Docstoc

FLIGHT TEST VALIDATION REPORT – D39

Document Sample
FLIGHT TEST VALIDATION REPORT – D39 Powered By Docstoc
					                          TECHNICAL REPORT
CONTRACT N° : GRD1-2000-0228

PROJECT N° :

ACRONYM : MA-AFAS

TITLE :     THE MORE AUTONOMOUS - AIRCRAFT IN THE FUTURE
                 AIR TRAFFIC MANAGEMENT SYSTEM

           FLIGHT TEST VALIDATION REPORT – D39
AUTHOR : QINETIQ (UK)

PROJECT CO-ORDINATOR : BAE SYSTEMS

PRINCIPAL CONTRACTORS :
         Airtel ATN Ltd (Ireland)   QinetiQ (UK)
         ETG (Germany)              EUROCONTROL (France)
         NLR (Netherlands)

ASSISTANT CONTRACTORS:
          Airsys ATM (France)       Galileo Avionica (Italy)
          AMS (Italy)               DLR (Germany)
          FRQ (Austria)             Indra Sistemas (Spain)
          NATS (UK)                 SCAA (Sweden)
          S-TT (Sweden)             Skysoft (Portugal)
          SOFREAVIA (France)        Stasys Limited (UK)



Report Number : QINETIQ/S&E/AVC/CR031041 – D39
Project Reference number :
Date of issue of this report : 19 Jun 2003
Issue No. 1.0

PROJECT START DATE : 01 Mar 2000                  DURATION : 40 months




                                    Project funded by the European Community
                                    under the ‘Competitive and Sustainable
                                    Growth’ Programme (1998-2002)
                            Customer Information
Customer Reference Number
Project Title                         The More Autonomous-Aircraft in the
                                      Future Air Traffic Management System
Company Name                          BAE SYSTEMS
Customer Contact                      A Hanna

Contract Number                       GRD1-2000-0228
Milestone Number                      D39
Date Due (dd/mm/yyyy)                 30/06/2003




ii                                                         QinetiQ/S&E/AVC/CR031041
Authorisation


 Prepared by               I Mansfeld
           Title           Senior Flight Test Engineer


           Signature


           Date
           Location        QinetiQ, MoD Boscombe Down



 Approved by               Dr D Walker
          Title            CM – Telematic Solutions


           Signature


           Date


 Principal authors
 Authorised by             Mr A Hanna
            Title          BAE Systems


           Signature
           Date
           Name            Mr A Wolfe
           Appointment     Senior Scientist
           Location        QinetiQ Boscombe Down



           Name
           Appointment
           Location




QinetiQ/S&E/AVC/CR031041                                 iii
Record of changes




             Issue    Date                           Detail of Changes
              A      29/04/03   Draft
              B      10/06/03   Revised Draft after internal review
              C      19/06/03   Completion of all chapters and additional update from review
              D      30/06/03   Minor refinements




     iv                                                            QinetiQ/S&E/AVC/CR031041
List of contents



Authorisation                                                                           iii

Record of changes                                                                       iv

List of contents                                                                         v

List of Figures                                                                        vii

List of Tables                                                                         viii

1         Introduction                                                                   1
1.1       Scope of Document                                                              1
1.2       Programme                                                                      1

2         BAC 1-11 Validation Flights                                                    2
2.1       Sortie Overview                                                                2
2.2       Summary Table                                                                  2
2.3       Trial Routes                                                                   3
2.4       Trial Scripts                                                                  3
2.5       Sortie Details                                                                 3

3         Boscombe Down Trials                                                           4
3.1       14th January 2003                                                              4
           th
3.2       7 February 2003                                                                9
               th
3.3       19        February 2003                                                       13
               th
3.4       27 February 2003                                                              19
               th
3.5       11 March 2003                                                                 26
               th
3.6       14 March 2003                                                                 27
               th
3.7       19 March 2003, Sortie No. 762.                                                32
               th
3.8       19 March 2003                                                                 34
               th
3.9       19 March 2003                                                                 39
3.10      20th March 2003                                                               40
3.11      Summary of GBAS Approach Guidance Presented on the Primary Flight Display.    46
3.12      Summary of Boscombe Results                                                   53

4         Rome Trials                                                                   56
               th
4.1       24 March 2003                                                                 56
QinetiQ/S&E/AVC/CR031041                                                                 v
4.2        25th March 2003                                                 58
             th
4.3        25 March 2003                                                   62
4.4        26th March 2003                                                 64
4.5        26th March 2003                                                 67
4.6        28th March 2003                                                 69
4.7        28th March 2003, Sortie No. 774.                                72
4.8        Summary of Rome Results                                         73

5          Overall Summary of MA-AFAS Flight Trials                        76

6          Conclusions and Recommendations                                 78

7          References                                                      80

8          List of abbreviations                                           81

A          Appendix                                                        83
A.1        Original QNQ1 UK Trials Route                                   83
A.2        UK route QNQ1 (revised)                                         85
A.3        Rome Route QNQ6                                                 87

Report documentation page                                                  89




      vi                                              QinetiQ/S&E/AVC/CR031041
List of Figures


