can still concentrate where the sequences need to be built, whereas without, they are no longer able to anticipate the building of sequences, thus being forced into a reactive position. No significant impact of time versus distance could be observed. As an initial assessment of quality of service provided, inter-aircraft spacing and closure rate at initial approach fix were analysed. Airborne spacing enables a more homogeneous flow, with more aircraft getting the optimal spacing value, and fewer cases of catching up situations. Once again, no significant impact of time versus distance could be observed. Project AGC-Z-FR - EEC Report No. Volume I v

6 EUROCONTROL COSPACE TMA An initial assessment of the usability of spacing instructions in TMA was made through controller feedback and observations. The spacing instructions developed for E-TMA seem to be usable in TMA, in particular for the integration of flows onto final approach. However their use implies changes in working methods use of standard trajectories (as opposed to radar vectoring), final integration to be performed on a point (as opposed to on an axis), and unique approach control position (i.e. initial and intermediate positions not de-grouped). The controllers mentioned that airborne spacing can give them more availability, provides but also requires anticipation. In addition, it allows to smooth the traffic but gives the feeling of having less pressure and hence of losing capacity. They preferred the time based spacing, in particular due to the progressive reduction of the distance between aircraft, but stressed the issue of having consistent criteria for wake turbulence. Controller reluctance to cancel airborne spacing led them to treat groups of spacing instructed aircraft as single entities, and integrate them accordingly. The traffic load might have an influence on the use of airborne spacing some airborne spacing initiated by E-TMA might have to be cancelled which would increase workload. Their main concern was the recovery of abnormal situations and the need for clear fallback procedures. Despite this concern, it was observed that, sometimes, they decided to use spacing instructions although they were aware that the applicability conditions might not be respected, and that there was a risk that the spacing would not be maintained. As an initial investigation of the impact of airborne spacing on the controller activity, the geographical based analysis of instructions introduced in E-TMA was re-used. It was observed that without airborne spacing, most of the heading (and speed) instructions are issued near final approach fix, whereas with airborne spacing, instructions seem to be issued all over the sector with fewer late instructions. This analysis suggests a positive impact on controller activity airborne spacing seems to partly relieve the controller of issuing late vectors for integration onto final approach, while providing anticipation. A global reduction in the number of instructions can also be observed. CONCLUSION This experiment enabled to confirm trends obtained in, and to get an overall understanding of impact of airborne spacing on controller activity and on control effectiveness in E-TMA. When correctly used i.e. fitting in with current sequencing practices spacing instructions seem to be beneficial (increased controller availability and better stability of flows transferred to TMA). On the opposite, it was observed that incorrect use can lead to degraded situations (increased controller workload). Finally, the spacing instructions developed for E-TMA seem to be usable in TMA. The main objective of the next ground experiment in late will be to assess the usability and usefulness of airborne spacing in TMA under very high traffic conditions. The assessment will rely on the same methodology and metrics already used for E-TMA. Secondary objectives will be to define and evaluate fallback procedures. The interaction between TMA and E-TMA, the use of arrival manager and datalink (uplink of target selection by the controller, and downlink of spacing parameters selected by the flight crew) will be studied next. vi Project AGC-Z-FR - EEC Report No. Volume I

7 COSPACE EUROCONTROL ACKNOWLEDGEMENTS This project is sponsored by the Eurocontrol Experimental Centre (EEC), the European Air Traffic Management Programme (EATMP) of Eurocontrol, and the European Commission (EC). The present experiment was made in collaboration with NUPII and MA-AFAS projects from EC. The authors would like to thank DGAC (France), ENAV (Italy), NATS (UK) for the secondment of their controllers. The authors wish to thank Jose Roca (EATMP/AGC) and Jean Lamonoca (DGAC, Aix-Marseille ACC) for their support over years. Thanks to Alain Zinger (DGAC, Paris Orly APP) for his enthusiastic help and encouragement. The authors would also like to thank all the people who worked with them on the experiment. Thanks to Bruno Favennec (Sofréavia), Mourad Abbas (Pacte Novation), Sébastien Mogue (Steria) and Caroline Aiglon (Transiciel) for their technical assistance. Thanks to Virginie Josselin, Nathalie Lépy, Anne Pellegrin and Luc Rodet (Novadis) for carrying out the eye movement analysis. Thanks to Gian Luca Fabiani, Enrico Lucini and Massimo Orsoni (Roma ACC), Didier Dubois (Aix- Marseille ACC), Hervé Le Jeannic and Noël Boussot (Paris ACC), Ludovic Boursier (Paris Orly APP), Claudio Colacicchi (Roma APP), Marcia Connick (Manchester APP) and Liz Jordan (London Gatwick APP) for their enthusiast participation and their fruitful comments. Thanks also to Noëlle Canto, Olivier Galichet and Ronald Granju (Paris Orly APP) who participate during the June prototype experiment. Project AGC-Z-FR - EEC Report No. Volume I vii

8 EUROCONTROL COSPACE viii Project AGC-Z-FR - EEC Report No. Volume I

9 COSPACE EUROCONTROL TABLE OF CONTENTS LIST OF FIGURES... XI LIST OF TABLES... XIV REFERENCES... XV ABBREVIATIONS... XVII. INTRODUCTION.... PRINCIPLES..... MOTIVATION..... STATE OF THE ART..... PROCEDURES..... TECHNICAL MEANS.... CONTEXT AND OBJECTIVES OF THE EXPERIMENT..... STRATEGY AND APPROACH..... PAST EXPERIMENTS..... VALIDATION MODEL..... OBJECTIVES HYPOTHESIS, METRICS AND MEASURES Human shaping factors Controller activity Control effectiveness Control safety..... CASE STUDY.... EXPERIMENTAL DESIGN..... EXPERIMENT SET-UP..... EXPERIMENT SCHEDULE Initial training Continuous improvement..... PARTICIPANTS..... SIMULATED ENVIRONMENT Geographical area Measured sectors Events Traffic samples PROCEDURES ATC procedures and constraints Separation and spacing..... EXPERIMENTAL PLAN Independent variables Dependent variables Run plans... Project AGC-Z-FR - EEC Report No. Volume I ix

10 EUROCONTROL COSPACE.7. FACILITIES AND EQUIPMENT Controller working environment Flight list Short term conflict alert Standard interface Airborne spacing interface.... DATA COLLECTION, VERIFICATION AND ANALYSIS..... DATA COLLECTION..... DATA VERIFICATION..... DATA ANALYSIS..... USE OF A BASELINE.... RESULTS E-TMA..... INTRODUCTION Factual data Statistical significance HUMAN SHAPING FACTORS Usability Motivation Workload Skills Teamwork Synthesis on human shaping factors..... CONTROLLER ACTIVITY Manage safe and expeditious flows of traffic Monitor and analyse traffic situation Assume and transfer aircraft Synthesis on controller activity..... EFFECTIVENESS Quality of flow management Quality of flight service provided Synthesis on effectiveness SAFETY Method Control errors Airborne spacing errors Synthesis on safety..... CASE STUDY FINDINGS TMA UNDERSTAND TMA SPECIFICITY IDENTIFY RELEVANT AIRSPACE AND PROCEDURES SPECIFY DETAILED AIRSPACE AND PROCEDURES ASSESS USABILITY SETUP Airspace and procedures Traffic Controller position Programme Limitations...9 x Project AGC-Z-FR - EEC Report No. Volume I

11 COSPACE EUROCONTROL 7.. ASSESS USABILITY FINDINGS Airspace and procedures Method of use Typical examples of use Controller activity.... CONCLUSION E-TMA Human shaping factors Controller activity Effectiveness Safety..... TMA Design and feasibility Controller activity NEXT STEPS... 9 TRADUCTION EN LANGUE FRANÇAISE... LIST OF ANNEXES See volume II of this report. LIST OF FIGURES Figure Converging situation.... Figure Controller interface with indications of aircraft under airborne spacing (green links between target and instructed aircraft)... Figure ASAS features on navigation display... Figure From to validation model... Figure E-TMA map... Figure TMA map... Figure 7 Simulation room... 9 Figure Label content and mouse buttons... Figure 9 Airborne spacing markings and the three steps... Figure Eye tracker worn by executive controllers during runs... Figure Iterative data verification process... Figure Mean use of airborne spacing number of spacing instructions compared to Figure number of concerned aircraft... 9 Mean duration of airborne spacing duration of airborne spacing compared to flight time of aircraft under airborne spacing... 9 Figure Respective rate of use of all applications in distance (left) and time based (right)... Figure Percentage of target selection not followed by a spacing instruction... Figure Proportion of single and multiple spacing instructions per aircraft... Figure 7 ISA summary recordings for a same team in three conditions... Figure Examples of temporal ISA recordings in the three conditions for an executive controller... Figure 9 NASA-TLX mental demand for both sectors (top),... Project AGC-Z-FR - EEC Report No. Volume I xi

12 EUROCONTROL COSPACE Figure NASA-TLX temporal demand... Figure Overall number of instructions, including spacing instructions, excluding select target... 7 Figure Overall number of instructions, including spacing and select target instructions... 7 Figure Impact of practice on speed variations following spacing instructions.... Figure Marking task repartition between executive and planning controllers... 9 Figure Types of manoeuvring instructions in the three conditions (without, distance and time).... Figure Mapping manoeuvring instructions over sectors... Figure 7 Mean geographical distribution of instructions.... Figure Anticipated sequence building. Example with two sessions (A and A).... Figure 9 Temporal distribution of instructions... Figure Percentage of fixations on the radar (working area only)... Figure Mean fixations in the three conditions... Figure Synthetic view of geographical distribution of fixations Figure Two examples of monitoring curves... Figure Example of monitoring curve for three periods of minutes each... 9 Figure Distribution of number of foveal and parafoveal fixations... Figure Example of temporal distribution of foveal and parafoveal fixation per aircraft... Figure 7 Distribution of mean maximum inter fixation period in foveal and parafoveal regions... Figure Distribution of max maximum inter fixation period in foveal and parafoveal regions. Figure 9 Mean and max inter fixation periods in foveal and parafoveal regions... Figure Geographical distribution of parafoveal and foveal fixations. AR sector.... Figure Example of distribution of maximum periods of fixation.... Figure Location of transfer to next frequency... Figure Examples of distribution of spacing in distance.... Figure Examples of distribution of spacing in time Figure Repartition of spacing quality, based on spacing value at IAF... Figure Closure rates of aircraft with optimal spacing value... 9 Figure 7 Total number of instructions per aircraft. All sessions... Figure Distribution of instructions per aircraft. Session A... Figure 9 Example of aircraft trajectories... Figure Mean speed profile... Figure Example of the impact of distance based spacing on speed profile... Figure Mean aircraft descent profiles... Figure Typical example of different descent profile in one condition... Figure Distance flown, fuel consumed and time flown... Figure Repartition of redundant orders, whose period is shorter than seconds... 9 Figure Repartition of redundant orders, whose period is longer than seconds... 7 Figure 7 Distribution of closing up aircraft, AR sector ( run)... 7 Figure Geographical distribution of instructions. AO sector Figure 9 Spacing at IAF. AO sector... 7 Figure 7... Initial conditions. sequences, merging to OKRIX. AF7JV is direct to MOLEK... 7 Figure 7... AF7JV inserted between the two sequences. AFBL given Figure Figure Figure merge to MOLEK Second sequence needs to be modified all aircraft are given a new merging point Lack of spacing and speed difference led to a drastic speed reduction for the AF7JV Sequence is broken to recover the degraded situation. AFBL is vectored... 7 xii Project AGC-Z-FR - EEC Report No. Volume I

13 COSPACE EUROCONTROL Figure 7... AF7JV is also given a heading... 7 Figure 7... AFBL has just received a speed instruction. AFSS and LBGJ are given speed instruction Figure Cancelling previous instructions and modifying the merging point Figure Speed variations s after spacing instruction ( runs)... Figure 9 Distribution of speed variations... Figure 7 Speed variations seconds after the instruction. Incorrect use. AR sector... Figure 7 Speed variations seconds after the instruction. Correct use. AR sector.... Figure 7 NASA-TLX mental and temporal demand. Incorrect use.... Figure 7 NASA-TLX mental and temporal demand. Correct use.... Figure 7 ISA ratings. Incorrect use... Figure 7 ISA ratings. Correct use... Figure 7 Manoeuvring instructions repartition Incorrect use.... Figure 77 Manoeuvring instructions repartition. Correct use... Figure 7 Geographical distribution of manoeuvring instructions. Incorrect use... Figure 79 Geographical distribution of manoeuvring instructions. Correct use... Figure Geographical distribution of fixations Incorrect use... Figure Geographical distribution of fixations Correct use... Figure Spacing value at IAF. Incorrect use.... Figure Spacing value at IAF. Correct use... Figure Closure rate (aircraft with optimal spacing at IAF). Incorrect use Figure Closure rate (aircraft with optimal spacing at IAF). Correct use... 7 Figure Aircraft mean speed profile. Incorrect use Figure 7 Aircraft mean speed profile. Correct use... 7 Figure Aircraft descent profile. Incorrect use... Figure 9 Aircraft descent profile. Correct use... Figure 9 Today E-TMA and TMA configurations... 9 Figure 9 June TMA sector (simplified map)... 9 Figure 9 Proposed E-TMA and TMA configurations... 9 Figure 9 November TMA sectors (simplified map)... 9 Figure 9 Example of graphical markings... 9 Figure 9 Resulting E-TMA and TMA configurations... 9 Figure 9 Example INIO, step...97 Figure 97 Example INIO, step...9 Figure 9 Example INIO, step...9 Figure 99 Example INIO, step...99 Figure Example INIO, step Figure Example INIO, step... Figure Example INIO, step 7... Figure Example INIR, step... Figure Example INIR, step... Figure Example INIR, step... Figure Example INIR, step... Figure 7 Example INIR, step... Figure Example INIR, step... Figure 9 Example INIR, step 7... Figure Example INIR, step... Figure Distribution of manoeuvring instructions (speed, heading and spacing if applicable) as a function of distance to merging point... Project AGC-Z-FR - EEC Report No. Volume I xiii

14 EUROCONTROL COSPACE LIST OF TABLES Table Spacing instructions for sequencing... Table A typical exchange between controller (left) and pilot (right)... Table Hypothesis related to the impact of spacing instructions on human shaping factors... Table Hypothesis related to the impact of spacing instructions on human activity... 9 Table Hypothesis related to the impact of spacing instructions on effectiveness... Table Hypothesis related to the impact of spacing instructions on safety... Table 7 Simplified experimental schedule (Low, Medium, High, Very high; Distance, Time). Table Run plan matrix for each E-TMA controller, who played each condition once... 9 Table 9 Data collection method and data attributes... Table Problems encountered during the data verification process and actions performed. Table Facts and figures regarding the measured runs... Table Total number of aircraft on the frequency during analysed period... 7 Table Spacing instructions rate of use... 9 Table Spacing instructions mean duration... 9 Table Spacing instructions respective rate of use... Table Impact of airborne spacing on workload assessed with the ISA device... Table 7 Synthesis of ISA ratings, per type, sector and position... Table Evaluation of controller description of traffic, as provided on the sector maps... Table 9 Number of aircraft overflying IAF (period of minutes)... Table Mean values of fuel consumption, flight duration and distance flown... Table Indicators of unsafety considered... 7 Table Synthesis of losses of separation per sessions, as a function of the airborne spacing and per sector... Table Cases of unsafe spacing (extract)... 7 Table Repartition of unacceptable transfer cases, for both sectors... 7 Table Examples of pairs of aircraft for which initial applications were incorrect (extract)... 7 Table Number of occurrences of the various types of airborne spacing related errors... 7 Table 7 Example of traffic with six sequences of four or five aircraft (for a duration of about minutes)... 9 Table Basic situations with two aircraft... 9 Table 9 Comparison of sequence configurations... 9 xiv Project AGC-Z-FR - EEC Report No. Volume I

15 COSPACE EUROCONTROL REFERENCES [] SHAPE web site factors/shape_framework.html. [] Amalberti R. (99). La conduite de systèmes à risque. PUF. [] Casso N. and Kopardekar P. (). Raytheon ATMSDI (Air Traffic management system development and integration). December. Draft guidelines Subtask Human factors metrics guidelines. [] CoSpace (). Towards the use of spacing instructions. Assessing the impact on flight deck. EEC Report. [] Dittmann, A., Kallus, K.W. & VanDamme, D. (). Integrated Task Analysis - Phase Baseline Reference of Air Traffic Controller Tasks and Cognitive Processes in the ECAC Area. EUROCONTROL, Brüssel, N HUM.ET..ST.-Rep-, ( S). [] Endsley, M.R., Farley, T.C., Jones, W.M., Midkiff, A.H. & Hansman, R.J. (99). Situation awareness information requirements for commercial airline pilots. Report ICAT-9-. Cambridge, MA Massachusetts Institute of technology International Centre for Air Transportation. [7] Endsley, M.R., Hansman, R.J. & Farley, T.C. (999). Shared situation awareness in the flight deck-atc system. Digital Aviation Systems Conference. Seattle, USA. [] Endsley, M.R. and Rogers, M.D. (99). Situation awareness information requirements for en route air traffic control. Report DOT/FAA/AM-9/7. Washington, D.C.Federal Aviation Administration Office of Aviation Medecine. [9] Farley, T.C., Hansman, R.J., Endsley, M.R., Amonlirdviman, K. & Vigeant-Langlois, L. (99). The effect of shared information on pilot/controller situation awareness and re-route negotiation. ATM Seminar. [] EUROCONTROL (). ATM+ Strategy Volume, ed.. [] Harwood K. (99). Defining Human-centered system issues for verifying and validating air traffic control systems. In J. Wise, V.D. Hopkin, and P. Stager (Eds.), Verification and validation of complex and integrated human machine systems. Berlin Springer-Verlag. [] Hammer, J. (). Preliminary analysis of an approach spacing application. FAA/Eurocontrol R&D committee, Action plan, ASAS Technical Interchange meeting. [] Kelly, J.R. and Abbott, T.S. (9). In-trail spacing dynamics of multiple CDTI-equipped aircraft queues. NASA TM-99. [] Lee, P.U., Mercer, J.S., Martin, L., Prevot, T., Shelden, S., Verma, S., Smith, N., Battiste, V., Johnson, W., Mogford, R., Palmer, E. (). Free maneuvering, trajectory negotiation, and self-spacing concept in distributed air-ground traffic management. USA/Europe Air Traffic Management R&D Seminar, Budapest, Hungary. Project AGC-Z-FR - EEC Report No. Volume I xv

16 EUROCONTROL COSPACE [] Oseguera-Lohr R.M., Lohr G.W., Abbott T.S., Eischeid T.M. (). Evaluation of operational procedures for using a time-based airborne interarrival spacing tool. Digital Avionics Systems Conference, Irvine, California, USA. [] Pritchett, A.R. and Yankovsky L.J. (99). Simultaneous Design of Cockpit Display of Traffic Information & Air Traffic Management Procedures. SAE Transactions - Journal of Aerospace. [7] Pritchett, A.R. and Yankosky L.J. (). Pilot performance at new ATM operations maintaining in-trail separation and arrival sequencing. AIAA Guidance, Navigation, and control, Denver, Colorado, USA. [] Rayner, K. (99). Eye movements and cognitive processes in reading, visual search, and scene perception. In J. M. Findlay, R. Walker, & R. W. Kentridge (Eds.), Eye Movement ResearchMechanisms, Processes, and Applications (pp. -). New York Elsevier Science Publishing. [9] Schilling, H.E.H., Rayner, K. & Chumbley, J.I. (99). Comparing naming, lexical decision, and eye fixation times Word frequency effects and individual differences. Memory and Cognition, 7-. [] Sorenssen, J.A. and Goka, T. (9). Analysis of in-trail following dynamics of CDTIequipped aircraft. Journal of Guidance, Control and Dynamics, vol., pp -9. [] Williams D.H. (9). Self-separation in terminal areas using CDTI. Proceedings of the Human Factors Society Annual Meeting. EXPERIMENT MATERIALS Grimaud, I. (). Controller handbook, version., November. Sheehan, C. (). Pilot working position, November. PROJECT WEB SITE xvi Project AGC-Z-FR - EEC Report No. Volume I

17 COSPACE EUROCONTROL ABBREVIATIONS Abbreviation ADS-B ART ASAS ATC ATM CDTI CWP EEC ETA E-TMA EXC FIR HMI IAF ISA N/A NASA-TLX PLC SID STAR STCA TIS-B TMA UIR WPT De-Code Automatic Dependant Surveillance Broadcast Analysis and Replay Tool Airborne Separation Assistance System Air Traffic Control Air Traffic Management Cockpit Display of Traffic Information Controller Working Position EUROCONTROL Experimental Centre Estimated Time of Arrival Extended Terminal Manoeuvring Area Executive Controller Flight Information Region Human Machine Interface Initial Approach Fix Instantaneous Self Assessment Not Applicable NASA Task Load Index Planning Controller Standard Instrument Departure Standard Terminal Arrival Route Short Term Conflict Alert Traffic Information Service Broadcast Terminal Manoeuvring Area Upper Information Region Waypoint Project AGC-Z-FR - EEC Report No. Volume I xvii

18 EUROCONTROL COSPACE xviii Project AGC-Z-FR - EEC Report No. Volume I

19 COSPACE EUROCONTROL. INTRODUCTION This report presents the results and findings of the CoSpace controller experiment conducted in November. This experiment fitted in with a series of air and ground validation exercises aiming at investigating the use of spacing instructions (denoted airborne spacing) for sequencing of arrival flows. The previous ground experiment conducted in November focussed on E-TMA (from cruise to initial approach fix) using distance based spacing. The present experiment aims at going a step further by introducing time based spacing (expected to be more efficient) and integrating TMA (from initial to final approach fix). More precisely, the objective was for E-TMA, to compare the respective impact of distance and time based spacing on controller activity and quality of service provided; for TMA, to assess the usability of spacing instructions. Flight deck experiments conducted in May and December are reported in a separate document (CoSpace ). The document is organised as follows Section introduces the principles of airborne spacing; Section introduces context and objectives of the experiment; Section describes the experimental design; Section describes the data collection and analysis; Section presents the results for E-TMA; Section 7 presents the main findings for TMA; Section draws conclusions.. PRINCIPLES.. MOTIVATION New allocation of spacing tasks between controller and flight crew is envisaged as one possible option to improve air traffic management. The motivation is neither to transfer problems nor to give more freedom to flight crew, but really to identify a more effective task distribution beneficial to all parties. This allocation of spacing tasks to flight crew denoted airborne spacing is expected to increase controller availability and to improve safety, which in turn could enable better efficiency and/or, depending on airspace constraints, more capacity. In addition, it is expected that flight crew would gain in awareness and anticipation by taking an active part in the management of his situation. Airborne spacing assumes new surveillance capabilities (e.g. ADS-B) along with new airborne functions (ASAS)... STATE OF THE ART Airborne spacing for arrival flows of aircraft was initially studied from a theoretical perspective through mathematical simulations, to understand the intrinsic dynamics of in-trail following aircraft and identify in particular possible oscillatory effects (Kelly & Abbott, 9; Sorenssen & Goka, 9). Pilot perspective was also addressed through human-in-the-loop simulations (Pritchett & Yankovsky, 99; Pritchett & Yankovsky, ; Williams, 9) and flight trials (Oseguera-Lohr et al., ) essentially to assess feasibility. The ATC system perspective was considered through model-based simulations, to assess impact on arrival rate of aircraft (Hammer, ). Initial investigations were also performed with controllers in approach (Lee et al., ). We used to call it limited delegation or delegation of spacing tasks, but the term delegation appeared to be misleading as it is sometimes understood as transfer of responsibility for separation. Project AGC-Z-FR - EEC Report No. Volume I

20 EUROCONTROL COSPACE.. PROCEDURES The principles of airborne spacing considered here is to provide the controller with a set of new instructions for sequencing of arrival flows. Through these new spacing instructions, the flight crew is tasked to acquire and maintain a given spacing with respect to a preceding aircraft (the target). Airborne spacing is composed of three phases Identification, in which the controller indicates the target aircraft to the flight crew. Spacing instruction, in which the controller specifies the task to perform. End of airborne spacing, which marks the completion of the task. As for any standard instruction, the use of spacing instructions is at the controller initiative, who can decide to end its execution at any time. The flight crew however can only abort it in case of a problem onboard such as a technical failure. In terms of responsibility, as opposed to visual separation, there is no transfer of separation responsibility. Four spacing instructions for sequencing are proposed (Table ). Table Spacing instructions for sequencing Maintain spacing Resume when spacing reached then maintain it In-trail Remain behind Heading instruction then remain behind Merging Merge behind Heading instruction then merge behind For illustration purposes, let us consider the situation of two arrival aircraft converging to a point, then following the same route to the airport. Today, the controller must ensure that the spacing is maintained, and therefore has to continuously monitor the situation and if necessary issue heading and/or speed instructions. With airborne spacing, the maintaining of the spacing through speed adjustments is transferred to the flight deck (Table and Figure ). Whereas the land after clearance can generally be given in final approach only (visual contact required), the spacing instruction can be issued earlier, typically before descent and regardless of visibility conditions, thanks to the display of the target aircraft onboard. However, applicability conditions need to be respected. In this example, prior to issue any spacing instruction, the controller must ensure that aircraft speeds are compatible, and that the spacing at the converging point is not lower than the desired spacing. For more details about applicability conditions, see Annex. XYZ, select target. Table A typical exchange between controller (left) and pilot (right) Selecting target, XYZ. The designation of the target aircraft is done through a unique identifier (here the SSR code). After selection and identification on the screen, the pilot replies The controller can then issue the spacing instruction XYZ, behind target, merge to WPT to be 9s behind. XYZ, target identified, o clock, Nm. Merging to WPT to be 9s behind target, XYZ. The pilot has to adjust his/her speed to maintain the spacing at the converging point and after the point. The airborne spacing will be ended by the controller when appropriate XYZ, cancel spacing, reduce speed knots. Cancelling spacing, reducing speed knots, XYZ. Project AGC-Z-FR - EEC Report No. Volume I

21 COSPACE EUROCONTROL Future positions Target WPT 9s Ownship Target 7 Ownship - Current Positions Figure Converging situation. The spacing instructed aircraft has to adjust its speed to maintain the spacing with the target aircraft at the converging point. Whereas the land after clearance can generally be given in final approach only (visual contact required), the airborne spacing can start earlier, typically before descending and whatever the visibility conditions, thanks to the display of the target aircraft onboard... TECHNICAL MEANS On the controller side, the only required modification to the current working environment is the knowledge of aircraft ASAS equipages, for example through a field of the flight plan. In addition, graphical marking capabilities on the controller screen would be useful as a reminder of on-going airborne spacing as well as a support for co-ordination when transferring spacing instructed aircraft to next sector (Figure ). Figure Controller interface with indications of aircraft under airborne spacing (green links between target and instructed aircraft) Project AGC-Z-FR - EEC Report No. Volume I

22 EUROCONTROL COSPACE On the cockpit side the airborne spacing requires the display of the target aircraft onboard the instructed aircraft. The Automatic Dependant Surveillance Broadcast (ADS-B) is a surveillance mean in a pre-operational state in which equipped aircraft transmit spontaneously their position and velocity (and eventually their trajectory). The Traffic Information Service - Broadcast (TIS-B) is an additional mean to be used when some aircraft are not ADS-B equipped position and velocity are transmitted via a ground station to equipped aircraft. The traffic data received through ADS-B or TIS-B is displayed on a screen in the cockpit. This capability is denoted Cockpit Display of Traffic Information (CDTI). In addition to the display of the target aircraft, assistance to maintain spacing is required, typically through graphical cues. This capability is usually denoted Airborne Separation Assistance System (ASAS). Despite the similarity between terms, it should be noted that ASAS is completely distinct from the collision avoidance system ACAS/TCAS which is a last resort system. An example of the display used in the CoSpace flight deck experiments is presented on Figure. Spacing scale Reference line Predicted spacing Suggested airspeed Target aircraft symbol Figure ASAS features on navigation display This term was introduced in 99 at a time where the distinction between separation and spacing was not clearly identified by the ASAS community. Project AGC-Z-FR - EEC Report No. Volume I

23 COSPACE EUROCONTROL. CONTEXT AND OBJECTIVES OF THE EXPERIMENT.. STRATEGY AND APPROACH One of the key aims of the study described in this paper is to build an understanding of the potential impacts of spacing instructions and of the evolutions induced or required with respect to today s ATC. Therefore, since its inception, the study follows an iterative process, in which every step (real-time experiment) helps defining the next one. The stepwise strategy followed can be described along three dimensions Operational start in cruise (in extended TMA) and progressively get closer to the runway (in TMA). Validation start assessing usability (e.g. concept, procedures, interface) and progressively address impact on user activity (controller, pilot) and eventually on the ATC system (e.g. quality of control/flying, safety, efficiency). Technology start with a basic working environment (e.g. paper strips, voice communications, no advanced tools, manual mode in the cockpit) and progressively introduce assistance and technology when need clearly identified (e.g. uplink of target selection, downlink of spacing parameters, controller spacing monitoring aids, automatic spacing mode in the cockpit). Because of the inherent air-ground nature of the concept, both controller and pilot aspects have to be considered. To limit risk in terms of development and execution, but mainly to be able to properly control experimental parameters, it was decided to conduct two separate streams of experiments (air and ground). Nevertheless, the consistency between the two streams is ensured essentially by relying on the same concept (applications, procedures, phraseology), the same operational environment (type of airspace, scenarios) and a unified validation framework (experimental plan, metrics)... PAST EXPERIMENTS In order to get feedback on the concept, an initial air-ground experiment was carried out in 999. Then, to assess the benefits and limits, two streams of air and ground experiments were conducted three ground experiments since with in total controllers from different European countries over 7 weeks; two air experiments since with in total pilots over days. Regarding the ground side, investigations started in upper airspace (E-TMA), from Nm until the exit point (IAF), with aircraft initially in cruise and starting their initial descent. For the last simulation (), the airspace comprised four en-route sectors adapted from Paris ACC handling south-east arrivals to Orly and Charles-De-Gaulle airports. The traffic simulated was derived from a real traffic. To allow for comparison, each exercise was played twice with distance based and without airborne spacing. It was observed (through the distribution of manoeuvring instructions and location of eye fixations) that the use of spacing instructions partly relieves the controller of the maintaining of sequences, and allows him to concentrate on the building of sequences. This (positive) impact on controller activity resulted in more stable and homogeneous spacing between aircraft at exit point. Project AGC-Z-FR - EEC Report No. Volume I

24 EUROCONTROL COSPACE Although promising, these investigations were exclusively focussed on the E-TMA. Regarding TMA, preliminary investigations were carried out through a prototype simulation (June ) involving four approach controllers (during. day each). The objective was to identify possible use and limits of airborne spacing. Although airborne spacing was perceived as potentially beneficial for some situations (e.g. downwind leg), key issues remain such as the final integration of flows already under airborne spacing... VALIDATION MODEL To analyse results from the previous experiment (), four dimensions were proposed acceptability, activity, effectiveness and safety. These dimensions can be seen as successive layers (lower levels being included in higher levels) and reflect wider perspective on impact of the concept (Figure left). The acceptability level referred to the relevance of the concept and the possible context for its application. It reflected how controllers perceive and implement it. The activity level investigated the integration of the concept within the overall control activity. It assessed the compatibility with existing tasks, procedures and strategies. It provides insight on the impact on active control and monitoring tasks. The effectiveness level questioned the consequence of using spacing instructions on the quality of control performed. In other words, it assessed the impact on the results of the controller activity, at a more systemic level, including interaction with adjacent sectors as well as with flight decks. Last of all, the safety level aims at analysing if induced changes are acceptable in terms of safety, and assessing the risks induced and the mitigation means. Acceptability Human shaping factors Activity Activity Effectiveness Effectiveness Safety Safety Figure From to validation model Project AGC-Z-FR - EEC Report No. Volume I

25 COSPACE EUROCONTROL Though the model provided a useful framework to support the analysis and put results in perspective, it had limitations. The main one was its inability to easily encompass human factors such as motivation, confidence or skills. The model was refined (Figure right). The first level (acceptability) was re-scoped in terms of human shaping factors and includes human factors and usability issues. It provides feedback on the impact of the concept on human factors such as workload, confidence and teamwork. It also includes feedback on the concept usability. The human activity level remained unchanged, except that the analysis now proposes to refine control tasks through a grid model of tasks. The second modification concerns the outer levels. To highlight the necessary trade-off between effectiveness and safety, the two dimensions are now considered at the same level. The validation consists in analysing not only the impact of the concept on each dimension, but also the mutual impact of dimensions. Typically, workload (human shaping factor) may influence activity, effectiveness, safety and acceptability and be influenced by them. Results from each level need to be combined and interpreted jointly... OBJECTIVES The objective of the experiment was twofold For E-TMA, to compare the respective impact of distance and time based spacing on controller activity and effectiveness. For TMA, to assess the usability of airborne spacing, in particular regarding the integration of flows of aircraft, transfer from and co-ordination with E-TMA. It should be highlighted that the two objectives reflect two distinct levels of maturity in TMA, still at an exploratory stage, the aim is to understand how airborne spacing could be used, whereas the investigations in E-TMA mostly consist in comparing the impact of variants (distance versus time) on user activity and on the quality of the service provided... HYPOTHESIS, METRICS AND MEASURES... Human shaping factors This first dimension addresses the impact of airborne spacing on human shaping factors, which include the instruction usability, and its impact on human factors. The set of human related factors denoted human shaping factors, influence the task performance and are also influenced by the task itself (HERA, ). Based on literature review (Amalberti, 99; Casso et al., ; Harwood, 99, SHAPE), we identified items which are thought to reflect the main human shaping factors possibly influenced by airborne spacing workload, skills, teamwork, motivation, confidence and usability. Human shaping factors which are all mutually dependent need to be analysed jointly. Because it is based on current practices, the spacing instruction should be nothing more than a new instruction, used at controller appreciation, if needed and when needed. It should not change controller strategy and should not require additional skills. However, possible reasons for concept rejection by controllers have been foreseen feeling of losing authority or losing the exciting part of the work, non confidence in pilots, possible difficulty in handling abnormal events. These items are typical issues that are investigated through the human shaping factor dimension. Project AGC-Z-FR - EEC Report No. Volume I 7

26 EUROCONTROL COSPACE Apart from workload, that is addressed through its perceived nature (informed along runs by controllers) and some more objective indicators (e.g. number of instructions, of fixations), the remaining items were addressed through the combination of observers notes, questionnaires and debriefing items. Hypothesis related to this dimension are presented in Table. Table Hypothesis related to the impact of spacing instructions on human shaping factors Metrics With (versus without) spacing instruction Hypothesis Time (versus distance) based spacing instruction Usability Motivation Workload Skills and training needs U Will be usable as it relies on current practices (rate of use; rate of errors). U Will be more or less usable depending on sector configuration (rate of use). M Will be motivating since controllers will recognise the benefits while preserving their authority (i.e. strategy, decisionmaking) (questionnaire; rate of use). M In contradiction to M, there is a risk of over-expectation (e.g. due to confusion of roles), thus potentially leading to frustration (questionnaire). W Will reduce workload, thus providing more availability (questionnaires, ISA, NASA-TLX, number of instructions fewer; number of fixations fewer; radiocommunications less, physiological measures). W Will reduce time-criticality of some actions (NASA-TLX temporal demand lower; resume instructions fewer), thus possibly smoothing of overall activity (temporal distribution of instructions smoother). Sk Will not change their skills nor require new control skills, but will require training on procedures, phraseology and awareness of change on the flight deck (questionnaire). Sk Will require careful training for the controllers to acknowledge that airborne spacing fits in with current practices (linked to over-expectation) (questionnaire; rate of errors). Sk There is a risk of losing skills in maintaining sequences (i.e. speed management in descent) (indicator would be linked to feedback from operational usage). Teamwork No No U Will be less intuitive thus possibly more difficult to use (linked to activity ) (questionnaire; rate of errors). M Will be perceived as more difficult initially, i.e. assessment of spacing in time different from today (questionnaire; rate of use). M Will be perceived as more efficient when used to it (questionnaire; rate of use). W Will initially increase mental demand (due to M above) (ISA, NASA-TLX mental demand). W Will decrease temporal demand (due to less time-critical management of descent) (NASA- TLX temporal demand). W May initially increase workload to assess transfer conditions as aircraft will be transferred in catching up situations (though spacing will remain valid) (questionnaire; geographical distribution of eye-fixations more located in transfer area). Note also valid for receiving sector. Sk Will require training and practice to assess spacing in time (questionnaire; rate of errors). Project AGC-Z-FR - EEC Report No. Volume I

27 COSPACE EUROCONTROL... Controller activity The human activity analysis aims at assessing the potential benefit of the concept for a given task (i.e. the task that the concept is expected to support) and at checking that it does not impact negatively other tasks. Based on literature review (Dittman et al, ; Endsley & Rodgers, 99; Endsley et al., 99; Endsley et al., 999; Farley et al. 99), we identified 7 main executive controller tasks Manage safe and expeditious flows of traffic, which refers in E-TMA and TMA to () detect and solve conflicts, and () build and maintain arrival sequences; Provide flight service, which correspond to satisfy aircraft request; Monitor and analyse traffic situation; Handle co-ordinations with adjacent sectors; Assume and transfer aircraft; Handle unexpected events; Hand over situations (i.e. shift). In the case of spacing instruction which involves controllers and flight crews, it is essential to identify if and how the control and the flight tasks are modified. In addition, the actions executed, the strategy used and the working methods need to be considered as a result of the human shaping factors, of the environment constraints (e.g. traffic load) and of the concept requirements (e.g. applicability conditions). Given the experiment focus, we restricted the analysis to three of these tasks, which are according to us the core ones in nominal conditions manage arrival flows, monitor and analyse situation and assume/transfer aircraft. Note that the other tasks were not considered for two reasons either we do not expect them to be impacted (e.g. handle co-ordination) or their assessment was not planned in the context of the present experiment (e.g. handle unexpected events, hand over situation). Hypothesis related to the three tasks are presented in Table. Table Hypothesis related to the impact of spacing instructions on human activity Metrics With (versus without) spacing instruction Hypothesis Time (versus distance) based spacing instruction Arrival flow management Seq Will not modify the nature of the building phase, e.g. same strategy used (sequence order same), same implementation (instructions same nature). Seq Will alleviate the active control of the maintaining phase (speed instructions fewer) and, when using heading then, of resuming step of building phase (resume instructions fewer). Seq Consequence of will allow to anticipate the building of sequences (geographical distribution of instructions upstream). Seq Seq Will make the building phase slightly longer and/or more difficult, due to Seq. Larger distance required at high altitude (thus imposing larger and/or longer deviations) (heading instruction larger value; resume/direct instruction later). Seq. More difficult assessment of spacing in time when initiating spacing (speed change after s more occurrences). Will facilitate the management of descent (e.g. top of descent) by eliminating the slowdown effect (speed and descent profiles similar to today s). To avoid or limit this effect, the use of airborne spacing could be delayed till descent phase (speed lower thus spacing smaller). Indeed, 9s corresponds to -Nm at high altitude with high ground speed, and to Nm at FL. Project AGC-Z-FR - EEC Report No. Volume I 9

28 EUROCONTROL COSPACE Metrics With (versus without) spacing instruction Hypothesis Time (versus distance) based spacing instruction Traffic situation monitoring and analysis Mon Will not change monitoring during building phase (eye-fixations peak over building area). Mon Will alleviate monitoring during maintaining phase (eye-fixation decrease over maintaining area). Mon May not change the quality /effectiveness of monitoring, i.e. still able to detect any event (eyefixation over aircraft similar period). Mon May change nature and content of monitoring (eye-fixation more on groups of objects). Mon Will impose more monitoring in the building phase (eye-fixation more over building area). Note building area larger (see hypothesis above). Mon Will cause more difficult assessment of spacing when initiating and during airborne spacing (eye-fixation more frequent on aircraft under spacing). Aircraft assume / Transfer TF Will improve transfer conditions, i.e. aircraft transferred in more stable situations (distance between aircraft and closing speed). No... Control effectiveness The way the tasks are performed results in a service provision, whose quality (control effectiveness) needs to be measured. We consider specifically flow management and flight service provision. Increased controller availability and more frequent speed adjustments performed by the flight crew are expected to improve the quality of flow management, in particular in enabling more homogeneous flows of aircraft. In addition, to assess possible impact on throughput, we consider the number of transferred aircraft and the transfer conditions (value and stability of spacing at transfer). Though the major expected benefit relies on the controller side, it is essential to check if airborne spacing is beneficial (at least not detrimental) to the flight deck. Quality of service provision is addressed through speed and descent profiles and flight efficiency, in terms of fuel consumed, time and distance flown. In addition, it is essential to assess if benefits are equally distributed between all aircraft, as opposed to being beneficial to some aircraft and detrimental to others. This is investigated by analysing the distribution of instructions per aircraft. This could be an indicator of the impact of airborne spacing on flight crew activity, even though it only provides a partial view (typically, manual speed adjustments not simulated). Benefit is for the receiving sector, e.g. TMA. Project AGC-Z-FR - EEC Report No. Volume I

29 COSPACE EUROCONTROL Hypothesis related to control effectiveness are presented in Table. Table Hypothesis related to the impact of spacing instructions on effectiveness Metrics With (versus without) spacing instruction Hypothesis Time (versus distance) based spacing instruction Quality of flow management FM Will enable more stable flows and at optimal rate (spacing at IAF more regular). No Quality of service provided SP Will induce fewer trajectory alterations/changes (number of instructions per aircraft fewer; trajectories straighter). SP SP Will improve speed and descent profiles (speed, descent profiles improved; consumption reduced). Will impose larger and/or longer deviations (distance, time flown increased).... Control safety Safety assessment has two objectives Assess if current level of safety is improved or at least maintained (including the verification that existing mitigation means are preserved). Ensure that potential new risks induced by spacing instructions are mitigated. Because the spacing instruction is based upon current practices, we expect it to be easy to understand and usable, thus not error prone. However new forms of errors might be introduced. In addition to analysing possible indicators of unsafety, a typology of airborne spacing-related errors, including possible causes and mitigation means needs to be proposed. Subjective feedback is provided by controllers comments, collected in questionnaires and during debriefing sessions, as well as events observed during the simulations. Objective indicators used when analysing the system recordings are separation infringements, transfer conditions (spacing infringements) and number of airborne spacing errors. Project AGC-Z-FR - EEC Report No. Volume I

30 EUROCONTROL COSPACE Hypothesis related to safety are presented in Table. Table Hypothesis related to the impact of spacing instructions on safety Metrics With (versus without) spacing instruction Hypothesis Time (versus distance) based spacing instruction Control errors CE CE May provide a situation less prone to error occurrence (due to increased controller availability and reduced time-critical actions) (rate of errors). May reduce the number of loss of spacing (due to flight deck involvement) (number of losses of spacing). No Airborne spacingrelated errors AS AS Means provided (in particular flight crew involvement) will enable new errors to be mitigated/recovered (number of errors detected). Might generate controller overconfidence in aircraft under spacing and/or disengagement in traffic management, e.g. resulting in poor error and/or conflict detection (questionnaire, rate of errors). No.. CASE STUDY The hypotheses previously presented assume a proper use of airborne spacing. Indeed, the correct use of airborne spacing is expected to provide availability (reduced workload) that could be converted in time spent on core tasks (e.g. build sequences, monitor situation and transfer aircraft) and possibly enabling a better quality of control. Conversely, an incorrect use of airborne spacing could lead to increased workload and possibly to degraded situations. For extreme cases, incorrect use could lead to situations more demanding than under conventional control. Two case studies will be used to assess this hypothesis. Project AGC-Z-FR - EEC Report No. Volume I

31 COSPACE EUROCONTROL. EXPERIMENTAL DESIGN.. EXPERIMENT SET-UP Considering the two distinct objectives, applying to two groups of controllers and two types of airspace, the experiment could have been split into two separate sub-experiments one on TMA, and one on E-TMA. However, to optimise time and resources, those two sub-experiments were combined as follows. First part exploring TMA, E-TMA building sequences to be transferred to the TMA for final integration. TMA and E-TMA controllers were jointly trained with airspace, traffic flows and spacing instructions. This was expected to allow for a qualitative measured session for TMA, whereas this still constituted training for E-TMA. It should be noticed however that the qualitative measurement for TMA highly depended on the quality of the sequencing (in particular, transfer conditions) made by the E-TMA. Four measured sectors (two E-TMA and two INI), two feed sectors and one pseudo-sector were simulated. The measured sectors were manned with two controllers (executive and planning). One controller was in charge of the two feed sectors and one of the pseudosector. While the E-TMA controllers of the measured sectors built sequences of (delegated) aircraft according to their own appreciation, the role of the controller playing on the pseudo-sector was to create one flow of (delegated) aircraft according to a predefined script to be transferred to each INI sector (see examples of the FEED task list in annex ). This gave the TMA controllers opportunities to integrate two flows of similar importance of aircraft (delegated in E-TMA). Second part measuring E-TMA, E-TMA building sequences to be transferred to the TMA, not manned. The significant training period was expected to allow for a quantitative measured session for E-TMA with very high traffic. Not manning the TMA suppressed all possible interventions on aircraft from the approach controllers, and thus allowed to measure the intervals between aircraft at initial approach fix. Only the two E-TMA sectors and two feed sectors were simulated. The measured sectors were manned with two controllers (executive and planning) and the feed sectors with one controller. The role of the E-TMA controllers remained identical. Note two pseudo-pilots were attached to each measured sector, and one pseudo-pilot to the pseudo-sector. No pseudo-pilot was attached to the feed sectors. Project AGC-Z-FR - EEC Report No. Volume I