Figure 3.1-1: Route Flown for First MFMS Trial                                      4
Figure 3.1-2 Demanded and Actual Bank Angle, 1st Circuit, 14th Jan. 03              6
Figure 3.1-3 Demanded and Actual Computed Air Speed, 1st Circuit, 14th Jan. 03      7
Figure 3.2-1 Route used for MFMS Flight, 7th Feb. 03                                9
Figure 3.3-1 Tracks for BAC 1-11 and SIM00                                         14
Figure 3.3-2 Spacing of BAC 1-11 from SIM00                                        14
Figure 3.3-3 Tracks of BAC 1-11 and SIM02                                          15
Figure 3.3-4 Spacing of BAC 1-11 from SIM02                                        15
Figure 3.3-5 Tracks of BAC 1-11 and SIM03                                          16
Figure 3.3-6 Spacing of BAC 1-11 from SIM03                                        16
Figure 3.4-1 GBAS Position Reports during Precision Departure                      20
Figure 3.4-2 Accuracy of GBAS 3D Position Data during Departure                    20
Figure 3.4-3 Freeze of SBAS Data within the FMS, 27th Feb. 03                      22
Figure 3.4-4 Tracks of BAC 1-11 and SIM04                                          23
Figure 3.4-5 Spacing of BAC 1-11 from SIM04                                        23
Figure 3.4-6 ADS-B Flight Data Received by EEC                                     24
Figure 3.6-1 Tracks of BAC 1-11 and SIM03                                          29
Figure 3.6-2 Spacing of BAC 1-11 from SIM03                                        29
Figure 3.8-1: QNQ5 Route used for Test Flight in Rome, 19th March 03               34
Figure 3.8-2 Tracks of BAC 1-11 and SIM02 for 1st Pass-Behind Manoeuvre            35
Figure 3.8-3 Spacing of BAC 1-11 from SIM02 during 1st Pass-Behind Manoeuvre       35
Figure 3.8-4 Tracks of BAC 1-11 and SIM02 for 2nd Pass-Behind Manoeuvre            36
Figure 3.8-5 Spacing of BAC 1-11 from SIM02 during 2nd Pass-Behind Manoeuvre       36
Figure 3.8-6 Plot of ADS Position Data during Rome Test Flight                     38
Figure 3.10-1 Precision Departure Route With Indication of GBAS Operational Mode   41
Figure 3.10-2 Accuracy of GBAS Position Fix Referenced to GPS Truth Track          41
Figure 3.10-3 Tracks of the BAC 1-11 and SIM02                                     42
Figure 3.10-4 Spacing of BAC 1-11 from SIM02                                       42
Figure 3.10-5 Approach Route with Indication of GBAS Operational Mode              44
Figure 3.10-6 Accuracy of GBAS Position Fix Relative to GPS Truth Track            44
Figure 3.10-7: MFMS Lateral Guidance Performance                                   45
Figure 3.11-1: Representative STAR for Boscombe Down                               46
Figure 3.11-2: 7th February GBAS Approach 1 using 3º Glide Slope                   47
Figure 3.11-3: 7th February GBAS Approach 1 using 4.5º Glide Slope                 48
Figure 3.11-4: 19th February GBAS Approach using 3º Glide Slope                    49
Figure 3.11-5: Glide Slope Comparisons for Approaches on 14 March                  51
Figure 3.11-6: GBAS and MA-AFAS Glide Slope Differences                            52
Figure 4.2-1 Spacing during first Pass-Behind Manoeuvre, 25th Mar. 03              59
Figure 4.2-2 Spacing during second Pass-Behind Manoeuvre, 25th Mar. 03             60
Figure 4.4-1: Spacing Distance during Merge Manoeuvre, 26th Mar. 03                65
Figure 4.6-1: Spacing of BAC 1-11 from ATTAS during Merge Behind, 28th Mar. 03     70




QinetiQ/S&E/AVC/CR031041                                                           vii
List of Tables


Table 2.2-1: Flight Trials using MA-AFAS on the QinetiQ BAC 1-11                         2
Table 3.11-1: GBAS Guided Approaches                                                    46
Table 4.2-1: Updated VDL4 Identity Codes                                                58




     viii                                                          QinetiQ/S&E/AVC/CR031041
                           This page is intentionally blank




QinetiQ/S&E/AVC/CR031041                                      ix
1        Introduction

1.1      Scope of Document

         This document describes the validation flight testing of the Avionics Package (AvP),
         developed within the MA-AFAS project, in the QinetiQ BAC 1-11 avionics research
         aircraft and the DLR VFW-614 ATTAS aircraft. The initial test flights, using only the
         BAC 1-11, took place from QinetiQ Boscombe Down in the UK, during the period
         January to March 2003. These flights were intended to verify the performance of the
         MA-AFAS FMS (MFMS) and the associated air/air and air/ground data link
         communications environment. This also included the procedures developed for the
         different Airborne Separation Assurance System (ASAS) manoeuvres, these being
         carried out in a real airborne environment. While these flight trials took place,
         development continued of the various MA-AFAS functional components. The flight
         trials at Boscombe Down were the precursor to the main trials that took place from Rome
         Ciampino Airport in Italy between 24th and 28th March 2003. Both aircraft participated in
         these trials to investigate the performance when the ASAS manoeuvres were performed
         with live aircraft in the roles of both the host aircraft and the target aircraft.

         A further set of taxi management and flight trials, using additional developments of the
         ASAS functionality, were conducted on the DLR ATTAS aircraft in May 2003. The
         results of these trials are provided in a separate document [4], Annex A to this flight
         trials report. An additional document, Annex B of this current D39 report, contains a
         detailed summary by NATS of the performance of the Ground- and Satellite-Based
         Augmentation Systems (GBAS and SBAS) that were used to support the flight trials at
         Boscombe Down [5]. The GBAS provided approach guidance information to the pilot
         during these flights and it was intended that the MFMS would also use this system to
         provide a precision approach capability. A further MA-AFAS document, D40 [6],
         describes the airline pilot evaluation of the MFMS and the ASAS manoeuvres conducted
         on the NLR simulator in Amsterdam.

         Prior to the flight trials programme and in support of the continued system development
         once it had started, ground simulation testing of the MA-AFAS components was carried
         out at the different trials sites. This was intended to resolve any functional performance
         problems with the system before it was committed to the real airborne environment. The
         results of these ground tests are described in the MA-AFAS document D37 [3].

1.2      Programme

         The planned flight test programme and deliverables are fully described with the MA-
         AFAS D32 document [1]. In particular, Annex A of document D32 refers to the flight
         tests that were to be carried out from Boscombe Down, Annex B refers to the tests
         conducted by DLR from Braunschweig and Annex C to flight tests from Rome
         Ciampino. Finally, Annex D of D32 defined the tests that were to be performed at NLR
         for the pilot evaluations.



QinetiQ/S&E/AVC/CR031041                                                                Page 1 of 90
2              BAC 1-11 Validation Flights

2.1            Sortie Overview

               The initial development sorties were all flown out of the QinetiQ Boscombe Down
               airfield in the southern UK. Not all of these used the entire MA-AFAS FMS (MFMS)
               equipment, for example when flights were dedicated to assessing problems with the VHF
               Data Link mode 4 (VDL4) data link. Due to these data link problems, the trials
               programme was re-arranged to allow for a one day visit to the Rome operating area so
               that both the VDL4 and the proposed route could be tested prior to the final flight phase.
               The final week of the trials was flown out of Rome Ciampino airport, and during this
               period the MFMS was exercised for the first time using a live ‘target’ aircraft using
               ADS-B via the VDL4.