32 EUROCONTROL COSPACE.. EXPERIMENT SCHEDULE The experiment took place between November th and 9 th. Two main phases characterise the experiment training and measured exercises. Simplified version of the experimental schedule is presented on Table 7. The detailed schedule of the simulation is presented in Annex. Table 7 Simplified experimental schedule (Low, Medium, High, Very high; Distance, Time) Week Sector Day Training All Environment (airspace, traffic, L) Spacing instruction (Distance, L) Spacing instruction (Time, M) Spacing instruction (D-T, M) Spacing instruction (D-T, M) a TMA Measured (H) Debriefing b E-TMA Spacing instruction (D, T) Training (H) Spacing instruction (D, T), plus eye-tracker Spacing instruction (D, T), plus eye-tracker Measured (V) E-TMA Measured (V) Debriefing Studies on problem solving (Bisseret, 99) showed that experts happened to fall back to novice behaviour when confronted to a situation they do not recognise as a known one. Even if previous strategies are still valid, they need time to figure it out. Therefore, because we expect such risks of expert destabilisation to initially occur with airborne spacing, a special attention has been paid to training issues.... Initial training Controller handbook presented an overview of the concept, of the simulation objectives and of the sectors. The initial training was decomposed between getting familiar first with the controlled environment and second with the spacing instructions. After a general presentation of the objectives of the experiment, hands on practice of the airspace and the traffic flows at a working position took place during the first day. On the second day, controllers practised distance based spacing and time based on the third day. The next two days provided more opportunities to use the spacing instructions in the two conditions (distance and time) with traffic load gradually increasing up to very high. The motivation was to confront controllers early with such a high level of traffic. Combining E-TMA and TMA sectors aimed at putting E-TMA controllers in realistic conditions measurements of TMA exercises required E-TMA controllers to respect as much as possible transfer conditions. The second week was also used to accustom E-TMA controllers to eye tracker devices. Project AGC-Z-FR - EEC Report No. Volume I

33 COSPACE EUROCONTROL... Continuous improvement Every morning a collective debriefing aimed at collecting controller feedback and identifying problems encountered. Whereas controllers did comment their respective performance, the role of the team members was rather to explain situations and confirm how the concept could be used (or should have been). These debriefings worked as a continuous improvement in the use of airborne spacing. In addition, an operational expert conducted on a daily basis individual debriefing sessions using a replay tool. These sessions helped discussing specific situations with each controller and answering their questions. They also complemented collective debriefings in enabling people to evoke more openly encountered difficulties or misunderstanding. In addition to the daily debriefings, a final one took place on the last experiment day. The objective was to collect a more synthetic feedback on the overall experiment. Whereas questionnaires aimed at collecting structured comments, the debriefing aimed at discussing perceived benefit, limits and perspectives... PARTICIPANTS Six en-route controllers (one from Aix-Marseille ACC, two from Paris ACC and three from Roma ACC), four approach controllers (Manchester, London Gatwick, Orly and Roma) and a total of ten pseudo-pilots took part in the experiment... SIMULATED ENVIRONMENT... Geographical area The simulated airspace was part of Paris Southeast area, handling arrival flows from cruise to landing. It was thought to be representative of a dense area and generic enough to allow an easy assimilation by the controllers. The airspace consisted of two E-TMA sectors (AR and AO) of Paris FIR/UIR (Figure ), and two TMA sectors, denoted INIR and INIO (Figure ) for Charles De Gaulle and Orly airports (LFPG and LFPO). Each E-TMA sector was the combination of one pre-sequencing sector (AR or AO) and one sequencing sector (AR or AO). Each TMA sector was a simplified version of the existing sectors, in which INI and ITM positions were grouped. Purposes of the sectors were AR mainly sequencing arrivals to LFPG with one converging area (DJL and TINIL points). AO mainly sequencing arrivals to LFPO with two converging points (ATN and OKRIX). INIR integrating flows to LFPG coming from two entry points (OMAKO and LORTA), both base leg. INIO integrating flows to LFPO coming from two entry points (MOLEK and ODRAN), one base leg and one downwind leg. Feed sectors were designated FE and FW, and the pseudo-sector AF. No restricted area was simulated. Note for AR, AR, AO and AO, actual designations are respectively UJ, AR, UT/TU and AO/SU. Project AGC-Z-FR - EEC Report No. Volume I

34 EUROCONTROL COSPACE BONET LGL FW CHW MAROL INIO VASPO OD RAN CDN CLM VAS TSU FAO LFPO VAS DESCT ABITA MLNS VAS OMAKO KOTUN INKAK IPLAN MOLEK VDP TRO DORDI LAULY VERIX AO INIO INIR AR INIR BOLLY COSPACE NOV' FE LU VAL EPL USIMI OK RIX AO CHABY AR RLP RESPO IXILU LU L OSKIN AF AMB AR FE DJL TINIL VERDI DERAK REKLA PENDU DELOX FW RIGNI BENIP ATN AO FE BAGOL LISMO ROMTA TUROM AO 9/unl. AO /unl. AR 9/unl. AR /unl. INIO /. INIR /. AF /unl.9 FW /unl. FE /unl. GU ERE MOU LESPI ROA AMORO BUSIL ALURA GERBI BULOL LU SAR LOGNI SAUNI KELUK GVA MILPA GA LBI PAS Véro/ / Figure E-TMA map COSPACE NOV' FW VERMA OPALE NITAR KATIL KENAP AF AO 9/unl. AO /unl. AR 9/unl. AR /unl. INIO /. INIR /. AF /unl.9 FW /unl. FE /unl. BON ET Véro// LGL CHW AF MAROL CDN INIO USIMI BT ODR AN VASPO NITEN SOMTU MTD TALUN ANARU XERAM KOPOR INIR LFPG LORTA TARIM GIMER REM VELER BSN SUIPE FAG LOR CLM VAS TSU FAO FE LFPO VAS KOTUN OMAKO DESCT VAS AR INIR MLNS AO INIO INKAK IPLAN MOLEK DORDI VDP TR O ABITA BOLLY LA ULY VERIX FW AO OKR IX AR Figure TMA map Project AGC-Z-FR - EEC Report No. Volume I

35 COSPACE EUROCONTROL... Measured sectors Four measured sectors (two E-TMA and two INI) and two feed sectors were simulated during the first part, and only two measured E-TMA and two feed sectors during the second part. Each measured sector was manned with two controllers (one planning and one executive). Feed or pseudo sectors were manned with one controller at a time.... Events Because the objectives of the experiment were to assess the benefits of spacing instructions in nominal situations, no degraded event was introduced.... Traffic samples The traffic samples were derived from real traffic data, but were modified to create clusters of aircraft. For high and very high samples, traffic was increased by duplicating flights. Low and medium traffic samples were available for training purpose. High traffic samples were used for TMA measurements and still E-TMA training, corresponding respectively to and 7 arrivals per hours. Very high traffic samples were used for E-TMA measurements, corresponding to arrivals and about overflying per hour, leading to about aircraft on contact. The resulting traffic was close to a real high-density traffic. To allow for comparison, three different traffic samples were used. List of aircraft for each traffic sample are presented in Annex. All the traffic was equipped to receive spacing instructions, thus offering maximum opportunities... PROCEDURES... ATC procedures and constraints Arrivals to LFPG Arrivals to LFPO From OMAKO, traffic was cleared to FL. From LORTA, traffic was cleared to FL. From MOLEK, traffic was cleared to FL9. From ODRAN, traffic was cleared to FL. From south-east, traffic was cleared to FL7 MAX over DJL. Arrivals to LIMC, LIML, LIMF GERBI FL9 MAX. Arrivals to LSGG SAUNI FL MAX. To have balanced flows in TMA, predefined aircraft had to be killed during their transfer to TMA. This was however transparent for the TMA controller as these aircraft were not visible on his screen. Project AGC-Z-FR - EEC Report No. Volume I 7

36 EUROCONTROL COSPACE Departures from LSGG Were sent climbing to FL MAX to DJL. Departures from LFLL via ATN Were sent to AO climbing to FL MAX to ATN.... Separation and spacing The minimum standard separation was Nm in en-route, Nm in approach. The standard spacing at transfer between en-route and approach had to be (unless explicit coordination with the approach) Nm without spacing instructions and with distance based spacing instructions, 9 seconds with time based spacing instructions... EXPERIMENTAL PLAN... Independent variables For both E-TMA and TMA, the main variable considered was the spacing instructions with three values without, distance, time. For exercises with distance and time, the use of spacing instructions was at controller discretion. In E-TMA, as seen during previous experiments, sector configuration has an impact on the use and benefits. It was thus added as a second variable with two values AR, AO. AR is composed of one early converging area while AO is composed of one early and one late converging area, reducing the opportunity to build early sequences of aircraft. In TMA, to assess if sector configuration has an impact, two distinct (single landing runway) configurations were simulated one with a base and a downwind legs (INIO), and the other with two base legs (INIR).... Dependent variables The metrics previously presented correspond to the dependent variables.... Run plans For E-TMA and TMA, the planning and the executive controller were of different nationalities. Two reasons guided this choice basically to force them speak in English and thus ensure that communications would be understandable for observers; beyond, to force controllers to step back, be open-minded and get back to basics (i.e. avoid the use of local rules of thumbs ). A same traffic was never played systematically without then with airborne spacing (nor the opposite). A same traffic was never played twice successively. A same controller would not play two successive exercises at the same position. Project AGC-Z-FR - EEC Report No. Volume I

37 COSPACE EUROCONTROL In TMA controllers alternated positions but not sectors (due to short training). Each controller played runs as executive controller in order to test the conditions (without, time and distance). runs were measured to enable all controllers to play all conditions. In E-TMA, the design required each controller to play conditions (without, time and distance) on the two different sectors at the executive position. Because we had controllers per sector, the run plan required measured runs. To summarise, we did alternate The traffic samples, in order to reduce risks of learning the traffic. The control positions (executive, planning) so that they all play as executive controller once every day. The sector manned not only to reduce a boredom feeling, but also to measure the impact of the sector configuration (E-TMA only). Table summarises E-TMA controller runs plan. Table Run plan matrix for each E-TMA controller, who played each condition once Airborne spacing Without Distance Time Sector AO AR.7. FACILITIES AND EQUIPMENT.7.. Controller working environment The environment was similar to today environment (Figure 7), making use of progress strips, along with flight lists, and a short term conflict alert. However, no arrival manager (sequencing tool) was available. Marking functions dedicated to airborne spacing were available. Figure 7 Simulation room Project AGC-Z-FR - EEC Report No. Volume I 9

38 EUROCONTROL COSPACE.7.. Flight list The flight list enabled the controllers to update the exit flight level (XFL) when needed. The entry point (EPT) and entry flight level (EFL) were not displayed. Indeed, the integration of flights was made using the paper strips. CALLSIGN XPT XFL SSR NS.7.. Short term conflict alert Short term conflict alert (STCA) was available callsigns of aircraft involved turned red, a red speed vector one minute long was displayed..7.. Standard interface Aircraft labels provided standard information, and the mouse allowed for interaction with the interface, and in particular to activate tracker, range & bearing (Figure ). A leaflet, used as a reminder was provided (see Annex ). Label content Mouse buttons Click A to acknowledge AFL (Mode C) SSR Code TOD MSR77-7 Click C extended label and Dynamic Flight Leg Ground speed A Action button B Tracker Range & bearing C Info button Figure Label content and mouse buttons.7.. Airborne spacing interface Specific markings have been developed enabling to display the different states of the airborne spacing process. They consist essentially of markers set around the position symbols of both delegated and target aircraft (permanent display) and of a link between them (conditional display). Three steps have been identified (Figure 9) Target selection markings displayed in orange (to remind the controller s/he still has an instruction to issue). Airborne spacing markings displayed in green (situation to monitor only). End spacing/target de-selection all markings removed. Project AGC-Z-FR - EEC Report No. Volume I

39 COSPACE EUROCONTROL Delegation marking inputs Target selection Click A over delegated Click A over target AZA - DLH - Delegation Click A on delegated AZA - DLH - End delegation or Target deselection Click C on delegated AZA - DLH - Figure 9 Airborne spacing markings and the three steps. Large circle indicates delegated aircraft, small one indicates target. Figure on the link displays current distance or time between aircraft. Project AGC-Z-FR - EEC Report No. Volume I

40 EUROCONTROL COSPACE. DATA COLLECTION, VERIFICATION AND ANALYSIS.. DATA COLLECTION For measurement purposes, two groups of data were collected objective and subjective. Objective data consisted of Aircraft data, pilot inputs (in response to controller instructions) and radio communications recordings; Monitoring and scanning patterns with eye tracking (Figure ) of E-TMA executive controllers. Subjective data consisted of Workload with the Instantaneous Self Assessment (ISA) device and NASA Task Load Index questionnaire; Situation awareness through the use of blank maps; Questionnaires and debriefings items. Figure Eye tracker worn by executive controllers during runs Data collection method, occurrence, relevance and attributes are summarised in Table 9. The list of all data collected is presented in Annex. Table 9 Data collection method and data attributes Occurrence Method/tool Metrics concerned Attributes Pre run data Questionnaires All, but at a high level Subjective Qualitative Observations All human shaping factors, cues about human activity, efficiency and safety Subjective Qualitative Continuous Communication recordings Some shaping factors (workload) Some human activity (coordination) Objective Quantitative data System recordings All, at a detailed level Objective Quantitative Eye tracker Some human shaping factors (workload) Some human activity (monitor situation) Objective Quantitative Periodic data Post run data ISA Some shaping factors (workload) Subjective Quantitative Questionnaires All, at a more detailed level Subjective Qualitative/Quantitative NASA-TLX Some shaping factors (workload) Subjective Quantitative Blank maps Efficiency of some human activity (monitor situation and handle unexpected events) Subjective Qualitative Debriefing All Subjective Qualitative Samples of questionnaires are presented in Annex. Synthesis of the final questionnaires are also proposed in Annex K. Project AGC-Z-FR - EEC Report No. Volume I

41 COSPACE EUROCONTROL The final questionnaire consisted of over 9 items, requesting controllers to choose the most appropriate answers among the ones proposed. Moreover, controllers were encouraged to provide as much comments as possible to justify their answers. They aimed at a subjective evaluation of the following elements Simulation characteristics (familiarity with airspace, traffic and interface, realism of the simulation). Relevance of phraseology. Task analysis (identification of solutions, preparation for spacing instructions, decision of using spacing instructions, communication, monitoring of aircraft, co-ordination with adjacent sectors, task sharing between planning and executive). Benefits analysis (effects and consequences of the airborne spacing). Concept assessment (relevance, usability, impact on capacity and safety). Usability of the marking functions provided on the interface (understandability, ease of use, readability, complexity and suitability for the task)... DATA VERIFICATION Prior to data analysis, a lengthy and tedious data verification process took place. Two points shall be noticed Some problems in data recorded were known (e.g. multiple occurrences of same order) and their correction could have been planned. Other problems (e.g. wrong time stamps) were discovered when analysing data and observing suspicious results (e.g. tails of distributions). Some of them even led to completely re-run the metrics already processed (Figure ). Some problems were linked to the simulation environment (e.g. technical troubles with the simulator, many aircraft handled by one pseudo-pilot), whereas others reflected errors that could occur in real operations. All the problems encountered and the actions performed are listed in Table. Even if the detection of problem could be done automatically, their correction usually requires human analysis. Typically, the correction of multiple occurrences of same order followed two main steps automatically detecting all cases and listening of audio recording to filter non relevant cases (e.g. double clicks on pseudo pilot interface). Out of the 7 cases of multiple occurrences identified for E-TMA, were considered as non relevant and were excluded from data analysis (see Annex 7 for more details). The remaining, corresponding to errors that could occur in real operations, are analysed in section... Project AGC-Z-FR - EEC Report No. Volume I

42 EUROCONTROL COSPACE Sim ulation data Sim ulation error filtering (e.g. m ultiple orders) Bug correction (e.g. time stamp) D ata verification Bug correction (e.g. replay tool) Data analysis extreme values Concept assessm ent Figure Iterative data verification process Table Problems encountered during the data verification process and actions performed Problem encountered Metric verification and design. Redundant orders same order same time, same order, consecutive time, without order in between. Erroneous transfer to frequency. Erroneous time stamp (tick) associated to recorded data; troubles encountered when trying to visualise the data in Art. Audiolan reliability (duration and number of messages). Erroneous geographical positions. Mudpie related problems Consistency of SSR code in Mudpie, Altitude message in Mudpie, Distance displayed on spacing link, Mach speed display on Mudpie, Range and bearing functionality. File loss. Process and actions Understand the objective, understand existing metric, modify, test, process and validate the obtained results (with expert). Detect cases, list them, listen to audio recordings to exclude repeated instructions (enabled to detect the problem of time stamps in Art which made it difficult to locate verbal instructions in the audio recordings), filter remaining doubles. Detect cases, analyse and correct erroneous transfers. Identify discrepancies, investigate the problem, create a TT, correct the problem, test the solution. Data excluded from analysis. In, the analysis of distribution of calls duration and number, enabled to detect too long messages or not equivalent number of messages on the two sides (controller / pilots). Detect and filter when AO/AO/FEED/FEED/AO/AO. Problem raised. Detect loss, look for restore, look for replacement solution, repeat previously run metrics. Project AGC-Z-FR - EEC Report No. Volume I

43 COSPACE EUROCONTROL.. DATA ANALYSIS The data analysis consists in objective and subjective parts. The objective analysis is composed of three parts A first quantitative analysis consisted in processing of simulation data to provide statistical figures. A second quantitative analysis consisting in processing eye movement data to investigate the monitoring tasks (only for E-TMA). A qualitative analysis requiring the involvement of an expert controller to understand controllers strategies and activity. Hours were spent analysing the exercises with the EEC Analysis and Replay Tool (ART) displaying recorded aircraft plots and controllers instructions. Following last year findings, it was decided to exclude the first minutes of the runs from the analysis. Indeed, eye tracker analysis last year showed that the first minutes of the runs, corresponding to traffic building, were not representative enough. Consequently, it was decided to start the analysis once the first aircraft was reaching the exit point. Replay tools helped us defining that the analysis period should start after minutes. The second step consisted in identifying the longest period common to all measured runs. A period of minutes for analysis was thus identified. The subjective analysis is based on different sources of information Questionnaires given to the controllers before, during, and after the simulation (including NASA-TLX rating form). Instantaneous Self-Assessment of Workload (ISA). Blank maps used to assess executive controller situation awareness at the end of each run. Collective debriefings conducted on a daily basis. While the questionnaires and the debriefing sessions investigated multiple aspects (relevance of the method, acceptability of the procedures, safety and workload issues), ISA, NASA-TLX exclusively focused on workload assessment. The main difference between the two subjective techniques used during the simulation is their temporal dimension. Whereas the NASA-TLX collects a global assessment of the workload perceived during an exercise, the ISA technique gathers continuous assessment over the whole exercise. It should be noticed that, for TMA, because of its exploratory nature (compared to E-TMA), the focus was on subjective analysis, and primarily relied on feedback, comments, and ideas coming from the participants... USE OF A BASELINE The experiment design consists in comparing different conditions (without, distance, time spacing) for a given group of controller. The controllers had no experience in airborne spacing (except one who participated during two weeks in ). Results obtained with a controller expert in the use of airborne spacing will be used as a baseline to confront respective results. Project AGC-Z-FR - EEC Report No. Volume I

44 EUROCONTROL COSPACE. RESULTS E-TMA This section is organised as follows factual data and statistical tests are first introduced; then, results are presented along the four dimensions previously introduced human shaping factors, activity, effectiveness and safety. Last of all, a case study of proper and non proper use of airborne spacing is presented. The notation of runs is First letter of sector Controller number ( session ) Spacing condition ( A for AO or AR) ( to ) ( N for without, D for distance and T for time).. INTRODUCTION... Factual data Facts and figures related to the measured runs are presented in Table and Table (for more detailed figures, see Annex H). It should be noticed that, although there were the same number of arrivals, there were more aircraft on frequency in AO due to more overflying traffic. Because of technical problems, the AD ended earlier. Fewer aircraft being controlled during this run, it was decided to exclude it from the analysis. Table Facts and figures regarding the measured runs Measured Analysed Number of runs Hours of control hmn hmn Number of aircraft controlled Flight hours hmn 7h7mn Number of spacing instructions Duration of spacing instructions hmn hmn Project AGC-Z-FR - EEC Report No. Volume I

45 COSPACE EUROCONTROL Table Total number of aircraft on the frequency during analysed period Run Aircraft number AR Aircraft number AO AD 7 AT AN 9 AD AT AN AD AT 9 AN 7 9 AD AT 9 AN 9 9 AD AT AN AD AT AN... Statistical significance Some statistical analyses were derived to study the independence of variables main characters. Because samples are reduced, the non parametric χ test was used. The null hypothesis was H Characters A and B are independent. The χ test provides a decision rule to assess the validity of hypothesis H relative to the independence of the two characters A and B. Let ˆχ be the value of the observed criteria, χ α(ν) the critical value at α level of significance and ν degree of freedom. If ˆχ <χ α(ν) the null hypothesis can not be rejected at level α. If ˆχ >χ α(ν) less than α times out of would give this value of ˆχ even though variables are independent; this means that characters A and B are dependent at level α. The following hypothesis were tested Are the type of instructions and the sessions independent? Are the geographical repartition of instructions and the sessions independent? Are the number of spacing instructions given and the sessions independent? Project AGC-Z-FR - EEC Report No. Volume I 7

46 EUROCONTROL COSPACE Detailed results of the statistical tests are presented in Annex I. Summarised results are as follows The type of instructions (spacing/heading/speed) and the sessions are dependent with a probability of.9. For most of the sessions the influence of the team is not significant at a level of.. The geographical repartition of the instructions and the sessions are dependent with a probability of.9. The number of spacing instructions given and the sessions are independent. The influence of the teams on the number of spacing instructions is not significant at a level of.. Note statistical results were obtained too late to influence the results analysis. Consequently, some summary results presented in the next sections shall be considered carefully... HUMAN SHAPING FACTORS... Usability... Subjective feedback Controllers all rated the airborne spacing as mostly (/) or completely (/) understood. For all of them, airborne spacing is compatible with their working method and with the sequencing tasks. The major critic was related to the difficulty and the cost to integrate late traffic in chains of airborne spacing instructed aircraft.... Rate and duration of use Since controllers were not forced to use airborne spacing (but rather invited to if they felt it could be helpful), we considered the rate of use as an indicator of usability a low rate could reflect either difficulties or reluctance to use airborne spacing. We analysed both the percentage of aircraft involved in airborne spacing, and the duration of airborne spacing. The rate of use was obtained in comparing the number of aircraft instructed to the total number of concerned 7 aircraft. To reflect the duration of airborne spacing for a given controller, in charge of a given sector, we decided to consider the time aircraft spent on frequency, rather than in the geographical sector. The duration was obtained in comparing the duration of flight under airborne spacing to the total duration of flight. According to our baseline, the maximum possible rate of use for a controller expert in airborne spacing is in distance based condition % in AO and 7% in AR, in time based condition 7% in AO and 7% in AR. Table and Figure show that in distance based, the rate of use is % in AO and 7% in AR. In time based, the rate of use is 7% in AO and 7% in AR. In both conditions (distance and time) results show a higher rate in AR than in AO. Note that in time based, the maximum rate of use was similar in both sectors. 7 Concerned aircraft correspond to the arrival aircraft, i.e. overflights excluded. Project AGC-Z-FR - EEC Report No. Volume I

47 COSPACE EUROCONTROL Regarding the duration of use, results show that spacing instructions lasted longer in AR than in AO (Table and Figure ). In distance based, it lasted % in AO and % in AR. In time based, it lasted % in AO and % in AR. Compared to time based, distance based spacing was longer in AO, but of comparable duration in AR. The trends are in line with our baseline figures. Results are in line with previous experimental results. However, it shall be stressed that even though present results suggest a rate higher than our expert's, the quality of control needs to be checked. If the rate of use reflects controller willingness to use the instructions, and possibly confidence in the benefits provided, it might not reflect correct use but possibly cases of over use. One mean to address efficient use is to consider the respect of initial applicability conditions (see the speed variation metric in...). Table Spacing instructions rate of use Use of Distance Time airborne spacing (%) higher mean baseline higher mean baseline AO AR Table Spacing instructions mean duration Duration of Distance Time airborne spacing higher mean baseline higher mean baseline AO 7 7 AR Mean use of delegation % Aircraft instructed / concerned aircraft (Duration of analysis hmn) Distance Time Mean percentage of use AO AR Mean percentage of use AO AR Mean percentage of time Mean time under spacing/ Mean flight time (%) (Duration of analysis hmn) Distance Time AO AR Mean percentage of time AO AR Figure Mean use of airborne spacing number of spacing instructions compared to number of concerned aircraft. Figure Mean duration of airborne spacing duration of airborne spacing compared to flight time of aircraft under airborne spacing. Project AGC-Z-FR - EEC Report No. Volume I 9

48 EUROCONTROL COSPACE... Type of spacing instructions used To identify the appropriateness and possibly relative ease of use of each application (remain, merge, heading then remain, heading then merge), we analysed their respective rate of use (Figure ). The remain instruction was nearly never used. This could be explained by its restrictive applicability conditions typically, to be used aircraft should be already spaced on the same trajectory. Whereas it could be useful and usable to maintain spacing within a single existing flow, it is less usable for sequence building, where converging flows need to be integrated. The most used application was the merge (more than 97% of the cases), while the heading then remain was not used at all. Reasons evoked to explain the limited use of the heading then merge instruction are the perceived unpredictability of the resuming trajectory and the lack of control they feel with this instruction. Because they felt the merge application was easier to use, controllers preferred to vector the traffic, decide of the appropriate time to resume and then give a merge instruction. However, it should be stressed that when applicability conditions are correct, heading then merge instructions enable less time-critical instructions to be issued. Given the initial conditions and heading instruction, controllers should know quite precisely where the aircraft should resume. The controller task should not correspond to assess where the aircraft will resume, but rather ensure that it is resuming where expected. This point is strongly related to skills, workload and situation awareness issues. Table Spacing instructions respective rate of use Remain Merge Heading then Remain Heading then Merge Distance Time Both AO AR Total AO AR Total Baseline % 97% 7 97.% % 7 9% %,% % Total 7 The link with skills is in terms of understanding the aircraft expected behaviour in correct conditions and feeling the similarity with existing practices. The link with workload is in terms of cognitive cost we assume that it is more demanding to observe and interpret situation rather than compare it with expected events. The link with situation awareness is in terms of predictability. Project AGC-Z-FR - EEC Report No. Volume I

49 COSPACE EUROCONTROL 9 7 Number of instructions Application usage - Distance 9 7 Number of instructions Application usage - Time AO AR AO AR Figure Respective rate of use of all applications in distance (left) and time based (right) To assess difficulties in setting up or assessing applicability conditions, we looked for spacing instructions cancelled, modified or not given (despite target selection phase). The automatic detection of such cases was complemented with expert analysis of each case in order to filter normal end of spacing instructions. Figure shows that in distance based conditions, the sector has an impact on the number of spacing instructions initially planned (target selected) and finally not implemented. It shall be noted that % actually correspond to occurrences instead of in the other conditions. Distribution in time does not show any impact of training the number of cases did not decrease with time. Target selection not followed by a spacing instruction (%) 9 7 AO Distance Time AR Figure Percentage of target selection not followed by a spacing instruction Synthesis on usability The high rate of use reflects the usability of the spacing instructions, and more specifically the merge instruction, in both conditions. Controllers felt they had difficulties to assess initial applicability conditions or to maintain them, which led to cancelling or modifying ongoing spacing instructions. Hypothesis U, U and U are confirmed. Project AGC-Z-FR - EEC Report No. Volume I

50 EUROCONTROL COSPACE... Motivation We considered as indicators of motivation to use the new instructions controller feedback on usefulness, on their understanding and the compatibility with current practices. For most of the controllers ( out of answers) the spacing instructions are considered as generally or totally useful. The most useful applications is the merge ( out of ), whereas the heading then is considered as the least useful by all controllers. For most of them, airborne spacing is a workload reduction (/) and/or a stimulant (/) rather than a loss of control (/) or a concern (/). Figure shows that controllers gave most of the time (i.e. %) one spacing instruction per aircraft. Analysis of multiple spacing instructions per aircraft showed that they correspond to situations when a spacing instruction had to be cancelled (e.g. integration of a late flow in AO) and a new one given. The modification of a spacing instruction could indicate controller motivation in using spacing instructions. (as opposed to fall back to using conventional instructions). Proportion of single and multiple spacing instruction AO AR Both instruction More than instruction Figure Proportion of single and multiple spacing instructions per aircraft Synthesis on motivation Controllers perceived the spacing instructions as useful, beneficial in terms of workload reduction and compatible with their current working methods. They extensively used the spacing instructions (7% rate of use, for about % of the flight time), and more specifically the merge instruction, in both distance and time conditions. Hypothesis M is confirmed. Hypothesis M and M are not confirmed. M was not tested.... Workload... Overall perceived workload Feedback gathered in the questionnaires is that for out of the controllers, using spacing instructions induces a workload reduction. All controllers consider that the most demanding application is the heading then because it requires them to wait for delegated aircraft to resume before using it as a target. The benefit gained in relieving controllers from identifying the resume point (and reducing risk of omission) seems limited since controllers have to wait for aircraft to resume before using them as target. Project AGC-Z-FR - EEC Report No. Volume I

51 COSPACE EUROCONTROL For out of the controllers the most demanding tasks are assessing applicability conditions and integrate flows of aircraft. We asked the controllers to rate between very easy and highly complex the effort to perform the following airborne spacing-related tasks identify when a situation becomes delegable; make a situation delegable (set up applicability conditions), initiate the airborne spacing process (select target), communicate spacing instructions, monitor instructed aircraft, remember current instructed aircraft, end spacing instructions. For out of the controllers, the most demanding airborne spacing-related task is making a situation delegable. With airborne spacing, out of the controllers consider communications were less frequent but out of them feels they were longer. However, out rate the communication load as lower or much lower. out of the controllers rated the monitoring as not stressful and less frequent with spacing instructions.... Instantaneous Self Assessed Workload ISA recordings provide both summary and temporal analysis of workload for both executive and planning controllers. Five levels of workload are defined (very low, low, fair, high and very high). Example of the recordings for a same team in the three conditions (Figure 7), showing a slight decrease of workload in time based for both the executive and the planning controllers. It also shows higher level of perceived workload for the planning controller 9. Detailed ISA recordings correspond to the executive controller (Figure ). Note that the absence of input could be caused either by a very high workload or simply a lack of attention. % Executive and planning controller workload (ISA) AR/EXC AR/PLC No AR/EXC AR/PLC Distance Very High High Normal No Answer Low Very Low AR/EXC AR/PLC Time Figure 7 ISA summary recordings for a same team in three conditions 9 Note that it is difficult to compare the workload perception between controllers, as despite fixed definition of the ratings, each individual has his/her own definition. Project AGC-Z-FR - EEC Report No. Volume I

52 EUROCONTROL COSPACE Project AGC-Z-FR - EEC Report No. Volume I Sector AR Executive controller AT Nb. of aircraft AD Nb. of aircraft AN Nb. of aircraft Sector AR Executive controller AT Nb. of aircraft AD Nb. of aircraft AN Nb. of aircraft Figure Examples of temporal ISA recordings in the three conditions for an executive controller Results show that with distance, 9% of controllers (whatever their position) feel either no change in their workload or a workload reduction. In time based, % of the planning controllers and 7% of the executive controllers feel a workload reduction or no change. The analysis of ISA ratings shows only a few cases of negative impact of airborne spacing on workload. Negative impact only represents % of the exercises in distance based and 7% in time based. Most of them concerned executive controllers ( out of cases). No impact of sector on workload was identified. Table summarises the observations and Table 7 proposes a synthesis of the results. Table Impact of airborne spacing on workload assessed with the ISA device. Arrows reflect workload evolution compared to baseline. Bold arrow reflects large change, light arrow slighter change and dash no change. AO AR Traffic sample Session Position Distance Time Distance Time EXC A PLC EXC A PLC EXC A PLC EXC A PLC EXC A PLC EXC A PLC

53 COSPACE EUROCONTROL Table 7 Synthesis of ISA ratings, per type, sector and position. Note that figures show that sector has no impact, whereas position and type of airborne spacing have. AO AR EXC PLC Total % D T D T D T D T D T D T Same Decrease 7 Increase 7... Post run mental and temporal NASA-TLX ratings Even though controllers assessed six dimensions (physical, mental, temporal, performance, effort and frustration), we focused our analysis more specifically on mental and temporal demands. Mental demand (Figure 9) the use of airborne spacing does not change mental demand. For both controllers, in both distance and time conditions and in both sectors, the workload is perceived as similar. Planning controller mental demand is rated lower than the executive controller mental demand. Lots of effort Rather effortful Some effort Little effort No effort NASA-TLX Mental demand No Distance Time EXC PLC Lots of effort Rather effortful Some effort Little effort No effort NASA-TLX Mental demand (AO) No Distance Time Lots of effort Rather effortful Some effort Little effort No effort NASA-TLX Mental demand (AR) No Distance Time EXC PLC EXC PLC Figure 9 NASA-TLX mental demand for both sectors (top), AO (left) and AR (right) Temporal demand (Figure ) compared to conventional situation, the use of airborne spacing does not modify temporal demand for the executive controller. Time based spacing seems to induce a slight increase in temporal demand for the planning controller. The temporal demand is similar in both sectors. Apart in time based condition, where both executive and planning controller temporal demands are similar, in other conditions the executive controller temporal demand is higher than the planning controller s. Project AGC-Z-FR - EEC Report No. Volume I

54 EUROCONTROL COSPACE NASA-TLX Temporal demand No Distance Time EXC PLC NASA-TLX Temporal demand (AO) NASA-TLX Temporal demand (AR) No Distance Time No Distance Time EXC PLC EXC PLC Figure NASA-TLX temporal demand Both sectors (top), AO (left) and AR (right) NASA-TLX ratings for both mental and temporal demand suggest that airborne spacing does not modify the executive controller load, whereas time based spacing seems to slightly increase the planning controller load. It is possible that the increased workload is related to the difficulties to update situation awareness, in particular in situations where it is not intuitive to assess visually spacing in time. The executive controller being directly involved in the situation might encounter less difficulty to update situation awareness. However, the large standard deviations show large inter-individual differences, thus preventing to draw any definitive conclusion.... Communication load To assess the workload induced by communication, we could have considered the frequency occupation, in terms of number and duration of messages and take into account both instructions and information requests. However, because we questioned the reliability of audio recordings, we limited the analysis to the number of instructions given, that were reflected by the number of pseudo pilots inputs.... Number of manoeuvring instructions The counting of manoeuvring instructions in all conditions suggests that airborne spacing induces an overall reduction, in both sectors and in both conditions (distance and time based) the total number of instructions is reduced by % (Figure ). When considering the select target instructions (Figure ) we still notice a % reduction of the overall number of instructions given. The observed trends are in line with last years results, even if compared to baseline figures, we expected larger benefits (% reduction without the select target instructions and % with them). Project AGC-Z-FR - EEC Report No. Volume I

55 COSPACE EUROCONTROL Mean number of instructions Baseline ALL AO AR AR No Distance Time Figure Overall number of instructions, including spacing instructions, excluding select target Mean number of instructions Baseline 9 9 ALL AO AR AR No Distance Time Figure Overall number of instructions, including spacing and select target instructions Synthesis on workload Answers to questionnaires show that controllers perceive a workload reduction when using spacing instructions. ISA and NASA-TLX ratings suggest that spacing instructions do not modify the executive controller mental and temporal demand, while time based spacing seems more demanding for the planning controller. Using spacing instructions induces a % reduction of the manoeuvring instructions. Monitoring load is also perceived as lower in both conditions. Hypothesis W, W, W are confirmed. W is not confirmed. W was not tested.... Skills Because airborne spacing is based on current practices, we did not expect new skills to be required, except conditions of use of each instruction and dedicated phraseology. Training was used to present the environmental set-up, the new instructions and help controllers understanding the similarity with current practices. Regarding the phraseology for spacing, out of the controllers felt it was generally understandable (or very clear) and had no difficulty to learn it. All of them rated the time needed to learn it as acceptable and qualified their familiarity with it at the beginning of the measured runs as quite acceptable. They all considered they were often able to comply with it. Some asked for some instructions to be shortened. Project AGC-Z-FR - EEC Report No. Volume I 7

56 EUROCONTROL COSPACE Even though some of the controllers were not familiar with interacting with the interface, they did not have difficulty marking the aircraft under airborne spacing on screens. They describe the marking function as easy to use. The main critics regarding interface are not related to airborne spacing (e.g. label overlapping). During debriefing sessions, controllers evoked the difficulties related to the visual perception of time spacing. Similarly to previous simulation, we noticed that controllers had difficulties with the applicability conditions. Even though they are performing similar tasks today (e.g. assess aircraft respective performances, current speed, position), they sometimes failed to perform them with airborne spacing. We identified two possible reasons either they transfer the applicability condition assessment to the flight crew (due to concept misunderstanding and/or over expectations) or they need more practice and time to realise the similarity between their current practice and the spacing instructions (skills assimilation still in progress). To assess controller ability to identify applicability conditions, we considered the speed variations following the spacing instructions. Appropriate applicability conditions should not induce large speed variations (e.g. greater than knots). When considering the distribution of mean number of speed variations as a function of order of run played (Figure ), we notice a learning effect with practice, the number of acceptable variations increased from to, whereas the number of excessive reductions decreased from to. Speed variations s after instruction Runs measured < kts OK > kts Figure Impact of practice on speed variations following spacing instructions. Mean for all controllers and all sectors, as a function of run (from to ). Synthesis on skills Phraseology and interface markings were easy to learn. However, controllers had difficulties with respecting applicability conditions. Even though controllers should be using the same skills, they did not. After days of training, controllers were still learning how to use efficiently the instructions. This could be explained by the reduced support provided during the training period, due to TMA session running in parallel. Hypothesis Sk and Sk are confirmed, Sk is not. Sk was not assessed. Project AGC-Z-FR - EEC Report No. Volume I

57 COSPACE EUROCONTROL... Teamwork It shall be reminded that even though one task distribution was suggested (same as today and graphical markings handled by the executive controller), it was left up to the controller preferences. Apart from one controller who felt the planning controller role would change with airborne spacing, all controllers felt airborne spacing would not change executive nor planning controller respective roles. Typically, they expected from the planning controller some support in terms of Situation assessment, including aircraft equipage and performance compatibility. Initial identification of potential spacing instructions. Monitoring of aircraft descent. Suggestions regarding the marking task diverge controllers suggest that markings are input by the executive, whereas the others suggest the task to be performed by the planning. Indeed, we could observe during runs that the task distribution could vary, not only depending on the traffic load, but also on controller preferences. Some executive controllers relied completely on their planning controllers, whereas other preferred marking themselves the aircraft under airborne spacing (Figure ). Note that no impact of the type of airborne spacing (distance / time) was identified. Marking task repartition EXC / PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC EXC PLC AO AR AO AR AO AR AO AR AO AR AO AR A A A A A A Target selection Spacing instruction Figure Marking task repartition between executive and planning controllers Marking aircraft under airborne spacing helps both controllers keeping an up to date knowledge of the situation. However, in very high traffic load, it happened that the controller in charge of the markings had difficulties to cope with it. The result was a loss of situation awareness for both controllers. Results show that under pressure it is difficult for the planning controller to keep track of the executive decisions. These results lead us to strengthen the attribution of marking task it should be an executive task, similar to strips updating. Synthesis on teamwork The use of spacing instructions does not change the current task distribution between executive and planning controllers. Support from the planning controller (e.g. for situation assessment) is strongly requested in both conditions. Attribution of marking tasks to either executive or planning controller depended on controllers, however for efficiency and safety it should be attributed to the executive controller. Hypothesis T and T are confirmed. Project AGC-Z-FR - EEC Report No. Volume I 9

58 EUROCONTROL COSPACE... Synthesis on human shaping factors The rate (over 7% of aircraft) and duration of use (% of the flight time) of the spacing instructions suggest that controllers accepted to work with these new instructions. The most used instruction was the merge. ISA and NASA-TLX ratings suggest that spacing instructions do not modify the executive controller mental and temporal demand, while time based spacing seems more demanding for the planning controller. The use of spacing instructions induces a % reduction of the number of manoeuvring instructions... CONTROLLER ACTIVITY... Manage safe and expeditious flows of traffic... Types of manoeuvring instructions To understand impact on controller strategy, the first step consists in analysing the type of manoeuvring instructions used. Three types are considered heading (including direct), speed and spacing instructions (when applicable). The use of spacing instructions in time and distance conditions induces a significant reduction of speed and heading instructions (respectively 7% and %). In AR, in both conditions, with 7% rate of use, the reductions of speed and heading instructions are respectively % and %. In AO, in time, with 7% rate of use, the reductions of speed and heading instructions are respectively 7% and %. In distance, with % rate of use, the reductions are respectively % and %. It should be noticed that the merge instruction includes a direct instruction. 7 AR 7 AO 9 7 Heading Speed Spacing Heading Speed Spacing No Distance Time No Distance Time Figure Types of manoeuvring instructions in the three conditions (without, distance and time). AR (left) and AO (right). Project AGC-Z-FR - EEC Report No. Volume I

59 LFPN LFPN LFPN LFPO LFPO LFPO MOLEK MOLEK MOLEK OMAKO OMAKO OMAKO OKRIX OKRIX OKRIX INKAK INKAK INKAK ATN ATN ATN LFPN LFPN LFPO LFPO MOLEK MOLEK OMAKO OMAKO OKRIX OKRIX INKAK INKAK ATN ATN COSPACE EUROCONTROL... Geographical distribution of instructions To go further in the understanding of impact on controller activity, we plotted instructions over the sector. Maps show the same trend as in previous experiments and baseline (Figure ). Session A Without delegation Session AN (7 - ) COSNOV Baseline Without delegation Session AN (7 - ) COSNOV LFPB LFPG LFPB LFPG LFPN LFPO OMAKO MOLEK INKAK OKRIX ATN Without Distance delegation Session AD (7 - ) COSNOV Without Distance delegation Session AD (7 - ) COSNOV LFPB LFPG LFPB LFPG Time delegation Session AT (7 - ) COSNOV Distance Time delegation Session AT (7 - ) COSNOV Distance LFPB LFPG LFPB LFPG Time Time Figure Mapping manoeuvring instructions over sectors. Example from the session A (left) and baseline (right). As defined in previous experiments, a more synthetic view of the geographical distribution of instructions is obtained by representing instructions as a function of distance to exit point (Figure 7). This geographical distribution of manoeuvring instructions reflects controller strategies. Typically, in conventional control, it enables the distinction between the sequence building (heading instructions) and the sequence maintaining (speed instructions). Project AGC-Z-FR - EEC Report No. Volume I

60 EUROCONTROL COSPACE Results show that airborne spacing modifies flow management in reducing the number of instructions used, but does not modify controllers strategies. Heading (or equivalent spacing instructions) are used to build sequences, whereas speed adjustments are used to maintain them (even if they are no longer visible at controller level, but should be detected on the flight crew side). In other words, with airborne spacing, controllers remain in charge of the strategy, and task flight crew to implement it. These results are comparable to our baseline observations and in line with our previous experiments the main reduction of instructions still occurs in the second part of the sectors. The sector configuration has an effect the impact of airborne spacing is more noticeable in AR than in AO. In AR, with spacing instructions, controllers are still using early heading instructions to build the sequences (between and Nm from the IAF). Then, from Nm from the IAF, only few speed instructions are used to maintain sequences. In AO, using spacing instructions also induces a reduction of speed instructions in the second part of the sector. However, heading and spacing instructions are still used in the second part of the sector. This is due to the late integration of Eastern traffic (on OKRIX) leading to either cancel ongoing airborne spacing or delay issuing spacing instruction. In both sectors, there are more instructions near the exit point in time than in distance. It reflects recovery from spacing-related errors, possibly due to the difficulty to assess spacing in time. Compared to our baseline, we noticed some differences in the three conditions Instructions are given very close to IAF; Speed and heading instructions are given passed the building area (after NM from the IAF). These situations were analysed with the replay tool. Most of the time, the late manoeuvring instructions reflect cases of spacing instructions cancelled and fallback to conventional control. This was due to applicability conditions initially not respected or no longer fulfilled. Typically, in case of predicted spacing much lower than required, the instructed aircraft would reduce its speed within its envelop limits. Sooner or later the controller will realise that the aircraft will not make it. To recover from this situation, the spacing instructions will be cancelled and the aircraft vectored to increase the spacing. Depending on the situation, if the aircraft was itself part of a long chain, the whole sequence might need to be modified. Hence additional instructions given late in the sector. Project AGC-Z-FR - EEC Report No. Volume I

61 COSPACE EUROCONTROL Mean number of instructions Mean number of instructions Mean number of instructions No (All sessions) Distance ( All sessions) Time (All sessions) AR Sector AO Sector Distance to IAF (nm) Distance to IAF (nm) Distance to IAF (nm) Distance to IAF (nm) Mean number of instructions Mean number of instructions Mean number of instructions No (All sessions) Distance (All sessions) Time (All sessions) Speed Heading Spacing Distance to IAF (nm) Distance to IAF (nm) Figure 7 Mean geographical distribution of instructions. AR (left) and AO (right), without (top), with distance (middle) and time spacing (bottom). Comparing the occurrence and duration of sequence building shows that with spacing instructions, in both time and distance conditions (Figure ) sequences are built earlier (NM instead of NM from the IAF). No Distance Time Mean number of instructions Mean number of instructions Mean number of instructions Distance to IAF (nm) Distance to IAF (nm) Speed Heading Spacing Distance to IAF (nm) Figure Anticipated sequence building. Example with two sessions (A and A). Project AGC-Z-FR - EEC Report No. Volume I