2.2            Summary Table

    Serial      Date         Location        S/W                                 Notes
                                            version
      1         14/01    Boscombe Down        C5      Initial Shakedown using short route. Basic FMS operation
                                                      only, no ASAS.
      2         07/02    Boscombe Down        D2      Short route. Simulated static ASAS target used.
      3         19/02    Boscombe Down      D2-ER1    Long Route with simulated ASAS targets. VDL4 coverage
                                                      explored.
      4         27/02    Boscombe Down        D3      Long Route with added ‘loop’ for Merge. Western extent
                                                      reduced for VDL4. SID flown.
      5         11/03    Boscombe Down        N/A     VDL4 Range checks only. No MFMS.
      6         14/03    Boscombe Down       E1A-2    Met data introduced.
      7         19/03    Boscombe Down       E1A-3    Airways transit to Rome for VDL4 checks.
      8         19/03     Rome Ciampino      E1A-3    Rome route using simulated targets. Voice and VDL4
                                                      coverage were explored.
      9         19/03     Rome Ciampino      E1A-3    Airways transit.
      10        20/03    Boscombe Down       E1A-3    The full ‘Rome’ version exercised with simulated targets
      11        24/03    Boscombe Down       E1A-3    Airways transit to Rome. Checks of ATTAS VDL4 link.
      12        25/03     Rome Ciampino      E1A-3    ATTAS target via VDL4
      13        25/03     Rome Ciampino      E1A-3    ATTAS target via VDL4
      14        26/03     Rome Ciampino      E1A-3    ATTAS target via VDL4
      15        26/03     Rome Ciampino      E1A-3    ATTAS target via VDL4
      16        28/03     Rome Ciampino      E1A-3    ATTAS target via VDL4
      17        28/03     Rome Ciampino      E1A-3    Airways transit

                                    Table 2.2-1: Flight Trials using MA-AFAS on the QinetiQ BAC 1-11



Page 2 of 90                                                                       QinetiQ/S&E/AVC/CR031041
2.3      Trial Routes

         The trials were flown over planned routes. The initial planning for the UK trials required
         co-ordination through controlled airspace to permit the BAC 1-11 to fly on non-deviating
         tracks (i.e. not to be subject to ATC tactical changes of heading or altitude). QinetiQ
         Boscombe Down, as a part of the normal trials planning and support, carried out liaison
         with the UK Airspace Utilisation Service (AUS) to produce an Aeronautical Co-
         ordination Notice (ACN) [2]. This was promulgated 6 weeks before commencement of
         the trials, and is contained in Appendix A.1. The route it describes is the one referred to
         as ‘QNQ1’ within the MFMS.

         During the development, several modifications were made to the route in order to enable
         the ASAS functionality to be fully exercised. The most used version of this is shown as
         Appendix A.2.

         For Rome trials, a similar and lengthy liaison process with ENAV was successfully
         followed to produce the route QNQ6, promulgated by ENAV as a NOTAM (Notice to
         Airmen). This route is shown as Appendix A.3.

2.4      Trial Scripts

         For each sortie, a comprehensive script was developed to suit both the route to be flown
         and the functions to be exercised .

2.5      Sortie Details

         Each of these sorties will now be covered in more detail as a series of flight reports. The
         sorties have been divided up to cover those associated with development of the MFMS
         (described in Chapter 3) and those relating to the live aircraft trials in Rome (described in
         Chapter 4). Each flight report is presented in a series of functional groups representing
         the systems that integrate to provide the overall MA-AFAS. Due to the nature of the
         development programme, data is not available for all sections for all of the sorties.
         Within each functional group, the information has been collated from that provided by
         the system specialists.




QinetiQ/S&E/AVC/CR031041                                                                   Page 3 of 90
3              Boscombe Down Trials

3.1            14th January 2003

3.1.1          Sortie Objectives

               The primary intention of this first flight with the MFMS on-board the BAC 1-11 was to
               ensure that the MFMS would operate correctly when integrated with the various other
               avionics systems installed on the aircraft. Of greatest interest was the interaction of the
               MFMS with autopilot in order to confirm that the MFMS demands could be handled by
               the autopilot without any significant problems in the airborne environment and that there
               was smooth tracking of the transitions in the demand values by the autopilot. The other
               key factors included verification of simulation results conducted with the ground based
               avionics rig to confirm that the MFMS would continue to function throughout the
               duration of the flight without any system failures or runtime problems.

               At this stage, there was still a limitation in the basic functionality of the MFMS so that
               the system could only predict a trajectory containing climb and cruise phases, but no
               descent. The FMS was also assuming zero wind conditions and an ISA temperature
               gradient with altitude when predicting the aircraft's trajectory. Additionally, only the
               lateral version of the Navigation Display was available to the pilot at this stage of the
               system development.




                                                        Figure 3.1-1: Route Flown for First MFMS Trial
Page 4 of 90                                                                   QinetiQ/S&E/AVC/CR031041
         For this first flight, a short route (stored as company route BOSC02) was flown in order
         to verify the basic FMS and validate experience with the ground test simulation. This is
         shown in Figure 3.1-1, the circular route being flown in the clockwise direction. No
         VDL4 transponder was available for use on this flight.

3.1.2    MFMS Report

         The FMS equipment had been installed in the cockpit of the BAC 1-11 to allow the left-
         hand pilot to operate the system. The right-hand pilot had the standard instrumentation
         available in order to act as the safety pilot during the flight. The MCDU had been
         mounted in the centre pedestal area just aft of the throttle quadrant so that it could be
         viewed by both pilots and allowing the safety pilot to monitor the intended actions of the
         FMS. The left-hand pilot was also provided with a track-ball, mounted on the right-hand
         armrest that permitted the pilot to interact with the soft keys on the electronic Navigation
         Display. This display was also only available to the left-hand pilot, having been mounted
         in the console directly in front of the pilot.