62 EUROCONTROL COSPACE... Temporal distribution of instructions The traffic were designed so that the load would be quite equivalent all along the run. We expected airborne spacing to help controllers smoothing their own activity typically, peaks of instructions could have been replaced by more homogeneous instructions giving. Similarly to previous experiments, the use of spacing instructions does not seem to modify the distribution of instructions in time (Figure 9). In all conditions, and in both sectors, spacing instructions induce a general reduction of instructions, all along the run. The number of spacing instructions is quite equivalent every minutes. The main differences are related to conventional manoeuvring instructions. Additional analysis helped identifying run per run what led controllers to give more numerous instructions (e.g. th minute, time condition, AR sector). Difficulties in assessing and maintaining applicability conditions led to spacing instructions cancellation followed by the use of additional conventional instructions to recover manually the situation. Mean number of observations Mean number of observations AO No Distance AR No Distance Mean number of observations Time Time interval (mn) Time Time interval (mn) Standard instructions Spacing instructions Figure 9 Temporal distribution of instructions Synthesis on flow management-related activity With spacing instructions, manoeuvring instructions are reduced speed instructions are reduced up to % and heading instructions are reduced by %. Reduction of manoeuvring instructions is observed in both sectors and both conditions. Geographical distribution of instructions show that using spacing instructions leads to anticipate sequence building and controllers are partly relieved from maintaining the sequences. No smoothing of activity was observed when looking at temporal distribution of manoeuvring instructions. Hypothesis Seq and Seq are confirmed. Seq could not be confirmed (activity smoothing). Project AGC-Z-FR - EEC Report No. Volume I

63 COSPACE EUROCONTROL... Monitor and analyse traffic situation... Collecting information Subjective feedback from controllers was collected via questionnaires items. Controllers thought airborne spacing modified their monitoring, typically in reducing its frequency ( controllers out of ). They all felt their monitoring was more global (as opposed to focused on individual aircraft) and less complex. No difference was felt between the spacing condition (distance versus time). Aircraft under airborne spacing were perceived as less often monitored. However, in very high traffic load situation, controllers could no longer update all markings either aircraft remained in target selected mode (orange marking) or were not marked at all. This led to difficulties to maintain an up to date awareness of the current situation. This stresses the importance of markings in terms of support for the monitoring, and the need for consistent information. Eye movement tracking technique provided more objective evaluation of the impact of airborne spacing on monitoring. Unless mentioned, general results are statistically significant (Chi-test, p<.). Summary results are presented in this section.... General characteristics of fixations The initial investigation consisted in measuring the number and duration of fixations inside the radar screen as opposed to outside. In the AR sector (one converging area), in the three conditions, the controllers spent more than 7% of their time monitoring the radar (Figure ). In the AO sector (one early and one late converging areas), controllers spend % of their time monitoring the radar screen. However, percentage of time spent on the radar screen is more dependent on controllers, than on the condition (see standard deviations). Fixations inside the radar (mean) Fixations inside the radar % % % % % % % % ND TD DD % % % % E F C A B D Baseline ND TD DD Fixations inside the radar (mean) Fixations inside the radar % % % % % % % % ND TD DD % % % ND TD DD % A B D E F C Figure Percentage of fixations on the radar (working area only). As a function of the experimental conditions (left), as a function of controller (right), AR (top) and AO (bottom). Project AGC-Z-FR - EEC Report No. Volume I

64 EUROCONTROL COSPACE... Geographical distribution of eye fixations We investigated if, similarly to what was observed with instructions, there was a correlation between number, duration and geographical position of fixations. To address the overall monitoring, we analysed the geographical distribution of fixations. The mean monitoring curves for AO and AR (Figure ) have similar shapes. The use of spacing instructions leads controllers to focus their attention (or at least their monitoring) on the first part of the sector. This is an area requiring careful analysis of the situation. In AO, even though the focus is on the first part of the sector, controllers still need to monitor the second part, to take into account flows arriving from the East, to be integrated in this second building area. Mean values do not show significant differences between the two spacing conditions (distance versus time). % AR - Mean value % AO - Mean value % % % % % % % % % % % % % % % % % % ND DD TD ND DD TD Figure Mean fixations in the three conditions. AR (left) and AO (right). A more synthetic view of the distribution consists in gathering fixations according to two main areas corresponding to the sequence building (-NM from IAF) and the sequence maintaining (- NM from IAF). The global figures show that the use of spacing instructions (in both distance and time conditions) induces a positive change in the main area of interest. With spacing instructions, controllers focus on the building area (% of fixations) whereas without spacing instructions their fixations are more numerous over the maintain area. The same trend is observed when considering sectors. Project AGC-Z-FR - EEC Report No. Volume I

65 COSPACE EUROCONTROL However, whereas the trend is very noticeable in distance based in AR, it is more noticeable in time based in AO. 7 Fixations distribution (AR sector) Maintain [-NM[ 7 Build [-NM[ Fixations distribution (AO sector) 7 9 No Distance Time Maintain [-NM[ Build [-NM[ Fixations geographical distribution 7 7 Maintain sequence (-NM) No Distance Time Build sequence (-NM) Inter-individual differences Figure Synthetic view of geographical distribution of fixations. AR (top left), AO (top right) and both sectors (bottom). The comparison between controllers led us to identify diversity in the monitoring curves. In both sectors, results show cases of neutral impact and positive impact of spacing instructions (in terms of focus on early part of the sector). We did not find curves showing a negative impact of spacing instructions. It is interesting to notice that similar patterns were found for each controller, whatever the sector monitored. Typically, when a positive impact of spacing instructions is noticed for a controller on a given sector, it is also observed in the other. Similarly, when observed for a given controller, the absence of impact took place in both sectors. Positive impact concerns controllers and the absence of impact the others. Disparity between two types of controllers may be explained by individual differences in learning process. Project AGC-Z-FR - EEC Report No. Volume I 7

66 EUROCONTROL COSPACE % AR % AR % % % % % % % % % % % % % % % % % % % % % % % AO % AO % % % % % % % % % % % % % % % % % % % % % % Distance to IAF Distance to IAF No Distance Time Figure Two examples of monitoring curves. Positive impact (left) and no impact of spacing (right). AR (top) and AO (bottom). Intra-individual differences Beyond the comparison between controllers, we analysed the monitoring curves all along the runs. Typically, we selected periods of minutes each and investigated if the monitoring curves were changing during this period. For illustration purpose, examples of three periods within a same run are presented for the AR sector (Figure ). When considering each condition (without, with time and with distance based), it appears that changes are more important without the spacing instructions. Whereas initially (first minutes) controllers were focusing on the transfer area (i.e. reactive position), they managed to focus on earlier area after minutes. It looks like situation is degrading during the last minutes. For distance based, curve seems to show that controllers are focusing on earlier areas with time. It suggests that spacing instructions gradually enable them to focus where sequences need to be built. With time based, the example shows that whereas spacing instructions also enabled the focus to be put on earlier area, the situation seems to degrade during the last minutes. Project AGC-Z-FR - EEC Report No. Volume I

67 COSPACE EUROCONTROL AR Example - [- minutes] AR - Example - [- minutes] AR Example - [- minutes] % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % No Distance Time Figure Example of monitoring curve for three periods of minutes each Global monitoring data seem to show that without spacing instructions, when confronted to the very high level of traffic, controllers are in a more reactive position, the building of sequences being no longer anticipated. With spacing instructions, both distance or time based, controllers seem to be focusing over the most important area, where sequences need to be built.... Contextual monitoring To try understanding what triggers controller monitoring, we performed a contextual analysis of fixations. The reduced monitoring in the second part of the sector with airborne spacing raised the issue of safety in terms of frequency of monitoring per aircraft. Typically, the questions to be investigated were are aircraft still monitored once spacing instructed? Are changes equally impacting all aircraft? To answer them, we analysed the frequency of monitoring of each aircraft, i.e. the interval between two consecutive fixations on a same aircraft. Previous results (Zeghal et al., ) suggested that controllers were fixating between groups of related aircraft rather than on each of them. We assume that the foveal perception of the area between aircraft corresponds to parafoveal perception of each aircraft. Studies on vision (Schilling et al, 99) showed that visual perception was based on the combination of foveal and parafoveal vision. In the foveal region ( degree of visual angle to the left and right of fixation), acuity is the sharpest. In the parafoveal region (extending to degrees of visual angle on either side of fixation) and in the peripheral region (everything on the line beyond parafoveal region) acuity drops off markedly so that our ability to identify letters is not very good even in the near parafovea. According to Rayner (99), the purpose of eye movements in reading is to place the foveal region on the part of text to be processed. Similarly, we assume that in the case of controllers, foveal vision corresponds to active monitoring and parafoveal vision to peripheral monitoring. As opposed to the global analysis of fixations, the contextual one required a fine synchronisation between air traffic events and eye measures. Spatial and temporal synchronisation was ensured through controller mouse clicks on reference beacons every minutes. An additional post processing synchronisation took place later. From these files, statistical computation was realised. In addition to analysing the distributions of fixations on the radar, we examined the frequency of monitoring of each aircraft, depending on its status not involved in a spacing instruction, having selected a target or spacing instructed. Project AGC-Z-FR - EEC Report No. Volume I 9

68 EUROCONTROL COSPACE The calculation of fixations per aircraft was made by allocating to every fixation one or several concerned aircraft. For each aircraft, we calculated the mean number of fixations as a function of their type (foveal versus parafoveal). A similar pattern is observed in all conditions, including the baseline run (Figure ) parafoveal fixations are twice more numerous than foveal fixations. Distribution of foveal and parafoveal fixations N D T N D T N D T N D T AO AR Both Baseline (AR) Foveal Foveal and Parafoveal Figure Distribution of number of foveal and parafoveal fixations Beyond the number of fixations in each visual area, we then investigated the period between two successive fixations on a same aircraft. For each aircraft, we displayed every foveal only and parafoveal fixations in time (Figure ). Then, inter-fixations periods were computed. Timeline (minutes) Enter (frequency) Foveal Parafoveal (when overlapping) Inter (parafoveal) fixation period Exit (frequency) Figure Example of temporal distribution of foveal and parafoveal fixation per aircraft For all runs, the mean and maximum fixations periods were calculated (Figure 7). In both sectors and in all conditions parafoveal fixations were more frequent than foveal fixations. For individual aircraft, this could correspond to some directed fixations (foveal regions) supporting decision making and many peripheral information taking when monitoring the overall situation. Project AGC-Z-FR - EEC Report No. Volume I

69 COSPACE EUROCONTROL A complementary indicator would be the temporal analysis of occurrence of each type of fixation for a given flight we could assume that foveal fixations would occur mainly during the sequence building phase (sector entry) while parafoveal fixations would occur all over the sector. Distribution of mean maximum period in foveal and parafoveal regions Distribution of max maximum period in foveal and parafoveal regions N D T N D T N D T mean N D T N D T N D T mean AO AR Both AO AR Both parafoveal and foveal foveal parafoveal and foveal foveal Figure 7 Distribution of mean maximum inter fixation period in foveal and parafoveal regions Figure Distribution of max maximum inter fixation period in foveal and parafoveal regions Summary results (Figure 9) confirm the trend. They also suggest that mean mean and mean maximum periods are very close. Mean and max maximum inter fixation period in foveal and parafoveal regions mean_mean mean_max max_max parafoveal and foveal foveal Figure 9 Mean and max inter fixation periods in foveal and parafoveal regions Project AGC-Z-FR - EEC Report No. Volume I

70 EUROCONTROL COSPACE We assumed that foveal fixations would occur mainly during the sequence building phase (sector entry) and parafoveal fixations all over the sector. However, initial results do not confirm this hypothesis (Figure ). Percentage (%) Geographical distribution of parafoveal and foveal fixations 9 7 Build sequence Maintain sequence Parafoveal Foveal Figure Geographical distribution of parafoveal and foveal fixations. AR sector. We also looked at the distribution of the maximum periods (Figure ). In addition to confirming the mean maximum period, this step enabled us to identify atypical cases of very long inter fixations period. Extremely long periods could raise safety issues. Typically, if confirmed by extended analysis, seconds between two successive fixations on an aircraft would suggest that it was not monitored during more than minutes, which seemed doubtful. Number of occurrences Maximum duration distribution (foveal + parafoveal) Period duration (seconds) Not spacing instructed Target selected Spacing instructed Figure Example of distribution of maximum periods of fixation. With distance based spacing. AR sector. Project AGC-Z-FR - EEC Report No. Volume I

71 COSPACE EUROCONTROL To distinguish real cases of aircraft omission from analysis-related errors (e.g. data loss), we looked at every aircraft whose fixation period was greater than seconds in fovea region and seconds in fovea and parafovea region. cases were identified. Each case was analysed using replay tools. cases could not be analysed for technical reasons. cases correspond to clean situations. Less frequent monitoring is then acceptable. Typically, aircraft concerning the other sector and not interfering with other aircraft are less frequently monitored. 9 cases raise issues in terms of eye tracker data reliability. Typically, it happened that no fixation was detected on some aircraft when receiving instruction, or even when being graphically marked. cases correspond to absences of monitoring that needed to be carefully analysed. In most of the cases, this less frequent monitoring concerned aircraft not yet transferred but within a chain of already transferred aircraft. It is assumed that the controllers thought the aircraft had been sent. These cases do reflect consequences of no using strips or not updating them. cases correspond to risky absence of monitoring caused by focus on other traffic.... Interpreting situations In questionnaires, controllers felt their resulting situation awareness was similar in both conditions (distance and time) according to all of them, the use of spacing instructions enabled a better awareness of the established sequences and of forthcoming problems.... Situation awareness assessment As pointed out in many studies (Endsley, 99), situation awareness is difficult to assess. In the context of air traffic control, it leads us to question controllers knowledge of the traffic. Typically, we do not expect controllers to remember exactly details about every aircraft (position, level, speed, status). We assume the necessary information to be the accurate location of areas where traffic still has to be acted upon and a more roughly location of areas where traffic is already sequenced or not conflicting. The details about executive controller annotations on the blank maps have been summarised in the Table. It is difficult to draw conclusions from these data. First of all, some controllers had difficulty to fill the documents. Second, each case seems to be very specific, and no generic trend can be identified. However, in most of the cases, the same level of detail is provided in the three conditions. We notice an effect of the controller rather than of the condition. However, with respect to the impact of situation awareness on safety, two issues will have to be tackled. First, whereas controllers seem to have a general awareness of traffic location with spacing instructions, is this awareness sufficient to perform efficiently their task? The analysis of quality of control might provide some answers to this question. In addition, the introduction of abnormal events (or diverting situations) might help us assessing controller ability to detect problems. Second, does airborne spacing induces changes in the information collected? Typically, with airborne spacing, we expected controller objects of attention to evolve from aircraft flight parameters (flight level, speed) to spacing information (distance between aircraft). Even though it could not be measured, observations suggest that the slow down effect induced by spacing instructions was sometimes not detected soon enough by controllers. This will need further investigation. Project AGC-Z-FR - EEC Report No. Volume I

72 EUROCONTROL COSPACE Table Evaluation of controller description of traffic, as provided on the sector maps A A A A A A AO No Distance Time No Distance Time No data Plots nd part of sector Plots entry Plots exit Plots entry Plots exit Plots entry Plots middle Plots all over sector Plots entry Plots exit Plots nd part of sector No data Plots entry Plots exit Plots entry Plots middle Plots all over sector Plots entry Plots exit Plots nd part of sector Plots entry Plots exit Plots entry Plots exit Plots all over sector Plots all over sector No data Plots all over sector Plots entry Plots exit Plots all over sector Plots all over sector Plots entry Plots middle AR Plots entry Plots exit Plots entry Plots exit No data Plots all over sector Plots all over sector Plots entry Plots middle Plots all over sector Plots all over sector Plots entry Plots exit Plots all over sector Plots all over sector Plots entry Plots middle Synthesis on traffic situation monitoring and analysis Monitoring over the building area was similar in all conditions. With airborne spacing, less monitoring took place in the second part of the sector, corresponding to the maintaining phase. With airborne spacing, parafoveal fixations were twice more frequent than without it. Hypothesis Mon, Mon are confirmed. Mon and Mon are not confirmed. Mon and Mon were not tested.... Assume and transfer aircraft Because E-TMA controllers received traffic from a feed sector, we did not analyse the task of assuming aircraft. However, because they were instructed to respect the transfer conditions, we considered the location of transfer as a possible indicator of safety (in terms of possible omission to transfer aircraft). Mean values (Figure ) show that in all conditions, and in both sectors, most transfers occur between and miles from the IAF. The distribution of transfer location show that many aircraft were transferred very late, even after aircraft had passed the IAF. When analysing results per controller we could not identify generic trends. Neither the sector nor the conditions could explain late transfer. Depending on controller, all variables seem to have an influence some transfer similarly in distance and without, but later in time, other later in distance, other without. Not using paper strips could be a cause of transfer omission without strips (or if strips are not updated), controllers might no longer be aware of the transfer status of aircraft. Rather than providing an absolute trend, we will discuss in the safety section (omission to transfer) possible explanations for these late transfers. Project AGC-Z-FR - EEC Report No. Volume I

73 COSPACE EUROCONTROL AR AO Total number of aircraft IAF > Nm [;[ [;[ [;[ [; [ [ ;] [;] < Nm before IAF after IAF Spacing No Distance Time Total number of aircraft 9 7 IAF > Nm [;[ before IAF [;[ [;[ [; [ [ ;] [;] < Nm after IAF Spacing No Distance Time Figure Location of transfer to next frequency. Mean values for AR (left) and AO (right). Synthesis on transfer activity During the measured runs, controllers did not respect the implicit experimental constraint related to the transfer conditions. Therefore, no correlation can be made between occurrence of transfer and independent variables. Hypothesis T could not be tested.... Synthesis on controller activity The main impact of airborne spacing is a removal of speed instructions, and reflects the transfer of the maintaining task to the flight deck. Once the sequences built, there is no need for the controller to issue further instructions, as the spacing is maintained by the flight deck. Reduction of manoeuvring instructions is observed in both sectors and both conditions. Airborne spacing enables anticipation in sequence building. No smoothing of activity was observed when looking at temporal distribution of manoeuvring instructions. The use of airborne spacing does not modify the monitoring over the building area, but alleviates the monitoring over the area where spacing has to be maintained. With airborne spacing, twice more peripheral monitoring corresponding to parafoveal fixations is observed... EFFECTIVENESS... Quality of flow management In E-TMA, the controller has to sequence aircraft in order to respect a sequencing constraint at the exit point (e.g. NM, not catching up, at FL9). The quality of control could thus be addressed through three elements number of aircraft overflying IAF, spacing conditions at transfer to TMA and spacing conditions at IAF. Spacing conditions cover inter-aircraft spacing and closure rate. The Table 9 shows, for each sector and each condition, the number of aircraft overflying IAF during a minute period. Results show no impact of the conditions. However, the experimental conditions (fixed number of aircraft entering the sector) might not allow to analyse the impact if any. Project AGC-Z-FR - EEC Report No. Volume I

74 EUROCONTROL COSPACE Table 9 Number of aircraft overflying IAF (period of minutes) AO AR No Distance Time No Distance Time A A A A A Total Mean As seen previously (section..), some transfers were missing or happened too late (probably due to TMA not manned). Consequently, spacing conditions at transfer could not be considered for comparison in terms of quality of flow management (some cases are considered in terms of safety, see...). We only considered the spacing conditions at IAF, first through distribution of spacing value at IAF. In both sectors (Figure ), with distance based spacing, a peak is observed around the expected spacing value (i.e. NM), whereas spacing values are spread between and Nm in conventional control situations. Numberof Aircraft 9 7 Spacing at IAF ADN - Distance / No - INIO (7 - ) Spacing (nm) Distance () No () Number ofaircraft Spacing at IAF ADN - Distance / No - INIR ( ) 9 7 Spacing (nm) Distance () No () Figure Examples of distribution of spacing in distance. Without and with distance. AO (left) and AR (right). Project AGC-Z-FR - EEC Report No. Volume I

75 COSPACE EUROCONTROL Number of aircraft 9 7 Spacing at IAF Session A Time AO sector ( ) Spacing (s) mean AT AT Number of aircraft Spacing at IAF Session A Time AR sector ( ) 9 7 Spacing (s) mean AT AT Figure Examples of distribution of spacing in time. AO (left) and AR (right). To get a more synthetic view of the spacing distribution, we identified four categories unacceptable spacing value (below NM or s), small spacing value ([; 7,NM[ or [; s[), optimal spacing value ([7,;,NM] or [; 9s[) and large spacing value (above.nm or 9s). Results are shownin Figure. Effect of airborne spacing (without versus with) Spacing instructions enable a more homogeneous flow at IAF, with more aircraft getting the optimal spacing value. With spacing instructions, % of the aircraft are optimally spaced at IAF against % without spacing instructions. In addition, spacing instructions enable fewer aircraft overflying IAF with too small spacing. Last of all, it reduces the number of aircraft overflying IAF with too large spacing. Because the same traffic was used in the three conditions, spacing instructions seem to enable a better use of the airspace, in reducing the number of aircraft overflying IAF with more than.nm spacing. Effect of airborne spacing type (distance versus time) Distance and time based spacing seem to have similar effect. Effect of sector (AO versus AR) In AO, % of aircraft gets the optimal spacing at IAF, against only % without spacing instructions. In AR, % of aircraft are optimally spaced at IAF against % without spacing instructions. The impact of the spacing type (distance/time) is different in the two sectors. Whereas results seem similar in AO, they are different in AR time based spacing provides more benefits than distance based. Typically, compared to distance based, time based spacing enables much fewer aircraft overflying IAF with small spacing ( against ), and more aircraft with the optimal spacing value (7 against 9). Note that it is not possible to decide from plain figures if large spacing values reflect non optimal spacing or traffic characteristics. Unacceptable spacing values were analysed from a safety perspective (see...). Project AGC-Z-FR - EEC Report No. Volume I 7

76 EUROCONTROL COSPACE 9 7 Spacing at IAF (AO) 7 <nm [;7.nm[ [7.;.nm[ <s [-s[ [-9s[ Not acceptable Small Optimal 7 >.nm >9s Large <nm <s Not acceptable Spacing at IAF (AR) [;7.nm[ [-s[ Small 9 7 [7.;.nm[ [-9s[ Optimal 9 >.nm >9s Large No Distance Time No Distance Time <nm <s Not acceptable Spacing at IAF (both sectors) 9 [;7.nm[ [-s[ Small [7.;.nm[ [-9s[ Optimal No Distance Time >.nm >9s Large Figure Repartition of spacing quality, based on spacing value at IAF. AO (top left), AR (top right) and both sectors (bottom). To go a step further, we analysed closure rates for aircraft having the optimal spacing at IAF. Three categories have been defined aircraft slower, stable or faster than preceding aircraft (Figure ). Apart from having more cases with optimal spacing with distance, it could be noticed that few cases are catching up. With airborne spacing, most of the aircraft fly over IAF in a stable situation. Project AGC-Z-FR - EEC Report No. Volume I

77 COSPACE EUROCONTROL Closure rate (AR) Closure rate (AO) Slower <- kt Stable [-;+ kt] Faster >+ kt Slower Stable <- kt [-;+ kt] Faster >+ kt No Distance No Distance Closure rate (both sectors) Slower <- kt Stable [-;+ kt] Faster >+ kt No Distance Figure Closure rates of aircraft with optimal spacing value. AR (top left), AO (top right) and both sectors (bottom). Synthesis on flow management The number of aircraft overflying IAF does not seem to change. With distance and time based spacing, spacing at IAF between aircraft is more regular and optimal. Instead, without airborne spacing, there are more cases of unacceptable spacing. Hypothesis TF, FM and Seq. are confirmed.... Quality of flight service provided... Instructions per aircraft Despite a limited realism of the flight crew perspective, it is nevertheless possible to have an initial insight on the impact of airborne spacing on it. More precisely, it is essential to ensure that potential benefits for the controller (e.g. overall reduction of instructions) are not detrimental to some aircraft. For that purpose, we compared the distribution of instructions per aircraft in the three conditions (Figure 7 and Figure ). Note that the total of all instructions also includes spacing instructions. Results, which are in line with previous years, show that with spacing instructions more aircraft get fewer instructions. Project AGC-Z-FR - EEC Report No. Volume I 9

78 EUROCONTROL COSPACE Number of instruction per aircraft Number of instruction per aircraft Number of instruction per aircraft > 9 7 AO (All sessions) All instructions 7 9 Number of aircraft > AO (All sessions) Heading instructions 9 7 > Number of aircraft AO (All sessions) Speed instructions Number of instruction per aircraft Number of instruction per aircraft r aircraft Number of instruction pe > 9 7 > 9 7 > 9 7 AR (All sessions) All instructions 7 9 Number of aircraft AR (All sessions) Heading instructions 7 9 Number of aircraft AR (All sessions) Speed instructions Number of aircraft Number of aircraft Distance Time No Not instructed Spacing instructed Figure 7 Total number of instructions per aircraft. All sessions. Project AGC-Z-FR - EEC Report No. Volume I

79 COSPACE EUROCONTROL > > A - AO - All instructions A - AR - All instructions Number of aircraft Number of aircraft > > A - AO - Heading instructions A - AR - Heading instructions Number of instructions Number of instructions Number of instructions Number of aircraft Number of aircraft > > A - AO - Speed instructions A - AR - Speed instructions Number of aircraft Number of aircraft Not instructed Spacing instructed Distance Time No Figure Distribution of instructions per aircraft. Session A.... Trajectories To assess the impact of the spacing instructions on the trajectories optimisation, we considered the trajectories flown in the three conditions. Results show that with both distance and time based spacing, trajectories are more direct and most of the late alterations disappear (Figure 9 top and Figure 9 middle). However, since time based spacing requires more physical space in the building phase (e.g. 9s corresponding to NM), there are larger deviations in the building area (Figure 9 bottom). Project AGC-Z-FR - EEC Report No. Volume I

80 EUROCONTROL COSPACE ALURA AMORO AOSTA ARPUS ARSIL ATN AVLON BAGOL BAMES BANKO BARAK BASUD BEGAR BEGEL BENIP BERAP BIBOT BLM BUBLI BULOL BUSIL ALURA AMORO AOSTA ARPUS ARSIL ATN AVLON BAGOL BAMES BANKO BARAK BASUD BEGAR BEGEL BENIP BERAP BIBOT BLM BUBLI BULOL BUSIL CACHI CDN CERVI CHABY CHW CIV CLM CMB CMF CTL DANBO DELOX DERAK DIDOR DJL DPE EPL ETAMP ETREK FIJAC GALBI GELTA GEMRA GERBI GIPNO GTQ GUERE GVA HOC HR IXILU KATIL KELUK KOPOR KOTUN LASAT LASON LAULY LIRKO LISMO LOGNI LSA LUL LUREN LUSAR LUVAL MADOT MANAG MEL MELEE MELKO MENOX MILPA MOLUS MOROK MOTAL MOU NEBUL NEV NITAR NURMO OBORN OKRIX OPALE OSKIN CACHI CDN CERVI CHABY CHW CIV CLM CMB CMF CTL DANBO DELOX DERAK DIDOR DJL DPE EPL ETAMP ETREK FIJAC GALBI GELTA GEMRA GERBI GIPNO GTQ GUERE GVA HOC HR IXILU KATIL KELUK KOPOR KOTUN LASAT LASON LAULY LIRKO LISMO LOGNI LSA LUL LUREN LUSAR LUVAL MADOT MANAG MEL MELEE MELKO MENOX MILPA MOLUS MOROK MOTAL MOU NEBUL NEV NITAR NURMO OBORN OKRIX OPALE OSKIN PAS PENDU PILON POGOL PON PUNSA RBT REKLA REM RESPO RIGNI RLP ROA ROLAV ROMTA ROTSI SAUNI SONAT SOSAL SOTOR SPR STR TALUN TARIM TDP TELBO TINIL TIRSO TOLP A TORP A TRO TUNOR TUROM UTELA VADOM VAMDA VANAS VATRI VDP VERDI VERIX VERMA VEULE PAS PENDU PILON POGOL PON PUNSA RBT REKLA REM RESPO RIGNI RLP ROA ROLAV ROMTA ROTSI SAUNI SONAT SOSAL SOTOR SPR STR TALUN TARIM TDP TELBO TINIL TIRSO TOLP A TORP A TRO TUNOR TUROM UTELA VADOM VAMDA VANAS VATRI VDP VERDI VERIX VERMA VEULE No ALURA AMORO AOSTA ARPUS ARSIL ATN AVLON BAGOL BAMES BANKO BARAK BASUD BEGAR BEGEL BENIP BERAP BIBOT BLM BUBLI BULOL BUSIL ALURA AMORO AOSTA ARPUS ARSIL ATN AVLON BAGOL BAMES BANKO BARAK BASUD BEGAR BEGEL BENIP BERAP BIBOT BLM BUBLI BULOL BUSIL CACHI CDN CERVI CHABY CHW CIV CLM CMB CMF CTL DANBO DELOX DERAK DIDOR DJL DPE EPL ETAMP ETREK FIJAC GALBI GELTA GEMRA GERBI GIPNO GTQ GUERE GVA HOC HR IXILU KATIL KELUK KOPOR KOTUN LASAT LASON LAULY LIRKO LISMO LOGNI LSA LUL LUREN LUSAR LUVAL MADOT MANAG MEL MELEE MELKO MENOX MILPA MOLUS MOROK MOTAL MOU NEBUL NEV NITAR NURMO OBORN OKRIX OPALE OSKIN CACHI CDN CERVI CHABY CHW CIV CLM CMB CMF CTL DANBO DELOX DERAK DIDOR DJL DPE EPL ETAMP ETREK FIJAC GALBI GELTA GEMRA GERBI GIPNO GTQ GUERE GVA HOC HR IXILU KATIL KELUK KOPOR KOTUN LASAT LASON LAULY LIRKO LISMO LOGNI LSA LUL LUREN LUSAR LUVAL MADOT MANAG MEL MELEE MELKO MENOX MILPA MOLUS MOROK MOTAL MOU NEBUL NEV NITAR NURMO OBORN OKRIX OPALE OSKIN PAS PENDU PILON POGOL PON PUNSA RBT REKLA REM RESPO RIGNI RLP ROA ROLAV ROMTA ROTSI SAUNI SONAT SOSAL SOTOR SPR STR TALUN TARIM TDP TELBO TINIL TIRSO TOLP A TORP A TRO TUNOR TUROM UTELA VADOM VAMDA VANAS VATRI VDP VERDI VERIX VERMA VEULE PAS PENDU PILON POGOL PON PUNSA RBT REKLA REM RESPO RIGNI RLP ROA ROLAV ROMTA ROTSI SAUNI SONAT SOSAL SOTOR SPR STR TALUN TARIM TDP TELBO TINIL TIRSO TOLP A TORP A TRO TUNOR TUROM UTELA VADOM VAMDA VANAS VATRI VDP VERDI VERIX VERMA VEULE Distance ARSIL BAMES BANKO BARAK BASUD BEGAR BEGEL BENIP BERAP BUBLI ARSIL BAMES BANKO BARAK BASUD BEGAR BEGEL BENIP BERAP BUBLI CDN CHW CIV CLM CMB CMF CTL DIDOR DJL DPE EPL ETAMP ETREK FIJAC GALBI GELTA GEMRA GTQ KATIL KELUK KOPOR KOTUN LASAT LASON LAULY LUREN LUSAR LUVAL MADOT MANAG MEL MELEE MELKO NEBUL NEV NITAR NURMO OBORN OKRIX OPALE CDN CHW CIV CLM CMB CMF CTL DIDOR DJL DPE EPL ETAMP ETREK FIJAC GALBI GELTA GEMRA GTQ KATIL KELUK KOPOR KOTUN LASAT LASON LAULY LUREN LUSAR LUVAL MADOT MANAG MEL MELEE MELKO NEBUL NEV NITAR NURMO OBORN OKRIX OPALE PILON POGOL PON PUNSA RBT REKLA REM RESPO RIGNI RLP ROTSI SAUNI SONAT SOSAL SOTOR SPR STR TALUN TARIM TDP TELBO TOLP A TORP A TRO TUNOR TUROM UTELA VADOM VAMDA VANAS VATRI VDP VERDI VERIX VERMA VEULE PILON POGOL PON PUNSA RBT REKLA REM RESPO RIGNI RLP ROTSI SAUNI SONAT SOSAL SOTOR SPR STR TALUN TARIM TDP TELBO TOLP A TORP A TRO TUNOR TUROM UTELA VADOM VAMDA VANAS VATRI VDP VERDI VERIX VERMA VEULE Time GA OL GA OL TH THG THO TH THG THO LFPB LFPG LFPN LFPO GA OL GA OL TH THG THO TH THG THO LFPB LFPG LFPN LFPO GA OL GA OL TH THG THO TH THG THO LFPB LFPG LFPN LFPO ALURA AMORO AOSTA ARPUS ATN AVLON BAGOL BIBOT BLM BULOL BUSIL ALURA AMORO AOSTA ARPUS ATN AVLON BAGOL BIBOT BLM BULOL BUSIL CACHI CERVI CHABY DANBO DELOX DERAK GERBI GIPNO GUERE GVA HOC HR IXILU LIRKO LISMO LOGNI LSA LUL MENOX MILPA MOLUS MOROK MOTAL MOU OSKIN CACHI CERVI CHABY DANBO DELOX DERAK GERBI GIPNO GUERE GVA HOC HR IXILU LIRKO LISMO LOGNI LSA LUL MENOX MILPA MOLUS MOROK MOTAL MOU OSKIN PAS PENDU TINIL TIRSO ROA ROLAV ROMTA PAS PENDU TINIL TIRSO ROA ROLAV ROMTA Figure 9 Example of aircraft trajectories. Without spacing (top), with distance (middle) and with time based spacing (bottom). Project AGC-Z-FR - EEC Report No. Volume I

81 COSPACE EUROCONTROL... Speed profile The aircraft speed profile was also considered as an indicator of the impact of airborne spacing on flight efficiency. In distance based, due to the slow down effect, a negative impact was expected. Indeed, because of definition of the distance based spacing, instructed aircraft should have the same speed as its target (when spacing is acquired). With long sequences of aircraft, this means that once the first aircraft (at lower altitude) starts reducing, the rest of the chain should also reduce. In conventional control, although controllers are spacing aircraft in distance, there is no slow down effect as they are using some buffer to compensate for difference of speeds and altitudes. In time based, this effect is expected to disappear as the instructed aircraft has to replicate its target position and speed with a delay. To assess the impact on speed profiles, we compared the distribution of mean speed values in the second part of the sector (from Nm from the IAF until the IAF itself), passed the top of descent. Mean speed profile, for both AO and AR sector do not show difference between the three conditions (Figure ). Speed profile (All sessions, AO) Speed in kt Distance to IAF (nm) No Distance Time Speed profile (All sessions, AR) Speed in kt Distance to IAF (nm) No Distance Time Figure Mean speed profile. AO (top) and AR (bottom). Project AGC-Z-FR - EEC Report No. Volume I

82 EUROCONTROL COSPACE However, when looking more specifically at some runs, we notice example of the impact of distance based spacing on the speed profile. It shall be notice that the run shown here (Figure ) is considered as a successful run under distance based 7% of the aircraft received spacing instructions, overall instructions were reduced by %, speed instructions were reduced by %, geographical distribution showed a positive impact and 7% of the aircraft had an optimal spacing at IAF (]7.;.[). When looking at the time based spacing, we also notice an effect initial speeds are lower with time based spacing. This could be related to the fact that time based spacing requires more physical space during the sequence building phase, which might be obtained by speed reductions. Speed profile (AO) n kt Speed i Distance to IAF (nm) No Distance Time Figure Example of the impact of distance based spacing on speed profile... Descent profile The aircraft descent profile was also considered as an indicator of the impact of airborne spacing on flight efficiency. We expected distance based spacing to induce earlier descent as a result of the slow down effect, i.e. aircraft having to follow same speed (thus similar altitude) as its target. However, no impact can be observed mean profile and standard deviation are similar (Figure ). Project AGC-Z-FR - EEC Report No. Volume I

83 COSPACE EUROCONTROL Descent profile (All sessions, AO) D escent in feet (X) Distance to IAF (nm) No Distance Time Descent profile (All sessions, AR) D escent in feet (X) Distance to IAF (nm) No Distance Time Figure Mean aircraft descent profiles. AO (top) and AR (bottom). It shall be noticed that even though the top of descent was supposed to take place in the same area, this was not always respected, in all spacing conditions, and descent were sometimes given very early in the sector. Therefore, beyond looking at mean profile, when looking at atypical profiles, it is not possible to distinguish before very early descents and possible impact of distance spacing. Typically, the example presented on Figure could suggest an impact of distance spacing. However, the analysis of the runs leads us to attribute the modified profile to an effect of controller individual strategy rather than distance spacing. Project AGC-Z-FR - EEC Report No. Volume I

84 EUROCONTROL COSPACE Descent profile (AO) Descent in feet (X) Distance to IAF (nm) No Distance Time Figure Typical example of different descent profile in one condition... Flight efficiency Initial estimation of the efficiency variations was made through the record of time, distance and fuel consumption (Table and Figure ). Differences between conditions are not significant. Therefore, we can not show negative nor positive impact of the use of spacing instructions on the flight efficiency. Table Mean values of fuel consumption, flight duration and distance flown No Distance Time Fuel consumed (kg) 7 7 Time flown (hours) Distance flown (Nm) Fuel consumed (kg) Time flown (h) Distance flown (Nm) Fuel consumed Time flown Distance flown Without No Distance Time Figure Distance flown, fuel consumed and time flown Synthesis on flight service provision With airborne spacing, more aircraft get fewer manoeuvring instructions the global reduction of instructions is shared among all aircraft and not detrimental to some of them. With distance and time spacing, trajectories are more direct and most of the late alterations disappear. However, with time, larger deviations in the building area are observed. No difference is observed between speed and descent profile in the three conditions. No significant impact is found on flight efficiency. No impact of spacing condition (distance/time) on the quality of flight service provision was observed. Hypothesis SP is confirmed. Seq, SP and SP could not be confirmed. Project AGC-Z-FR - EEC Report No. Volume I

85 COSPACE EUROCONTROL... Synthesis on effectiveness If the number of aircraft overflying IAF does not seem to change, spacing at IAF between aircraft is more regular and optimal with airborne spacing. With airborne spacing, more aircraft get fewer manoeuvring instructions. Larger deviations in aircraft trajectories are observed early in the sector with time condition. No difference is observed between speed and descent profile. No significant impact is found on flight efficiency. No impact of spacing condition (distance/time) on the quality of flight service provision was shown... SAFETY... Method The objective of the previous section was to qualify the impact of airborne spacing on the quality of control. We considered the number of failures (separation and spacing infringements) as indicators of quality of control, as well as outcome of airborne spacing use. The objective of this section is to understand what caused inefficient or unsafe control. We consider as indicators of safety (or actually unsafety) any event, action or effect different from what was expected. Depending on the context and on possible consequences, the safety criticality of these indicators might vary. Typically, we consider error as an initial event (e.g. wrong instruction given) which might lead, under certain circumstances to a major failure (loss of separation) that is visible at the system level. Our method of analysis followed three steps we first defined indicators (Table ), then indexed and counted occurrences and finally used replay tools to understand the context of these occurrences. In addition to identify failures (that is the observable events), the method followed aimed at understanding the succession of events that caused the failures. Table Indicators of unsafety considered Control Airborne spacing Errors Loss of separation Loss of spacing Unstable situation Omission to descend aircraft Omission to transfer Non respect of applicability conditions Misuse of spacing instructions Indicator Distance between aircraft compared to minima (NM) Distance between aircraft compared to required spacing Closure rate at transfer Relative flight level Distance from exit point when transferred Closure rate when delegated Relative trajectories when delegated Relative flight level when delegated Speed variation after spacing instructions Incompatible instructions (e.g. spacing and speed) Superfluous instructions (e.g. merge plus direct) Project AGC-Z-FR - EEC Report No. Volume I 7

86 EUROCONTROL COSPACE... Control errors... Loss of separation The first objective of controller work is to ensure the safety of traffic, essentially in maintaining a minimum separation between aircraft. Therefore, following traditional analysis in air traffic control experiments, we looked at the number of losses of separation between aircraft at same flight level. Minor loss includes separation value between. and NM, serious loss between. and.nm and very serious loss below.nm. However, the duration of losses can increase the criticality of the loss (e.g..nm, but during 7 seconds will be serious, whereas.nm during seconds will be minor). To summarise figures presented intable, we obtain losses without the spacing instructions ( minor, serious and very serious), losses with distance based spacing ( serious and very serious) and with time based basing ( minor and serious). These figures do not show significant impact of spacing on safety, even though more minor losses are observed in conventional condition. It shall be stressed that surprisingly more losses occur in the AR sector, which was considered as the easiest sector. Most of the losses of separation were consequences of inappropriate initial conditions when using spacing instructions instruction was given to aircraft either too fast or too high compared to its target. This resulted in the aircraft not reducing speed and consequently infringing separation. In addition, the separation losses occurred in half of the cases in the area where aircraft should have been transferred to the TMA. It reflects the fact that with spacing instructions situation initially looks acceptable and takes time to degrade. This stresses the need for continuous assessment of the applicability conditions. Table Synthesis of losses of separation per sessions, as a function of the airborne spacing and per sector Very serious Serious Minor Session Condition AO AR AO AR AO AR A A A All No Distance Time No Distance Time No Distance Time No 7 Distance Time All All 9 Project AGC-Z-FR - EEC Report No. Volume I

87 COSPACE EUROCONTROL... Omissions of instructions We analysed if airborne spacing led controllers to omit to descend and/or transfer aircraft. Typically, we measured the relative altitude between successive aircraft and the geographical location of aircraft transfers. No impact of airborne spacing was noticed.... Repetition of instructions In section., redundant orders were analysed. Over the 7 redundant orders detected, needed to be investigated. Both controllers and pilots were at the origin of errors ( cases for controllers, by pilots). We expected short interval to reflect slips (e.g. accidental repetition, such as fast double-click on pseudo-pilot interface) and longer interval to reflect mistakes (e.g. intentional repetition of instruction, reflecting a loss of situation awareness). Over the 7 cases of orders repeated within less than seconds (Figure ), were from controllers and from pilots. No specific instruction was concerned by controller repetition, while pilots errors were more frequent with direct and level instructions. We assume that pilot errors are related to the simulation situation it could happen that their interface did not react instantaneously, leading to quickly repeat the action. However, it shall be noticed that these 7 cases represent only % of the total number of instructions given. Cospace ' ETMA repartition of doubles interval < s Direct Heading Level Speed Spacing Controller Pilot Figure Repartition of redundant orders, whose period is shorter than seconds Over the orders repeated within more than seconds (Figure ), were from controllers and from pilots. Pilots errors were more frequent with level and speed instructions, while controller redundant orders concerned direct and speed instructions. It is not possible here to identify if pilot errors are realistic and reflecting possible operational problem (e.g. implementing instruction given to another aircraft). Regarding controller redundant instructions, two causes are envisaged either controllers forget that they have previously given the instruction or they do not see the expected actions implemented. The first cause could occur when controllers are not using paper strips or are too busy to update them and lose track of the instructions given. The second cause could be observed when incompatible orders are given. For example, a Mach instruction could impair later descent in increasing the rate of descent. In this later case, redundant orders could reflect operational errors. In both cases, errors can be attributed to a loss of situation awareness. In the first case, the awareness of the current situation is lost, in the second case, the understanding of the consequences of previously given instructions is lost. Project AGC-Z-FR - EEC Report No. Volume I 9

88 EUROCONTROL COSPACE Cospace ' ETMA repartition of doubles interval > s Direct Heading Level Speed Spacing Controller Pilot Figure Repartition of redundant orders, whose period is longer than seconds... Unsafe transfer conditions In E-TMA, letters of agreement fix conditions of transfer typically, controllers are expected to transfer to TMA sectors flows of aircraft sequenced, with a given spacing (e.g. NM) and in stable condition (i.e. not catching up). Whereas the analysis of spacing value between transferred aircraft provides information in terms of quality of control, the stability of spacing is more related to safety issue. Not only catching up aircraft might infringe separation minima, but once transferred they will require actions from the receiving sector. To assess if aircraft are transferred in a stable situation (i.e. not catching up), we planned to analyse the closure rates between successive aircraft at transfer. For each run, each condition and in both sector, we considered the distribution of closure rates as a function of spacing value at transfer (Figure 7). For optimal spacing value ([7.;.[), significant closure rates are above knots. For spacing values below NM, we considered as relevant any closure rate, which would degrade an already unacceptable situation. We differentiated closing up aircraft leading to unsafe spacing (below NM) from closing up leading to small spacing (between and 7.NM). To complement this analysis, we compare the aircraft spacing value at transfer with the spacing value at IAF. Even though positive results in favour of spacing instructions were obtained, they can not be presented in this report. Indeed, as illustrated in Figure, transfers did not occur at a same given point. Depending on the run (i.e. not only the condition, but also the controller), conditions at transfer could correspond to conditions in E-TMA (NM before the IAF) or in TMA ( or NM after the IAF). Despite the training sessions conducted with active TMA sector, in the absence of TMA controllers during the measured runs, E-TMA controllers did not seem to respect transfer conditions or were omitting them. Session AN -AR- No Session AD -AR- Distance Session AT -AR - Time Number of aircraft Number of aircraft Number of aircraft Current spacing at transfer > Closure rate Current spacing at transfer > Closure rate Current spacing at transfer > Closure rate Figure 7 Distribution of closing up aircraft, AR sector ( run). Without (left), with distance (centre) and with time based spacing (right). 7 Project AGC-Z-FR - EEC Report No. Volume I