                                                                               The picture at left
                                                                               shows the Electronic
                                                                               Flight Instrument
                                                                               System (EFIS) in the
                                                                               BAC 1-11. This is in
                                                                               the Left Side cockpit
                                                                               only, and comprises
                                                                               a Primary Flight
                                                                               Display of Attitude,
                                                                               Speed and Altitude
                                                                               on the left, and the
                                                                               Experimental Map
                                                                               Display on the right.
                                                                               The Map display is
                                                                               showing the MFMS
                                                                               output during an
                                                                               ASAS Pass-Behind
                                                                               manoeuvre.




         In the cabin of the aircraft, the scientific crew that was monitoring the performance of the
         system could view a repeater of the Navigation Display. An IHTP laptop PC was also
         connected via ethernet to the PC cards in the FMS cabinet, allowing additional
         monitoring of the system behaviour, including the emulation of the MCDU and to detect
         any possible problems being encountered.

         For this first flight, the pilot was able to use the MCDU to initialise the FMS, including
         selecting the primary navigation data source to be the inertial reference system (IRS),
         inputting the fuel load and zero fuel weight of the aircraft and finally selecting the
         company route. The pilot also input the required cruise altitude (in this case FL240,
         although the MCDU required the entry in feet) before generating the trajectory while the
         aircraft was still on the ground at the stand. In its current form, the FMS used zero as the
         take-off time, viewing the generation in terms of an elapsed time. When predicting the
         lateral route, the FMS used 8nm constant radius turns. The vertical and speed profiles
QinetiQ/S&E/AVC/CR031041                                                                  Page 5 of 90
               were predicted based on a set of subphases with a continuous climb from take-off to the
               cruise altitude. The initial en-route climb speed was defined to be 230 knots CAS with a
               speed change on reaching FL100 to 250 knots CAS and a further acceleration to 260
               knots CAS once the aircraft was at the cruise level.

               Once the aircraft had taxied to the holding point for runway 23 at Boscombe Down, a
               further trajectory generation was carried out successfully. The trajectory was not
               activated until after the aircraft had taken off. With the autopilot first engaged in its basic
               modes, the FMS demands were then engaged by the pilot selecting the Lateral
               Navigation (LNAV) and Profile (PRFL) modes on the autopilot control panel. The
               aircraft was at about 5400ft once the FMS demands were engaged through the autopilot.
               The aircraft was just under 1.5nm off to the left of its originally intended track at this
               stage, but the bank demand from the FMS resulted in the aircraft capturing back on to
               track just at the start of the turn at waypoint B15. Lateral guidance was then maintained
               throughout the remainder of the flight until the aircraft was about 3nm (on track for
               runway 05) before being overhead Boscombe Down (waypoint EGDM) when the pilot
               disengaged the system in order to perform an approach to runway 23. Figure 3.1-2 shows
               how the autopilot was able to track the bank angle demand from the FMS, there being a
               latency of 1 or 2 seconds in the autopilot's response to a change in the demand. It can
               also be seen that the pilot had briefly disengaged the autopilot at about 2770 seconds,
               although the MFMS demands were soon re-engaged again.




                                 Figure 3.1-2 Demanded and Actual Bank Angle, 1st Circuit, 14th Jan. 03
               The Profile demands were also successfully used by the autopilot, the FMS initially
               demanding a climb to FL240 at 250kts CAS and with a climb power setting of 94% HP
               RPM. Although the aircraft was lower than FL100 when the demands were fully engaged
Page 6 of 90                                                                      QinetiQ/S&E/AVC/CR031041
         with the autopilot, the MFMS, in these earlier versions, was deriving the speed demands
         based on the aircraft's position along the route. Consequently, due to the aircraft having
         been delayed on its initial climb out from Boscombe Down, it was not as high as the
         MFMS had originally predicted and had passed the point on the route where the MFMS
         had expected it to reach FL100. Therefore the speed demand had already transitioned to
         250kts (see Figure 3.1-3). As seen with the ground test runs using the aircraft model rig,
         the FMS had occasional problems calculating the aircraft's relative position to the
         subphase transition points. In this case, it was known that it did not correctly determine
         the distance from the top of climb point where the aircraft was to accelerate to 260kts
         CAS. Therefore this speed change did not occur, although there was a brief increase in
         the speed demand during the climb itself, before returning to 250kts again. This was
         directly related to the problem with a relative distance calculation. Once the aircraft had
         reached waypoint C in the cruise, the Profile demands were disengaged from the
         autopilot and the pilot directly dialled in a descent to FL50 on the autopilot. The lateral
         bank demand from the FMS was still engaged, however




                 Figure 3.1-3 Demanded and Actual Computed Air Speed, 1st Circuit, 14th Jan. 03
         The aircraft landed at the end of the approach to runway 23 and the FMS system was
         reset in order to repeat the flight. The system behaved identically to the first flight,
         although there were some problems with the initial engagement of the autopilot due to it
         having not fully reinitialised after the landing (this was an issue on this first flight only
         and the problem was not encountered on any subsequent MFMS flight). The MFMS
         demands were not fully engaged through the autopilot this time until the aircraft was at


QinetiQ/S&E/AVC/CR031041                                                                   Page 7 of 90
               about FL120. As before, the Profile demands were disengaged at waypoint C while the
               lateral demands were maintained until about 3nm before EGDM at FL50.

               The flights showed consistency with the results obtained from the simulation runs. They
               proved that the basic FMS was functioning correctly with the other on-board systems and
               was capable of using the data from these various systems such as the inertial reference
               system, the air data computer, the engine instrumentation system and the autopilot.
               Similarly, the autopilot was able to function properly using the demands from the FMS
               with no divergent trends or other problems in tracking these demands to guide the aircraft
               both in the lateral and profile modes along the predicted trajectory.




Page 8 of 90                                                                  QinetiQ/S&E/AVC/CR031041
3.2      7th February 2003

3.2.1    Sortie Objectives

         For the second trials flight with the MFMS, advancements in the development of the
         system now permitted a complete flight profile to be predicted from take-off to the
         approach gate altitude at the arrival airport (in this case Boscombe Down again). The
         STAR produced for this flight took the aircraft to a waypoint, DM002, just prior to the
         turn on to the final approach for runway 23 at a distance of about 8nm from touchdown.
         The gate altitude was set to be FL40 and it was intended that the predicted trajectory
         should maintain FL40 throughout the region of the STAR.