89 COSPACE EUROCONTROL As illustrated in Figure, the use of spacing instructions enabled aircraft to be regularly sequenced. This is an indirect indicator of safety at a global level, in the sense that it might reduce workload of TMA controllers, or at least allow it to remain acceptable indeed, low quality of transfer (aircraft too close, too far or unstable) will require TMA controllers to recover the situation in addition to their tasks of integrating flows of aircraft. In addition, analysing spacing value at IAF enabled us to identify cases of aircraft with less than NM longitudinal separation. An operational expert used replay tools to analyse each case, in order to distinguish unsafe from low quality transfer. Typically, unsafe cases were aircraft not separated longitudinally nor vertically, whereas low quality transfers involve aircraft separated at least vertically. First of all, we investigated the overall context, including the aircraft concerned (given aircraft and possibly previous one), aircraft respective speeds, current and predicted spacing, respective flight levels (Table ). The table provides details about the involved aircraft condition, respective call signs, sector, current spacing at transfer, transfer time, type of instruction, requested spacing, respective speeds, closure rate, respective flight levels. Based on these data, replay tool was used to perform a contextual analysis of situations, in order to understand what initial decisions led to such events. This latest analysis required the expertise of an air traffic controller, highly experienced both in sequencing activity and in spacing instructions. Detailed comment of the analysis are presented in Annex. Table Cases of unsafe spacing (extract) Session SpacingUse Callsign_Del Callsign_target tosect currspacing_attransfer TransferTime DelegationBeginOrder TAS_Del RequestedSpacing TAS_tar dv_delminustar AT T AFUL LBQL INIO AT T AF7YM AFR9 INIO. MERGE 9-7 AN No AFR9 EUL INIR AT T TAR7 AF77LA INIO. MERGE AN No LBGJ AFSS INIO AN No AFXJ TAR7 INIO AD D AFKY PRXO INIO.7 MERGE AD D AFR9 LBQL INIO. MERGE AN No AFJV PRXO INIO AN No AF7YM AFR9 INIO AN No CRL9 CRL9 INIO AT T AFBL AF7JV INIO AFL_del AFL_tar transfers with spacing <NM were found by the system (distance to IAF at transfer of an aircraft minus distance to IAF of the preceding aircraft). After analysis, revealed unsafe, 7 were safe but reflected a poor quality sequencing, were actually optimal and one acceptable (late transfers, after the IAF, for which the computation we use does not reflect the actual situation). % of the unsafe transfers occurred without the use of the spacing instructions, % with the time based spacing and % with distance based spacing. It shall be stressed that one controller was responsible for % of the unsafe transfers. Errors related to spacing instructions, identified when analysing these unsafe transfers are presented in the following section. Project AGC-Z-FR - EEC Report No. Volume I 7

90 EUROCONTROL COSPACE Table Repartition of unacceptable transfer cases, for both sectors Unsafe transfers Poor quality Acceptable Optimal No Time Distance Synthesis on control errors More losses of separation occurred in condition without airborne spacing. However, these were medium losses of separation. The most numerous unsafe transfers were found in conditions without airborne spacing and with time based spacing. Repetition of instructions, indicating loss of situation awareness was observed. Hypothesis CE is confirmed, CE was not tested.... Airborne spacing errors... Spacing instructions usage Ensuring the feasibility of airborne spacing is part of the controller' tasks. In addition to initially assess applicability conditions, controllers are in charge of maintaining them. The main items defined in the applicability conditions are compatible speeds (e.g. ensure a slow aircraft is not asked to catch up on a much faster one), compatible flight levels and compatible trajectories (e.g. ensure an aircraft is not asked to merge behind an aircraft following a diverging path). One of the difficulties induced by airborne spacing is the mutual dependencies between delegated aircraft, and consequently the cognitive cost of maintaining appropriate situation awareness. For example, whereas it is quite easy for a controller to understand that an aircraft is reducing speed because its target is descending, it is harder to detect the consequence of speed variation of a target on the rate of descent of its delegated. In order to investigate systematically the conditions of airborne spacing, we defined what were the applicability conditions in most of the expected situations e.g. stable situation, descending target. Then, basic indicators, such as relative trajectories, relative speed and relative altitude were associated (and if possible combined) to each case. The third step consisted in a contextual analysis, focusing on the applicability conditions for each spacing instruction, from its start until its end. An initial typology of airborne spacing-related errors described their potential contexts of occurrence, causes (e.g. cognitive tunnel vision, slips, incomplete/incorrect knowledge) and possible means for their avoidance or tolerance. In order to automatically detect some of them, we defined indicators (Table ). Once these events were detected and documented (exercise and aircraft concerned), an operational expert analysed the conditions in order to understand possible causes (lack of training, misuse of spacing instructions, simulation pilot error). In addition, the analysis of unsafe transfer (see above) provided additional information about possible causes of errors related to the use of the spacing instructions. 7 Project AGC-Z-FR - EEC Report No. Volume I

91 COSPACE EUROCONTROL Applicability conditions could be either not respected initially or not maintained in time. Non respect of initial applicability conditions was indicated by Initial distance between aircraft below requested one, which led to drastic speed reduction (up to knots); Incompatible speeds (e.g. aircraft under airborne spacing much faster than the target when giving a heading then merge ), which could lead to too large deviations; Not same trajectory in case of remain behind (calculating angles between target and delegated); Not merging trajectories in case of merge (assessing if both aircraft are direct to the same waypoint). It shall be stressed that the two first indicators are not exclusive ones. For those, analysis needs to consider the combination of spacing value, distance from merging point and aircraft respective speed. Indeed, such cases could be acceptable if the speed difference between the two aircraft still enables the requested spacing to be achieved when the target is passing the merging point. To identify cases of erroneous initial conditions, we could have looked for pairs of aircraft with too small or too large spacing value, with not enough time to establish it and with incompatible speeds. However automatic search of such cases would still require operational expert to review each case to determine its acceptability. To reduce the analysis time, a better indicator was identified. Excessive speed variations, seconds after the instruction, is assumed to reflect inappropriate applicability conditions. Note that this indicator is discussed more extensively in the next section (...). Non maintaining of applicability conditions was indicated by Loss of spacing; Manoeuvre not compatible with the airborne spacing during merging phase (e.g. heading given to the target before the merging point or to the aircraft under airborne spacing) or during remaining phase (e.g. heading given to the target but not to the aircraft under airborne spacing); Not compatible flight levels (detecting successive aircraft with large difference in altitude NM before IAF); Late descent of the aircraft under airborne spacing. Before discussing the results, it shall be reminded that spacing instructions have been given. Table shows that erroneous initial conditions were twice more frequent in time based than in distance based conditions. % of the spacing instructions were given while initial applicability conditions were not appropriate. When investigating more specifically the cases (Table ) we could notice that same pairs of aircraft were involved, in different conditions (e.g. AFSF/AFGP in AD, AT, AD and AT). This would suggest that despite practice, controllers were confronted to the same difficulties. In addition, different controllers handling a same traffic were also experiencing similar difficulties. Project AGC-Z-FR - EEC Report No. Volume I 7

92 EUROCONTROL COSPACE Table Examples of pairs of aircraft for which initial applications were incorrect (extract) Session Spacing type Sector Spacing instruction Instruction time Callsign instructed Callsign target AT Time AO MERGE AEY AFHG AD Distance AO MERGE AEY AFHG AD Distance AO MERGE 9 AFSF AFGP AT Time AO MERGE AFSF AFGP AD Distance AO MERGE AFSF AFGP AT Time AO MERGE 9 AFSF AFMN AD Distance AO MERGE AF9BL LBDM AT Time AO MERGE AF9BL LBDM AT Time AO MERGE AF9BL LB9FR AD Distance AO MERGE AFTM AFZJ AT Time AO MERGE 9 AFTM AFZJ AT Time AO MERGE AFTM AFZJ AD Distance AO MERGE AFTM AFZJ Table Number of occurrences of the various types of airborne spacing related errors At init During Misused spacing (excessive speed reduction s after) Misused merge (no direct) Misused remain (bad angle) Incompatible orders with merge Incompatible orders with remain AO AR All Total Distance Time Distance Time Distance Time In addition, we also identified cases of misuse of airborne spacing, typically indicated by heading or speed instructions given to the instructed aircraft. It shall be noticed that assessing the required configuration and respecting compatibility between orders seem to be better understood by controllers, as errors related to these issues are much less frequent, representing only % of the cases. 7 Project AGC-Z-FR - EEC Report No. Volume I

93 COSPACE EUROCONTROL Beyond risks related to the erroneous use of spacing instructions, we also identified risks induced by this instruction. Typically, we identified risks of Failure to detect the catching up of two sequences under airborne spacing, controllers might pay less attention to the aircraft respective speeds; Omission to descent aircraft; Failure to anticipate and notice the impact of inappropriate spacing instructions on the following traffic (slowing down which can lead to no longer feasible airborne spacing).... Illustration of a degrading situation (Time condition, AO sector) For this run, the overall rate of use went up to 7%, which is a very high figure, compared to our baseline (7%). It shall be reminded that the AO sector is structured around two converging points, one early and one late. The latest one is used by controllers to integrate some Eastern flows. These Eastern aircraft do reduce the possibility to use airborne spacing in the sense that they have to be anticipated and sequence building has to take them into account. As mentioned in section..., the rate of use does not reflect the rate of efficient use. Typically, in this run, the use of spacing instructions induced only a % reduction of the number of instructions, and 9% of the speed instructions. The geographical distribution of instructions was not optimal many heading and speed instructions were given in the latest part of the sector, even after the merging point (Figure ). The spacing value at IAF (Figure 9) was homogeneously around the requested 9 seconds, apart from two cases below seconds (equivalent to NM). Number of instructions Geographical distribution of manoeuvring instructions AT AO Sector - (7 - ) 7 9 Distance to IAF (nm) Speed Heading Spacing Figure Geographical distribution of instructions. AO sector. Project AGC-Z-FR - EEC Report No. Volume I 7

94 EUROCONTROL COSPACE Number of Aircraft 7 Spacing at IAF A - Time - AO sector (7 - ) Spacing (s) Figure 9 Spacing at IAF. AO sector. The analysis of the run with a replay tool highlighted two points () late manoeuvring instructions were used to manually recover from degraded situation and () one consequence of this degraded situation was a pair of aircraft (AF7JV and AEY7) involved in a loss of separation (below NM or s). Case lack of anticipation, wrong assessment of initial conditions and use of wrong merging point. Consequence Modify merging point for the whole chain and rebuild the whole sequence. Context (initial situation). The controller has built two successive, but separate sequences of aircraft, using the merge instruction (Figure ). The first sequence is in remain behind situation while the second one is in merge situation to OKRIX. AF7JV is arriving from East and has to be integrated between the two sequences. AF_JV is faster (knots) and higher (FL) than the other aircraft. Figure 7... Initial conditions. sequences, merging to OKRIX. AF7JV is direct to MOLEK. In ideal conditions, the controller should have anticipated the AF_JV, planned space between the two chains. If not anticipated, in the same resulting situation, the controller should have vectored the AF7JV to put it behind the AEY7 and then vectored the following aircraft (AFBL, AFSS and LIBGJ) to create spacing behind the AF_JV. Because they had all received a spacing instruction, the controller should have cancelled them all. Possibly to avoid giving these instructions the controller tried to put the AF_JV between the two chains, despite improper conditions (different speeds, different levels, not enough space, and different merging points). 7 Project AGC-Z-FR - EEC Report No. Volume I

95 COSPACE EUROCONTROL Unfortunately, the time saved now in not cancelling the instructions will be lost later, in a much demanding situation. Let us illustrate how the situation will progressively degrade. To integrate AF7JV between the two chains, the controller asks AFBL to merge to MOLEK (Figure ), whereas aircraft behind it are merging to OKRIX. Figure 7... AF7JV inserted between the two sequences. AFBL given merge to MOLEK. As a consequence, spacing instructions previously given have to be cancelled and the merging point (OKRIX) is modified to be MOLEK (Figure ). All aircraft receive a cancel delegation, retain target followed by a merge to MOLEK. These spacing instructions given late in the sector are the ones displayed on Figure. Figure 7... Second sequence needs to be modified all aircraft are given a new merging point. Project AGC-Z-FR - EEC Report No. Volume I 77

96 EUROCONTROL COSPACE This modification enables spacing instruction to be given correctly (same merging point), but does not improve the situation in terms of current distance between aircraft, nor speed differences. Figure 7... Lack of spacing and speed difference led to a drastic speed reduction for the AF7JV. The initial conditions (not enough space and different speeds) led to a drastic speed reduction for the AF7JV (Figure ). In 9 seconds, the aircraft ground speed dropped from to knots, which would not be acceptable in real operations. Because of this speed reduction, the aircraft cannot descent, and is still much higher than the other aircraft. Even if aircraft are vertically separated, the AF_JV is gradually getting closer to both the AEY7 and the AFBL. To create space, the controller turns the AFBL (Figure ). However, to do so, he has to cancel the two spacing instructions in which the AFBL is involved the one with AF7JV and the one with AFSS. Then the turn can be given. Figure 7... Sequence is broken to recover the degraded situation. AFBL is vectored. At this moment, the AEY7 and the AF7JV have only.9nm separation. Safety is not endangered as they still are vertically separated (7 feet). To create spacing, a heading must be given to the AF7JV. To do so, its spacing instruction needs to be cancelled (Figure ). Figure 7... AF7JV is also given a heading. 7 Project AGC-Z-FR - EEC Report No. Volume I

97 COSPACE EUROCONTROL From now on, the controller will manually recover the situation, so that most of the aircraft will be transferred with a correct spacing (as illustrated on Figure 9). It shall be noticed that the aircraft involved in the current situation will be transferred late, even after they geographically passed the IAF. Figure 7... AFBL has just received a speed instruction. AFSS and LBGJ are given speed instruction. The AF7JV is still descending, and following its heading. Following aircraft (AFBL, AFSS and LBGJ) are vectored and given speed reductions to create spacing. Two more points need to be added the LBGJ has not been descended, the AFSS speed reduction will hardly be feasible knots at FL7 is not realistic. Again, these conventional manoeuvring instructions, given close to the IAF are the ones displayed between and nm from the IAF on Figure. When trying to force the situation, the controller ended up with a degraded situation requiring more time-critical actions. Not only the late recovery required more actions, but it also required attention from the controller, possibly detrimental to the situation awareness related to the rest of the traffic. Figure Cancelling previous instructions and modifying the merging point. To link the two sequences together the controller has to cancel previous instructions and modify the merging point. The modification of merging point for the second sequence even had an impact on a third sequence, ten minutes later instructions given had to be modified and new merging point given. Not only the initial condition required the controller to focus attention on the situation, but it even induced modification in the controller strategy. What the controller could have done is cancel the CRL spacing instruction, maintain it direct to OKRIX, possibly give it a speed reduction. No change would have then been required on the following aircraft. The situation induced a serious workload increase. Beyond requiring numerous fixations over the integration area, it induced numerous instructions, such as cancel spacing instructions and issue heading and speed instructions. Even though no loss of separation occurred, the increased workload and the focus on some aircraft could have led to a conflict not detected (e.g. in the sector entry). The situation awareness was degraded. Consequently, we consider that safety was endangered. Project AGC-Z-FR - EEC Report No. Volume I 79

98 EUROCONTROL COSPACE This example shows two points Airborne spacing does not modify the laws of physics. With wrong initial conditions, an aircraft can not comply with the instruction. Airborne spacing requires anticipation. Once the sequence is built, any modification requires numerous instructions. This example illustrates the difficulty of the AO sector and its late converging flow.... Speed variations following spacing instructions For every aircraft receiving a spacing instruction, we analysed the speed variations a certain time after the execution of the instruction. We identified the seconds delay as the most appropriate one it includes reactions to erroneous initial conditions, and should exclude normal speed reductions. We distinguished three categories speed decrease (kt or more), speed increase (kt or more) and between. Overall results show peaks between [-; +] knots in both sectors and in both conditions Figure ). The excessive speed variations (Table and Figure ) correspond to consequences of erroneous initial applicability conditions. To identify the ratio of initial erroneous applicability conditions that possibly led to an excessive speed variation, we cumulated the list of initial spacing too small with the list of initial speed difference too high and extracted redundant pairs of aircraft. different pairs of aircraft were counted. The number of excessive speed variations corresponds to % of the initial errors. It shall be reminded that during the experiment simulation pilots were not in a position to detect all the cases of not feasible spacing instructions. Following these investigations, it is envisaged to use a similar algorithm to detect in real time cases of erroneous applicability conditions. The objective is to involve simulation pilots who will refuse not feasible spacing instructions. It is expected that successive unable spacing will lead controllers to pay better attention to the respect of initial conditions. Speed variations s after instruction Number of occurrences AO 9 AR 7 AO AR D T <- kts OK >+ kts Figure Speed variations s after spacing instruction ( runs) However, as illustrated (Figure 9).differences were noticed between runs. Distributions show cases of excessive speed reductions following a spacing instruction. Following the detection of abnormal cases, their investigation was required. Project AGC-Z-FR - EEC Report No. Volume I

99 COSPACE EUROCONTROL Speed variation seconds after spacing instruction AD - AD - Number of aircraft < > Speed variation after sec < > Speed variation after sec Figure 9 Distribution of speed variations. Example of numerous excessive speed reductions (left) and acceptable speed variations (right). Synthesis on airborne spacing errors The most frequent errors, in the category of non respect of applicability conditions are related to non compatible speeds or initial spacing too small. Combination of incompatible orders (e.g. merge and heading) represents only % of the errors. The use of a new metric, called speed variation seconds after spacing instruction enables the detection of inappropriate applicability conditions. Hypothesis AS was not tested. AS was not confirmed.... Synthesis on safety The safety assessment consisted in analysing in detail traffic management activity. The definition of unacceptable events enabled their systematic detection. Expert analysis using replay tools helped identifying the initial conditions and possibly causes leading to such unacceptable events. Compared to current situation, airborne spacing does not induce more losses of separation. Airborne spacing induced errors are mainly related to the set up and the maintaining of applicability conditions. It is envisaged to insist on such risks of errors during training. The possible contribution of flight crews as mitigation means could not be assessed in the experiment. Project AGC-Z-FR - EEC Report No. Volume I

100 EUROCONTROL COSPACE.. CASE STUDY The main underlying hypothesis regarding airborne spacing is that, if correctly used, it should provide benefits in terms of controller availability that could be converted in improved safety and effectiveness. On the opposite, it is anticipated that incorrect use could lead to negative consequences, and inducing situations more demanding than if airborne spacing was not used. To assess these two hypotheses, it was decided to analyse the impact on a selected set of metrics in two typical cases of correct and incorrect use. Ideally, metrics should be analysed on limited period of time, corresponding to periods where all spacing instructions were either correctly used or not. However, two difficulties have to be stressed. First some metrics need to be calculated over a long period of time (e.g. geographical distribution of instructions), which might include both correct and incorrect cases of use. Second the restriction of the analysis on some cases has to be questioned, as it would exclude concurrent tasks performed simultaneously on aircraft included and excluded from the analysis (e.g. monitoring curves). Therefore it was decided to perform the case study on two runs representative of correct and incorrect use. Bad use of airborne spacing Rate of use aircraft were managed by the controller during the measured run. out of the arrival aircraft were under airborne spacing during the run, which corresponds to a low (7%) rate of use compared to the 7% mean rate. To assess the quality of the airborne spacing usage, we use the metric denoted speed variation seconds after the instruction (Figure 7). It illustrates that in half of the cases, the applicability conditions were not correct the instructed aircraft had to slow down. Good use of airborne spacing Rate of use out of aircraft were given a spacing instruction (%). The metric denoted speed variation seconds after the instruction (Figure 7) shows that in the present run, most of the spacing instructions were issued in suited conditions, as only very few instructions ( out of 7) were followed by a large speed reduction. Number of aircraft Speed variation seconds after spacing instruction Time - AT - AR < > Speed variation after sec Number of aircraft Speed variation seconds after spacing instruction Time - AD - AR < > Speed variation after sec Figure 7 Speed variations seconds after the instruction. Incorrect use. AR sector. Figure 7 Speed variations seconds after the instruction. Correct use. AR sector. Project AGC-Z-FR - EEC Report No. Volume I

101 COSPACE EUROCONTROL Bad use of airborne spacing Impact on human shaping factors (limited to impact on workload) Whereas the use of time spacing seems to increase significantly the executive controller mental and temporal demands, it has no impact on the planning controller workload. NASA-TLX ratings (Figure 7) show that compared to conventional situation (no), the executive controller mental and temporal demand are higher in time condition. Mental and temporal demands are similar for the planning controller in all conditions. However, both mental and temporal demands of the planning controller are much lower than the executive. Good use of airborne spacing Impact on human shaping factors (limited to impact on workload) NASA-TLX ratings (Figure 7) show similar trends for both executive and planning controller. Compared to conventional situation (no), the mental demand is similar in distance condition, and the temporal demand is similar in time condition. Both mental and temporal demands of the planning controller are lower than the executive. NASA-TLX - Executive and planning controller Mental demand Temporal demand No Distance Time No Distance Time EXC PLC NASA-TLX - Executive and planning controller Mental demand Temporal demand No Distance Time No Distance Time EXC PLC Figure 7 NASA-TLX mental and temporal demand. Incorrect use. Figure 7 NASA-TLX mental and temporal demand. Correct use. Project AGC-Z-FR - EEC Report No. Volume I

102 EUROCONTROL COSPACE Bad use of airborne spacing ISA ratings (Figure 7) show results similar to NASA-TLX. The executive controller workload is perceived higher in time than in conventional condition. Planning controller workload is perceived slightly higher in time condition. In both conditions the executive controller workload is much higher than the planning controller workload. Executive and planning controller workload (ISA) % Good use of airborne spacing ISA ratings (Figure 7) seem consistent with NASA-TLX ratings. The executive controller workload is perceived lower in distance condition. Planning controller workload is perceived similar in distance condition. In both conditions the executive controller workload is slightly higher than the planning controller workload. Executive and planning controller workload (ISA) % AR/EXC AR/PLC AR/EXC AR/PLC No Time Very High High Normal No Answer Low Very Low AR/EXC AR/PLC AR/EXC AR/PLC No Distance Very High High Normal No Answer Low Very Low Figure 7 ISA ratings. Incorrect use. Number of manoeuvring instructions The use of time spacing reduces the total number of manoeuvring instructions by 7% (7 instructions instead of 9). Heading instructions are reduced by % and speed instructions by 7% (Figure 7). Number of instructions Manoeuvring instructions repartition Time Figure 7 Manoeuvring instructions repartition Incorrect use. 9 9 Spacing Heading Speed No Figure 7 ISA ratings. Correct use. Number of manoeuvring instructions The use of time spacing reduces the total number of manoeuvring instructions by % (7 instructions instead of ). Heading instructions are reduced by % and speed instructions by 7% (Figure 77). Number of instructions Manoeuvring instructions repartition 7 No Figure 77 Manoeuvring instructions repartition. Correct use. Spacing Heading Speed Distance Project AGC-Z-FR - EEC Report No. Volume I

103 COSPACE EUROCONTROL Bad use of airborne spacing Geographical distribution of manoeuvring instructions As illustrated in section..., the use of airborne spacing usually improves the geographical distribution of instructions heading and spacing instructions are given early in the sector, while very few speed instructions are given (Figure 7). In addition, from nm before the IAF no more instructions are usually given. Good use of airborne spacing Geographical distribution of manoeuvring instructions In the present case (Figure 79), a positive impact is noticed heading and spacing instructions are given early in the sector (between and nm from the IAF). In addition, we can notice that with distance spacing, nearly no speed instructions are given. In the present case (Figure 7), no positive impact is noticed heading, speed and spacing instructions are given all over the sector, including in the exit area. No sequence building area can be identified. The incorrect use of spacing instructions in the early part of the sector induces degradations in the later part, requiring the controller to recover, using both conventional and spacing instructions. This continuous issuing of instruction can explain the high perceived workload. Number of instructions Geographical distribution of manoeuvring instructions Distance to IAF (nm) Speed Heading Spacing Number of instructions Geographical distribution of manoeuvring instructions Distance to IAF (nm) Speed Heading Spacing Figure 7 Geographical distribution of manoeuvring instructions. Incorrect use. Figure 79 Geographical distribution of manoeuvring instructions. Correct use. Project AGC-Z-FR - EEC Report No. Volume I

104 EUROCONTROL COSPACE Bad use of airborne spacing Geographical distribution of fixations The use of time based spacing does not modify the distribution of fixations similarly to conventional conditions, fixations are all over the sector (Figure ). The curve reminds the distribution of instructions, and suggests that the controller monitors aircraft when instructed. In addition, this result suggests that the controller is not in a position to anticipate events, but rather in a monitoring the overall sector, including essential and secondary areas. Good use of airborne spacing Geographical distribution of fixations Distance spacing enables the controller to focus attention on the early part of the sector, where the sequence needs to be built. Even though they are less numerous, fixations still occur on the second part of the sector. This curve suggests that the controller is in a position to anticipate its sequence building. In addition, the curve reminds the distribution of instructions and suggests that the monitoring is focused where instructions are issued. % Geographical distribution of fixations % Geographical distribution of fixations % % % % % % % % % % % % No Time No Distance Figure Geographical distribution of fixations. Incorrect use. Spacing at IAF Most of the aircraft had the required spacing (9 seconds) when passing the IAF (Figure ). This shows that the controller did recover the degraded situation and occasionally delayed the transfer in order to transfer the aircraft in good conditions. Figure Geographical distribution of fixations. Correct use. Spacing at IAF Compared to conventional situation (blue on Figure with airborne spacing, most of the aircraft have the required spacing ( nm) when passing the IAF. Aircraft of Number 9 7 Spacing at IAF with Number of Aircraft Spacing at IAF with 9 7 without Spacing(s) Spacing (nm) Figure Spacing value at IAF. Incorrect use. Figure Spacing value at IAF. Correct use. Project AGC-Z-FR - EEC Report No. Volume I

105 COSPACE EUROCONTROL Closure rate Bad use of airborne spacing With airborne spacing, among the aircraft with optimal spacing at IAF (i.e. spacing between and 9 seconds) were not in catching up situation and was even smaller than its leading aircraft. of the aircraft with optimal spacing were actually in catching up situation. 9 7 Closure rate (aircraft w ith op tim al spacing at IA F) Closure rate Good use of airborne spacing With airborne spacing, among the aircraft with optimal spacing at IAF, none was sent in a catching up situation (Figure ). Closure rate (aircraft w ith optim al spacing at IAF) CR négatif CR_OK CR positif CR Nnégatif CR_OK D C R positif Figure Closure rate (aircraft with optimal spacing at IAF). Incorrect use. Speed profile Compared to conventional situation, with (time based) spacing (green curve on Figure ), aircraft reduce speed too early. They are too slow, too far from the IAF. Aircraft mean speed profile Figure Closure rate (aircraft with optimal spacing at IAF). Correct use. Speed profile Compared to conventional situation, with (distance based) spacing (fushia curve on Figure 7), aircraft speed profile is very good aircraft speed is still very high, even close to the IAF. Aircraft mean speed profile Speed in kts Speed in kts Distance to IAF (nm) No Distance Time Distance to IAF (nm) No Distance Time Figure Aircraft mean speed profile. Incorrect use. Figure 7 Aircraft mean speed profile. Correct use. Project AGC-Z-FR - EEC Report No. Volume I 7

106 EUROCONTROL COSPACE Descent profile Bad use of airborne spacing Compared to conventional situation, with (time based) spacing (green curve on Figure ), aircraft generally descent earlier. Good use of airborne spacing Descent profile Compared to conventional situation, with (time based) spacing (green curve on Figure 9), aircraft are descended later, which is better in terms of flight efficiency. Descent in hundred on feet Aircraft mean descent profile Descent in hundred on feet Aircraft mean descent profile Distance to IAF (nm) No Distance Time Distance to IAF (nm) No Distance Time Figure Aircraft descent profile. Incorrect use. Figure 9 Aircraft descent profile. Correct use. Synthesis on case study In order to assess the consequences of correct versus incorrect use of airborne spacing, two case studies were performed on two representative runs. It was observed that when incorrectly used, the positive impact of airborne spacing is limited. In some extreme situations as illustrated in..., it may even endanger safety. When correctly used, the positive impact is even more significant than the average one of the measured runs. The analysis over the whole runs (as opposed to limited part) could explain why results are different, but not opposite. For the run considered as incorrect, correct cases of use might have counterbalanced incorrect cases and conversely. Project AGC-Z-FR - EEC Report No. Volume I

107 COSPACE EUROCONTROL 7. FINDINGS TMA The TMA experiment resulted from initial exploratory studies conducted prior to the November experiment. Before presenting the results from this latest real time experiment, it seems relevant to trace the process followed to understand first the TMA specificity, then if and how the airborne spacing could be used in TMA. Then, we will present the results from the experiments and discussed how to go further. 7.. UNDERSTAND TMA SPECIFICITY Participant Alain Zinger (Paris Orly). Following initial discussions with an expert approach controller, it appeared that, although airborne spacing seemed appropriate for E-TMA, its use would be more problematic in TMA. In particular, the anticipation needed to set-up sequences with spacing instructions seems hardly compatible with today practices (mainly late vectors for integration onto final approach). This raised the question what are the main differences between E-TMA and TMA? To address this question, we restricted the analysis to the E-TMA sectors already simulated (AO and AR) and associated approach sectors (Paris Orly and Charles de Gaulle). It is acknowledged that although these sectors are thought to be representative, a study of other dense TMA in Europe has to be performed. Two main differences have been identified standard trajectories in E-TMA versus radar vectoring in TMA, and integration on a point in E-TMA versus integration on an axis in TMA. In addition, in TMA, the high pressure to ensure the optimum runway capacity, the lack of space and the larger turns (e.g. from downwind leg to final axis), result in highly time-critical instructions. This timecriticality restricts even more the number of aircraft manageable for integration onto final approach, and generates uncertainty preventing an early planning of the final sequence order. Consequently, whereas in E-TMA the work consisted of a building phase then a maintaining phase, this is the opposite in TMA a maintaining phase to keep spacing on each flow, followed by a last building phase for integration of flows onto final approach. This is typically the case in E-TMA with AR sector, composed of AR and AR sectors, and the TMA sectors, composed of INI and ITM positions (Figure 9). At first sight, there could be benefits for maintaining spacing before integration onto final approach, typically on long downwind leg with flows already under airborne spacing (initiated by E-TMA). However, the relevance (usability and usefulness) of the existing spacing instructions for integration onto final approach was less obvious. The feeling was that, at least, a specific instruction (merge on an axis) had to be developed. Maintaining Building IAF Entry points AR AR ITM INI FAF IAF IAF Maintaining Building E-TMA TMA Figure 9 Today E-TMA and TMA configurations Project AGC-Z-FR - EEC Report No. Volume I 9

108 EUROCONTROL COSPACE 7.. IDENTIFY RELEVANT AIRSPACE AND PROCEDURES Participants Ludovic Boursier, Noëlle Canto, Olivier Galichet, Ronald Granju and, during debriefing, Alain Zinger (Paris Orly). In order to get a feedback on the relevance of the concept in TMA, and identify if existing spacing instructions could be used or if new ones had to be developed (e.g. merge on an axis), an initial experiment took place in June. It consisted of a series of small-scale simulations with approach controllers. Four controllers from Paris Orly participated, each for half a day a first exercise on an E-TMA sector (AO) to understand the concept and get familiar with the procedures, then a second exercise on a TMA sector. A collective debriefing took place at the end of the session. The TMA sector consisted of a simplified Orly TMA in a West configuration and had to integrate flows coming from two IAF (MEL at FL7 and EPR at FL). One flow (EPR) came through a long downwind leg, the other one (MEL) entered directly through a base leg (Figure 9). The level of traffic envisaged allowed grouping INI and ITM positions (as today in such traffic conditions), and manning this position with one controller. Compared to today s operations (radar vectoring), the major change was the use of a unique standard trajectory for each flow from IAF to FAF (OYE). TSU INIO.7 ITM. OYE PO7 MEL EPR CDN OKRIX Figure 9 June TMA sector (simplified map) The traffic was built to allow the controllers to test successively three different configurations maintaining of a sequence, integration of one aircraft in a sequence, and integration of two sequences. It entered the TMA sector already under airborne spacing (initiated by the feed sector), still merging to their respective IAF. The traffic level was low ( arrivals per hour) with sequences of up to four aircraft, equally balanced between the two IAF. All the traffic was equipped to receive spacing instructions. There was no departure. During the experiment, the need for a dedicated merging point arose, allowing using the merge instruction to integrate aircraft from different flows, in the same way as in E-TMA. The merging point was created as the first common point between base and downwind legs (point PO7, Figure 9). The remain instruction was also used with aircraft of a same flow. During the collective debriefing, we tried to find out to which extent airborne spacing could be used in TMA. More precisely, we tried to address the potential use and benefits of airborne spacing, step by step, from IAF to FAF maintaining spacing before integration onto final approach, maintaining spacing for one flow until FAF, and eventually managing integration onto final approach. Controllers mentioned that there could be a benefit to receive aircraft under airborne spacing since situations will be and will remain stable, in particular with a long downwind leg. To reuse part of the E-TMA preparation, itself in West configuration. 9 Project AGC-Z-FR - EEC Report No. Volume I

109 COSPACE EUROCONTROL However, they stressed that the benefit would be greater with time based spacing due to the progressive reduction of distance (9s would be equivalent to a reduction from Nm at about kts, to Nm at about kts then Nm at kts). In addition, one flow could maintain airborne spacing until FAF, but controllers mentioned that using standard trajectory may be less optimal than radar vectoring (enabling fine adjustments). The integration onto final approach under airborne spacing remains a key issue. In particular, the controllers mentioned the need to create gaps for inserting aircraft from different flows. To allow for more flexibility, they also suggested the creation of multiple merging points (and standard trajectories associated) and, in particular, some on the runway axis. Other aspects were also mentioned would intensive vectoring (e.g. under heavy traffic) be compatible with airborne spacing? How the interaction with E-TMA could be managed, and how arrival manager could be used? How degraded situations could be handled? 7.. SPECIFY DETAILED AIRSPACE AND PROCEDURES Since the existing spacing instructions seemed to be usable for integration onto final approach, no specific ones were developed (e.g. no merge on an axis). In addition, to avoid adding complexity, it was decided to keep the idea of a unique standard trajectory for each flow with a unique merging point. The merging point was defined as the first common point between the different converging trajectories. It was acknowledged however that the use of standard trajectories and the integration on a point would imply a modification of the working method. In fact, this new TMA configuration becomes similar to the one used in E-TMA standard trajectories and integration on a point. In E-TMA, the potential increase of availability with airborne spacing (findings from previous experiments) enables the grouping of pre-sequencing and sequencing sectors, even under very high traffic. Considering this similarity and the need for anticipation to set-up airborne spacing, it was decided to group the INI and ITM positions, and to man this unique position with an executive and a planner controller (Figure 9). Although this grouping is used today under low traffic, it would constitute, with the level of traffic envisaged (higher than June ), a second modification of the working method. Maintaining Building IAF FAF Merging point Building Maintaining Entry points IAF IAF E-TMA TMA Figure 9 Proposed E-TMA and TMA configurations Project AGC-Z-FR - EEC Report No. Volume I 9

110 EUROCONTROL COSPACE 7.. ASSESS USABILITY SETUP Participants Ludovic Boursier (Paris Orly), Claudio Colacicchi (Roma), Marcia Connick (Manchester) and Liz Jordan (London Gatwick). The objective of the November session was to assess the usability of airborne spacing in approach, in particular regarding the integration of flows, the transfer from and the co-ordination with E-TMA. The airspace, procedure and traffic were based on the principles identified in June Airspace and procedures The environment relied on the following principles A unique standard trajectory for each flow (based on the radio failure trajectory) with a unique merging point; INI and ITM positions grouped; Sectors manned with an executive and a planning controller. The airspace consisted of two approach sectors (INIR and INIO) for Charles De Gaulle and Orly airports (LFPG and LFPO), each with a single landing runway, in a West configuration (Figure 9) INIO integrating flows to Orly coming from two IAF (MOLEK at FL9 and ODRAN at FL), one base leg and one downwind leg; INIR integrating flows to Charles De Gaulle coming from two IAF (OMAKO at FL and LORTA at FL), both base leg. Though INIO layout is close to reality (there is today a third IAF North of ODRAN with integration on the downwind leg, but used by a very small proportion of the traffic), INIR is quite far from the existing sector (there are today two independent landing runways, each fed separately by a base leg and a downwind leg). The motivation was to simulate two distinct (single landing runway) configurations one with a base and a downwind legs, and the other with two base legs. Three conditions were simulated without airborne spacing (conventional control but with standard trajectories), with distance and with time based spacing. During exercises with airborne spacing, the use of spacing instructions was at controller discretion. With airborne spacing, the proposed merging points were VAS for INIO and LOR for INIR. The minimum standard separation was Nm in TMA, and the standard spacing at transfer between E-TMA and TMA were (unless explicit co-ordination) Nm without airborne spacing (as in today s letter of agreement) and with distance based spacing, or 9 seconds with time based spacing. Compared to June, IAF and trajectories have been changed to reproduce the recent modification of the Paris TMA. 9 Project AGC-Z-FR - EEC Report No. Volume I

111 COSPACE EUROCONTROL XERAM LORTA INIR BSN LFPG FAG LOR INIO VASPO ODRAN VAS FAO LFPO VAS VAS DESCT MLNS OMAKO AO AR INIO INIR INKAK MOLEK AO OKRIX AR Figure 9 November TMA sectors (simplified map) 7... Traffic Similarly to June, the traffic was built to allow the controllers to test successively three different configurations maintain a sequence from IAF until FAF, integrate one aircraft in a sequence, and integrate two sequences. For exercises with airborne spacing, traffic entered the TMA sectors already under airborne spacing (initiated by E-TMA and feed sectors), and merging to their respective IAF (at Nm for distance based spacing, or 9s for time based spacing). The traffic level was medium-high ( arrivals per hour) with sequences of up to five aircraft, equally balanced between the two IAF (Table 7). All the traffic was equipped to receive spacing instructions. There was no departure. Table 7 Example of traffic with six sequences of four or five aircraft (for a duration of about minutes) Number of aircraft from IAF Number of aircraft from IAF Sequences 7... Controller position The environment was similar to today environment, making use of progress strips. However, no arrival manager (sequencing tool) was available. Graphical markings dedicated to spacing instructions were available, consisting of markers set around the position symbols of the aircraft under airborne spacing and of its target, and of a link between them (Figure 9 Example of graphical markings). These markings served as a reminder and also allowed to visualise aircraft coming from E-TMA under airborne spacing. Project AGC-Z-FR - EEC Report No. Volume I 9

112 EUROCONTROL COSPACE Figure 9 Example of graphical markings 7... Programme The TMA session lasted 9 days ½ days of training and exploratory use, ½ day of intermediate debriefing to agree on working methods, days of measured exercises and ½ day of final debriefing. Each TMA controller played as executive controller six training exercises representing low and medium traffic load, first without airborne spacing then alternatively using distance and time based spacing. Each played one measured exercise of medium-high traffic load three times without, with distance and time based spacing. To avoid introducing a bias, the order varied. The measured period lasted minutes Limitations Several limitations of the experiment have to be mentioned. First of all, considering the lack of maturity of the use of spacing instructions in TMA and the short training period (compared to E-TMA training which lasted days), controllers were probably not familiar enough with the concept during the measured period. Due to the performance limitations of the graphic controller position update, only a limited number of range-rings were displayed, forcing controllers to use intensively the range & bearing tool. This induced additional workload. Despite this workaround, the overall simulator performance still remained poor, causing sometimes disturbances and errors. In terms of implementation, two bugs impaired the use of spacing instructions a problem of beacon detection occurring in some specific situations, limiting the use of heading then merge ; a problem in the at least behaviour, preventing aircraft to re-accelerate once spacing had been reached, thus provoking controller puzzlement and inducing the slowing down of the following aircraft. Although identified, these two bugs could not be fixed during the simulation without risking side effects (comprehensive tests not possible). 9 Project AGC-Z-FR - EEC Report No. Volume I

113 COSPACE EUROCONTROL In terms of pilot interface, modifying the spacing value (e.g. from Nm to Nm in distance based) imposed a heavy procedure ( cancel spacing, retain target, remain at least Nm behind instead of remain now at least Nm behind ). In addition, there were a certain number of pilots errors due to pilot overload and bugs in their interface. It seems that these limitations did not impair controller understanding of the concept, however they impacted on the use of spacing instructions. Consequently, if the qualitative findings can be considered as valid, the quantitative results must be considered with caution. 7.. ASSESS USABILITY FINDINGS We describe below the main findings related to airspace and procedures, method of use and controller activity. A discussion synthesises the feedback from the controllers Airspace and procedures Early in the experiment (during the exploratory use), INIO controllers used to shorten the downwind trajectory, using VAS as merging point instead of VAS. However, this created conflicting situations as, at some point, this shortened trajectory was not separated from the runway axis. Therefore, VAS had to be moved further south, at mid-distance between VAS and VAS. For INIR, LOR was kept unchanged. Receiving the traffic still merging on the IAF was felt too constraining. Indeed, as long as the aircraft and its target have not passed the merging point, they cannot be vectored (restriction of the instruction). It was thus decided (during the intermediate debriefing) to replace the E-TMA merging points by the first point upstream common to all corresponding trajectories. MOLEK was replaced by OKRIX, OMAKO by INKAK, and LORTA by XERAM. ODRAN was kept as merging point because of the long downwind leg, there was no need to vector the aircraft before they had passed it. Interestingly, E-TMA controllers mentioned that this new configuration allowed checking spacing before transfer to TMA and taking corrective action if needed. It could be noted that E-TMA and TMA configurations are now similar (Figure 9). Maintaining Building IAF Merging point FAF Merging point Building Maintaining Entry points IAF IAF E-TMA TMA Figure 9 Resulting E-TMA and TMA configurations Project AGC-Z-FR - EEC Report No. Volume I 9

114 EUROCONTROL COSPACE 7... Method of use Controllers used spacing instructions even for integration onto final approach, i.e. for building and maintaining sequences. Although INIR was found more difficult than INIO (no downwind that would provide space and flexibility, sometimes head-on situations when vectoring aircraft), spacing instructions were used in the same manner for both sectors. Depending on situations (e.g. aircraft from the same flow or from different flows, spacing existing or to be created), they maintained, modified on-going airborne spacing, or created new ones. To describe the method used by controllers during the experiment, we detail below basic situations with two aircraft (Table ). Situations with multiple aircraft were managed by combining the method used for these basic situations. It was observed that with airborne spacing, controllers tended to integrate successively blocks of aircraft of one flow, then of the other, i.e. segregating flows. Whereas, without airborne spacing, controllers integrated aircraft per aircraft, i.e. mixing flows. From the measured exercises, the order at the FAF of the sequences with airborne spacing (distance and time) was compared to the order of the same sequences without. Four typical configurations were identified depending on flows being mixed or segregated, with and without (Table 9). This confirmed the observations and the controllers feedback. Table Basic situations with two aircraft Aircraft Spacing Method of use Spacing instruction From same IAF From different IAF Existing To be created Existing To be created Kept in remain, trajectory shortened by giving a direct Remain cancelled and use of either conventional heading followed by merge, or heading then merge Use of merge Use of either conventional heading followed by merge, or heading then merge Unchanged (same target, same type of instruction) Modified (same target, different type of instruction) Created (new target) Created (new target) Table 9 Comparison of sequence configurations Without With Occurrences Mixed Mixed Segregated Segregated Mixed Segregated Segregated Mixed Undetermined (aircraft removed ) In distance based, the desired spacing was possibly modified (from Nm exactly to Nm at least ). It could be noticed that, when giving a direct, the spacing definition is different. However, considering the small deviation angle in situations considered, the variation of actual spacing was small. A merge would be more appropriate but would have imposed to issue a new spacing instruction. In time based, to increase spacing from 9s to s for wake turbulence. To keep manageable control conditions when inconsistent behaviours resulting from limitations mentioned previously occurred. 9 Project AGC-Z-FR - EEC Report No. Volume I

115 COSPACE EUROCONTROL 7... Typical examples of use Below, we describe two typical examples of use on INIO and INIR. Each is illustrated by a series of figures (Figure 9 to Figure for INIO and Figure to Figure for INIR). A modified phraseology is given to easily follow the succession of steps. The phraseology proposed and used during the experiment is illustrated in section.. In particular, for the target selection, the target callsign is not used (SSR code instead) and for the spacing instruction, the target identifier (SSR code) is not repeated. The spacing is cancelled when the aircraft is transferred to tower. It should be noticed that the screen captures were made with the replay tool and not from the controller interface. Example INIO, initial situation IBE from ODRAN, number one in the sequence. AF7LZ to MOLEK, number two. LBAW (not visible) remaining behind AF7LZ and AF7PO (not visible) remaining behind LBAW. Example INIO, step IBE direct VAS. AF7LZ select target IBE. Figure 9 Example INIO, step Project AGC-Z-FR - EEC Report No. Volume I 97

116 EUROCONTROL COSPACE Example INIO, step AF7LZ continue heading then merge VAS behind IBE (initial spacing was too small). LBAW and AF7PO (not visible) continue present heading. Figure 97 Example INIO, step Example INIO, step LBFK select target AF7PO. Figure 9 Example INIO, step 9 Project AGC-Z-FR - EEC Report No. Volume I

117 COSPACE EUROCONTROL Example INIO, step AF7LZ reports merging VAS (spacing reached). Figure 99 Example INIO, step Example INIO, step LBAW direct VAS. Figure Example INIO, step Project AGC-Z-FR - EEC Report No. Volume I 99

118 EUROCONTROL COSPACE Example INIO, step AF7PO direct VAS. Figure Example INIO, step Example INIO, step 7 LBFK merge VAS behind AF7PO. Figure Example INIO, step 7 Project AGC-Z-FR - EEC Report No. Volume I

119 COSPACE EUROCONTROL Example INIR, initial situation DLH to LORTA DLH7 to LORTA, remaining behind DLH ANA to LOR AZA to OMAKO, remaining behind ANA AZA77 (not visible) to OMAKO, remaining behind AZA Example INIR, step AF7VC heading Figure Example INIR, step Example INIR, step AF7VC direct LOR DLH select target AF7VC ANA heading AZA cancel spacing, retain target (*) AZA77 cancel spacing, retain target (*) * The link should be orange (target selection status). Figure Example INIR, step Project AGC-Z-FR - EEC Report No. Volume I

120 EUROCONTROL COSPACE Example INIR, step DLH merge LOR behind AF7VC Figure Example INIR, step Example INIR, step AF7VC select target GIL9C AZA heading Figure Example INIR, step Project AGC-Z-FR - EEC Report No. Volume I

121 COSPACE EUROCONTROL Example INIR, step AF7VC merge LOR behind GIL9C AZA77 heading Figure 7 Example INIR, step Example INIR, step ANA select target DLH7 Figure Example INIR, step Project AGC-Z-FR - EEC Report No. Volume I