         The route had also been modified (now company route BOSC06) so that the aircraft
         would effectively fly the BOSC02 route in reverse, i.e. anticlockwise (see Figure 3.2-1).
         This was to allow a suitably long straight section of track on which to perform a pass-
         behind manoeuvre and also to provide a smoother feed into the STAR, which was to the
         south of the airfield. The cruise altitude was reduced to FL210 in order to permit a better
         duration of cruise.




                                             Figure 3.2-1 Route used for MFMS Flight, 7th Feb. 03
         The functionality of the system had been improved and it was intended to test a pass-
         behind manoeuvre being performed relative to a target aircraft selected by the pilot via
         the MCDU ASAS pages. In this case, a simulated target aircraft was created within the
         FMS that would conflict with the trajectory of the BAC 1-11. In order to ensure that the
         simulated target aircraft was going to be in the correct position and at the right time to
         create a conflict, this aircraft was defined to have a ground speed of only 1kt and to be at
QinetiQ/S&E/AVC/CR031041                                                                  Page 9 of 90
           a position about 5nm along the leg from waypoint AGIBS to DM005. This aircraft was
           also set up to be following a north-south track, which was about 100º to the proposed
           track of the BAC 1-11.

           Additional updates had been incorporated in order for the FMS to utilise the data from
           the two GPS systems, SBAS and GBAS, including the use of this data to assign UTC
           time within the FMS. The FMS defaulted to using GBAS as its primary navigation data
           source, the pilot being able to change this selection via the MCDU, if required.
           Improvements had also been made to the determination of relative distance from the
           aircraft to the subphase change points in order to overcome the problem where speed
           changes were being triggered at incorrect points along the route.

           As for the previous flight, the VDL4 transponder was not currently available for testing.

3.2.2      MFMS Report

           Similar to the first flight, the pilot used the MCDU to initialise the FMS and generate a
           trajectory for the BOSC06 company route. As well as inserting the required cruise
           altitude, the pilot also now had to input the Estimated Off-Blocks Time (EOBT) and the
           Computed Take-Off Time (CTOT). It was known from the ground tests with the aircraft
           model rig that the FMS was computing a descent that was far too shallow for what the
           aircraft would typically achieve and therefore the predicted descent was not reaching
           FL40 by the beginning of the STAR. The predicted location for the top of descent was
           suitable, however, for achieving this required altitude within the given distance when it
           was actually flown. For the purpose of this trial, this was not a significant problem.

           Prior to take-off, the primary navigation source was changed from GBAS to SBAS to
           ensure a continuous data source for the flight. As on the first flight, the trajectory was not
           activated until the aircraft was airborne. After the aircraft had been cleaned up (flaps and
           undercarriage retracted) and the aircraft established in the climb, then the FMS lateral
           and profile demands were engaged through the autopilot. The initial FMS speed demand
           was 230 knots CAS that was then increased to 250 knots as the aircraft passed the
           subphase change point on the route. A further increase to 260 knots CAS occurred when
           the aircraft attained the cruise altitude of FL210. Prior to the demand for descent to 5900
           feet (final altitude of the trajectory), the FMS demanded a deceleration back to 250 knots
           CAS, in accordance with the predicted flight profile.

           Once into the descent, the pilot was able to enter the details for the pass-behind
           manoeuvre into the FMS via the MCDU. This was to simulate the response to an
           instruction from ATC to perform a manoeuvre in which ATC are delegating partial
           responsibility to the pilot for ensuring the required manoeuvre spacing from the other
           conflicting traffic. Having identified that the minimum separation between the two
           aircraft will be compromised, ATC could request that one aircraft performs a pass-behind
           manoeuvre with a defined spacing distance to be observed and to return to the original
           route at an assigned waypoint. In this case, the BAC 1-11 was to pass-behind the other
           traffic, SIM01, with a minimum manoeuvre spacing of 5nm and to return to the original
           route at waypoint DM005. The traffic identity, the type of manoeuvre, the minimum
           spacing and the resume waypoint were all entered by the pilot through the MCDU.


Page 10 of 90                                                                QinetiQ/S&E/AVC/CR031041
         The pilot was also able to select an option on the lateral Navigation Display using the
         track-ball in order to display any other traffic in the vicinity for which ADS-B reports
         had been received. After entering the target aircraft's identity into the ASAS page of the
         MCDU, this traffic was highlighted on the Navigation Display in orange to ease
         identification. A trajectory could only be generated, however, once the BAC 1-11 was on
         a straight leg and therefore this delayed the manoeuvre until the aircraft had completed
         its turn at waypoint A. The conflicting aircraft, SIM01, was now about 2nm to the right
         of the leg from waypoint AGIBS to DM005. At this point, the pilot was able to use the
         MCDU to request the FMS to compute a trajectory incorporating the pass-behind
         manoeuvre. This was successfully achieved and the trajectory activated, although
         activation had to be performed within 15 seconds because this was the time period ahead
         of the aircraft that the FMS maintained current track before inserting any start of turn to
         avoid the conflict situation.

         With the target aircraft being effectively static, the trajectory generated by the FMS
         deviated the BAC 1-11 by 3.8nm to the left of its original track between AGIBS and
         DM005, allowing for the fact that the target aircraft was already 2nm to the right of this
         track (see Figure 3.2-1). The route distance for this pass-behind manoeuvre to its
         termination at waypoint DM005 was 38.6nm, which was only about 0.5nm greater than
         that for the original route. However, this partly benefited from the fact that the
         manoeuvre bypassed the waypoint AGIBS at which there was 7º turn to the left towards
         DM005, i.e. in the same direction as the pass-behind manoeuvre was turning the aircraft.

         Since this manoeuvre had not originally been intended for use during the descent, it did
         have an effect on the predicted speed during the course of the manoeuvre. The FMS
         manoeuvre generator assigned time constraints to each of the track change points that it
         defined to create the pass-behind. In this case, the system determined the time constraints
         based on the aircraft approximately maintaining its current ground speed. Clearly, during
         the descent, this is not the case, the ground speed continuously decreasing for a constant
         CAS value. Thus, the trajectory predictor computed that a higher CAS value would be
         needed to ensure the time constraints were met. Hence there was an increase in the CAS
         demand to the autopilot to 259 knots on activation of this new trajectory. The aircraft
         successfully followed the lateral route (the turn radii for the pass-behind manoeuvre
         being based on a bank angle of 20º) and the profile demands. The prediction had also
         now computed a descent to FL40 and therefore the aircraft continued down to this
         altitude for entry into the STAR. The profile demands were disengaged from the
         autopilot as the aircraft approached waypoint DM001 and then the lateral demands were
         disengaged mid-way between DM001 and DM002, so that the aircraft could be
         positioned for a manual approach to runway 23 using GBAS guidance signals.

         Following the landing, the FMS was reset on the ground and the pilot re-initialised the
         information via the MCDU in order to perform a repeat of the first flight. Once again, the
         trajectory was generated while the aircraft was on the ground with an EOBT of 1420
         UTC and a CTOT of 1425 UTC. The aircraft actually took-off at about 1420 UTC and
         the trajectory was only activated once the aircraft was airborne and the pilots had
         completed cleaning up the aircraft.

         With the lateral and profile demands engaged through the autopilot, on this flight an in-
         flight trajectory generation was also performed during the climb with the aircraft on the
QinetiQ/S&E/AVC/CR031041                                                                 Page 11 of 90
           leg to waypoint ADSON. This was successful and on activation, it was noted that there
           was no disturbance in the autopilot behaviour as the FMS started sending demands based
           on this new trajectory. Effectively, there were no changes in the profile demands and no
           significant variation in the lateral demands either.

           As before, the pass-behind manoeuvre was set-up by the pilot entering the target identity
           (SIM01), the minimum spacing distance (5nm) and the resume waypoint name (DM005)
           into the FMS via the MCDU. On this occasion, the FMS initially reported errors
           attempting to generate the pass-behind trajectory due to trouble meeting the time
           constraints. As previously mentioned, the implementation had not originally envisaged
           the use during the descent. However, a trajectory was successfully generated after a few
           attempts and then activated. As for the previous run, the speed demand increased, this
           time to 262 knots CAS. The aircraft correctly followed the planned lateral path that was
           intended to ensure a spacing distance of 5nm from the conflicting target aircraft.

           On this occasion, the conflicting aircraft, SIM01, was 1.1nm to the right of the track from
           AGIBS to DM005 and the resultant pass-behind trajectory deviated the BAC 1-11 by
           about 4.9nm to the left of this track leg. The total route length for this manoeuvre leading
           to DM005 was 35.3nm, which was 1nm longer than the originally intended route.

           Both the lateral and profile demands were subsequently disengaged from the autopilot
           prior to waypoint DM001, the aircraft being in level flight at FL40 and the FMS having
           demanded a speed reduction first to 250 knots CAS and then to 230 knots CAS for
           manoeuvring in the STAR.




Page 12 of 90                                                               QinetiQ/S&E/AVC/CR031041
3.3      19th February 2003

3.3.1    Sortie Objectives

         For this flight, the full trials route (company route QNQ1) was to be used for the first
         time (see Appendix A.1), this provided areas where the aircraft was cleared to perform
         ASAS manoeuvres when simulating delegation being granted by the controller to the
         pilot. The intention was to perform four pass behind manoeuvres during the course of the
         flight. As for the last flight, the target aircraft were being simulated within the FMS,
         although this time they were moving at typical ground speeds to create a more
         representative situation. The default turn radius used for the standard en-route trajectory
         prediction had also been reduced to 5nm, since this was the value that was likely to be
         used in the Rome trials. This also meant that there was a longer section of straight track
         for generating the various manoeuvres.

         This was the first flight of the VDL mode 4 and the primary objective was to assess the
         range and reliability of this data link. The designed route, QNQ1, remained within
         150nm of the Boscombe Down base station. This maximum route range had been
         decided so as to allow suitable signal strength throughout the trial. It had also been noted
         that if the transponder link was lost for a period of 2 minutes or more an MFMS CMU
         reset would occur which has the side effect of crashing the MFMS FMU. This was an
         undesirable effect for the flight so the transponder performance was observed using the
         SC-TT T5 viewer software.

3.3.2    MFMS Report

         The pilot initialised the FMS similarly to the previous flight, although on this occasion
         the company route selected was QNQ1 and the cruise altitude was FL240. This was all
         done with the aircraft still at the stand and a trajectory generated for the complete flight
         as far as waypoint DM002, where the aircraft would capture the final approach path. The
         company route was based on the aircraft departing from runway 23 at Boscombe Down,
         but, for this flight, the winds were such that the aircraft had to depart from runway 05. To
         accommodate this, after take off, the aircraft was flown through a dumb-bell turn in order
         to return along the centre-line of runway 23 at about 1200ft QFE. After traversing the
         length of the runway 23, the pilot activated the trajectory in the FMS and engaged the
         lateral FMS guidance demands through the autopilot, followed soon after by the profile
         demands, once the climb had been initiated. This occurred without any problems and
         with a smooth transition by the autopilot to the FMS demands.

         During the climb, the FMS updated the speed demands in accordance with the planned
         speed profile for the flight so that, once established in the cruise, the aircraft was flying at
         260kts CAS. At this point, the pilot selected the Cockpit Display of Traffic Information
         (CDTI) overlay on the lateral map of the Navigation Display (ND) using the track-ball to
         select the soft-keys on the map display. As the aircraft approached the turn at waypoint
         YYY, the simulated traffic was triggered to start automatically within the FMS and the
         symbols representing the location of the surrounding traffic appeared on the map. It was
         noted that, although the fore-point of the triangle symbol, representing the own aircraft
         on the map, indicated the position of the aircraft, it was actually the centres of the
         triangular-shaped symbols representing the other traffic that indicated their positions.
QinetiQ/S&E/AVC/CR031041                                                                      Page 13 of 90
           This might be considered an inconsistency in the implementation, but given the scale of
           the traffic symbols, the variation in the perceived position could be small compared with
           the influence of the update frequency of the ADS-B reports for the traffic.

           Due to the limited length of the straight legs that could be incorporated within the
           complete route structure in order to remain within the required trials area, the method of
           performing the pass-behind manoeuvres was not entirely as might be expected in a more
           operational environment. Primarily, rather than a conflict being identified and resolved
           when two aircraft were on converging legs, the pilot had to enter the relevant data for the
           manoeuvre via the MCDU on the preceding leg. This meant that a trajectory could be
           generated as soon as possible after the completion of the turn on to the leg on which the
           conflict would occur and thus there would be sufficient airspace available to incorporate
           the lateral route adjustment.

           For the first pass-behind, the pilot selected the target aircraft, SIM00, and entered a
           required minimum spacing distance of 5.5nm. When computing the pass-behind
           manoeuvre, an additional factor of 0.25nm was applied by the FMS to the minimum
           spacing value in order to provide a performance buffer during the execution of the
           manoeuvre. The path of the other aircraft would currently result in the spacing distance
           of the two aircraft being compromised while the BAC 1-11 was on the leg between
           waypoints EEE and FFF. On this leg, the aircraft SIM00 was predicted to cross about
           4.9nm ahead of the BAC 1-11 on a track that was approximately 83° to that of the 1-11,
           passing from right to left. This intercept point between the two tracks of the aircraft was
           15.7nm ahead of the BAC 1-11's current position when the pilot triggered the prediction.
           The trajectory generated by the FMS for the BAC 1-11 to pass-behind SIM00 resulted in
           the BAC 1-11 deviating about 4.4nm to the right of its original track to waypoint FFF.
           The extra route length created by this manoeuvre was of the order of 1nm. The tracks of
           the two aircraft are shown in Figure 3.3-1, the track of the BAC 1-11 being displayed in
           blue and that of the target aircraft, SIM00, in red. The BAC 1-11 was initially on a south-
           westerly track while the target aircraft, SIM00, was flying on a south-easterly track.




 Figure 3.3-1 Tracks for BAC 1-11 and SIM00          Figure 3.3-2 Spacing of BAC 1-11 from SIM00


           Following activation, the aircraft transitioned smoothly on to this new trajectory and
           accurately tracked the lateral path for this manoeuvre. The initial change in track angle
Page 14 of 90                                                              QinetiQ/S&E/AVC/CR031041
         for the manoeuvre was 16.5°. The speed demand remained unchanged at 260 knots CAS
         and the height demand continued at the cruise altitude of FL240. During the course of the
         manoeuvre, the minimum spacing distance that was encountered between the BAC 1-11
         and the simulated target aircraft was actually about 5.6nm (see Figure 3.3-2). The
         manoeuvre was regarded as complete once the BAC 1-11 had performed the turn at
         waypoint FFF to return to the original route on the leg to waypoint GGG.

         On this next leg, the pilot cancelled the selection of the target SIM00 and entered the
         identity of the target aircraft for the second pass-behind manoeuvre, SIM02. The
         minimum spacing distance was again set to 5.5nm (5.75nm including the performance
         buffer) and the resume waypoint was entered as HHH. The track of this aircraft would
         intercept that of the BAC 1-11 on the leg between waypoints GGG and HHH. On this
         occasion, the target aircraft was flying on a track that gave a much shallower intercept
         angle of 40°, crossing from left to right, than had been the case for the first pass-behind
         situation. The simulated aircraft, SIM02, was predicted to reach the intercept point 3.6nm
         ahead of the BAC 1-11. At the point on the route where the pass-behind trajectory was
         predicted, the BAC 1-11 was 24.3nm away from this intercept point. With the simulated
         aircraft flying a track that was closer to that of the BAC 1-11, this configuration was used
         to investigate whether the resultant track change required for the pass-behind would be
         greater than that encountered with the first example. As it was, the generated trajectory
         deviated the BAC 1-11 by about 4.5nm to the left of its original track and the initial track
         change was of the order of 18° (see Figure 3.3-3). The geometry of the pass-behind
         manoeuvre was therefore reasonably similar to that produced when the target aircraft had
         been flying almost at right-angles to the track of the BAC 1-11, although in this case, the
         predicted intersection in the original tracks was almost 10nm further away. The extra
         route distance created by this pass-behind manoeuvre was only about 1.2nm. As before,
         the aircraft flew this lateral manoeuvre with no problems being encountered and the
         speed and altitude demands remaining unchanged. The minimum spacing distance during
         the manoeuvre was 6.1nm compared to the defined value of 5.5nm (see Figure 3.3-4).




 Figure 3.3-3 Tracks of BAC 1-11 and SIM02         Figure 3.3-4 Spacing of BAC 1-11 from SIM02
         The system was then configured for a third pass-behind manoeuvre, this time to occur on
         the leg between waypoints III and JJJ. The selected target aircraft was SIM03 but this
         time the minimum spacing distance was defined as 8nm. This was due to the target
         aircraft running earlier than had been originally planned (this was partly affected by the
QinetiQ/S&E/AVC/CR031041                                                                  Page 15 of 90
           delay to the BAC 1-11 caused by flying the extra 2nm created by the previous two pass-
           behind manoeuvres). The track of SIM03 would result in it crossing that of the BAC 1-11
           at an angle of 92° from left to right, the angular intercept being similar to that with
           SIM00, but this time with the target aircraft heading very slightly towards the BAC 1-11.
           With this current track geometry, the simulated aircraft would reach this intercept point
           6nm ahead of the BAC 1-11. The trajectory for the pass-behind manoeuvre was
           generated when the BAC 1-11 was still 24.9nm from this intercept in the current tracks.
           The resultant trajectory required the BAC 1-11 to deviate by 5nm to the left of its
           original track to waypoint JJJ, requiring an initial track change of 18º and an extension of
           2.2nm to the previous route distance (see Figure 3.3-5). The minimum spacing from
           SIM03 reached 7.7nm during the manoeuvre, consequently this was 0.3nm inside the
           required minimum value of 8nm (see Figure 3.3-6).




  Figure 3.3-5 Tracks of BAC 1-11 and SIM03          Figure 3.3-6 Spacing of BAC 1-11 from SIM03


           During these manoeuvres, the variations from the intended lateral spacing, which equated
           to the defined minimum spacing distance plus the performance buffer, can be explained
           to a certain extent by the effects of the prevailing wind. The version of MA-AFAS that
           was flown assumed zero wind conditions in its prediction, but the actual wind was of the
           order of 45-50 knots from a direction of about 150º. This resulted in variations in the
           ground speed compared with the predicted value within the trajectory. For instance, for
           the pass-behind of SIM03, the BAC 1-11 encountered a significant direct tail wind
           component and therefore the spacing distance achieved a value below the required
           minimum.

           Towards the end of the manoeuvre to pass-behind SIM03, as the BAC 1-11 was
           approaching waypoint JJJ, a problem arose with the position data being output by the
           SBAS equipment. Just prior to this waypoint the SBAS receiver lost lock on the
           geostationary augmentation satellite, which provided not only the differential correction
           messages but also an additional pseudorange source. Upon reacquisition, the receiver
           incorrectly resolved the CA code ambiguity resulting in a 600km error in the
           measurement of this pseudorange to the geostationary satellite. This pseudorange was
           included in the position solution and, possibly due to the nature of the implementation of
           the software, the error was not detected causing the solution to jump by a number of
Page 16 of 90                                                               QinetiQ/S&E/AVC/CR031041
         miles. This position jump was influential on the behaviour of the FMS, which saw a step
         in the aircraft's position of just under 7nm on a bearing of 168º True, the aircraft
         currently flying a track of 54º True. Ground speed values reported by the SBAS also
         showed extremes of 1700kts. With this position data giving a cross-track deviation of
         about 5nm, the FMS demanded the maximum allowed left bank of 25º. At this point the
         FMS was disengaged and the aircraft manually flown to regain the trajectory. Once the
         primary position source had been changed, via the MCDU, to the inertial reference
         system, a new trajectory was generated direct to waypoint ZZZ and, with sensible
         demands now being produced, LNAV and PRFL were re-engaged through autopilot. The
         error with the SBAS was tracked down and the receiver reset and upon restarting normal
         operation was resumed.

         Due to the deviation from the route caused by this problem, it was not possible to
         perform the final pass-behind manoeuvre on the leg between waypoints JJJ and ZZZ.
         However, it had been noted that the performance of the VDL4 data link to the ground
         station at Boscombe Down had not been very consistent when the aircraft had been
         flying in the proximity of the furthest west point on the route (waypoint III). Therefore it
         was decided to fly a further circuit to the west of the airway A25 in order to investigate
         further the performance of the VDL4 data link. At the same time, the pilot inserted the
         additional waypoints GGG, HHH and JJJ into the FMS route between ZZZ and KKK on
         the LEGS page of the MCDU. A new trajectory was generated and activated to provide
         the routing for this extra circuit. The guidance demands were not permanently engaged
         through the autopilot, though, because of a requirement on occasion to deviate the
         aircraft either side of the track in order to check on the effects on the data link
         performance.

         The FMS demands were re-engaged, however, for the return to Boscombe Down and the
         FMS demanded a deceleration to 250 knots just prior to triggering a descent to FL75
         during the turn at waypoint CCC. The FMS was still computing too shallow a descent
         and therefore to overcome this problem, a further trajectory generation was carried out
         while the aircraft was in the descent, just after completing the turn at waypoint AAA.
         This now correctly resulted in a descent to FL40 for the STAR, the FMS demanding an
         additional speed change to 230 knots CAS on reaching this altitude. In order to prepare
         for the approach, the Profile demands were disengaged from the autopilot as the aircraft
         started the turn at waypoint DM001 and then the lateral demands were disengaged when
         the aircraft was about 1nm from the end of the route at DM002.

3.3.3    VDL4 Report

         The flight trials were performed to exercise the MA-AFAS ASAS manoeuvre
         capabilities during which dynamic Aircraft Position Reports using ADS-B were
         generated and received by a ground station installed on the QinetiQ Boscombe Down
         airfield.

         For this first flight there was no Point to Point traffic present on the link, just the position
         broadcasts from the air and ground transponders. The reporting rate of the ground station
         had been set to provide uplink reports at specific intervals in the repeated sequence of


QinetiQ/S&E/AVC/CR031041                                                                      Page 17 of 90
           16s, 12s and 32s. The reporting rate of the airborne transponder had been set to provide
           downlink broadcast reports at regular 4s intervals.

           The online assessment during the flight was made very difficult in that the only available
           data was presented on a laptop PC screen within two windows. The first, a scrolling
           screen of position reports, from the local transponder, at 1s intervals interspersed with the
           remote position report. The second window consisted of a two-line timestamped message
           representing the latest local and remote message data. It was noticed on the airborne
           viewer that aircraft speed was reported as being a factor of 10 too great. This was the
           opposite on the ground side as a factor of 10 too small.

           Several events were noticed during the flight where the uplink data was lost for periods
           approaching 60s. One such event lost the link for 60s when at 32nm during the initial
           climb. A further significant event took place as the aircraft exceeded 135nm from the
           base station when the uplink messages were lost for a period of greater than 2 minutes. A
           further 3 uplinks were observed over a period whilst the aircraft range from the base
           station was varied between 143nm and 68nm, the aircraft was manoeuvred to allow for
           any antenna interference and masking characteristics and both aircraft and ground
           transponders were reset. None of these actions had any effect until the aircraft reduced its
           range to within 50nm of the base station, when the link was re-established.

           Observations of the aircraft at the base station remained constant until the aircraft’s range
           exceeded 135nm when the downlinks became unreliable. However, the down link was
           not lost in the same manner as the up link and returned to normal performance when the
           aircraft range returned to within 135nm. It was, however, noticed that the base station log
           file contained a significant number of reports where the system clock was reported as
           being 00:00:58. This would indicate that the base station was losing lock on the satellites,
           from which the time is derived.

           As a consequence of this flight the MFMS route was re-designed to remain within 135nm
           of Boscombe Down airfield to allow a reliable communication link to be maintained.
           Further investigation was also required to determine the cause of the apparent reception
           range anomaly.




Page 18 of 90                                                                QinetiQ/S&E/AVC/CR031041
3.4      27th February 2003

3.4.1    Sortie Objectives

         This flight was planned as a repeat of the flight carried out on the 19th February, but
         using the revised version of the QNQ1 route that kept the aircraft within 135nm of the
         VDL4 ground station at Boscombe Down (see Appendix A.2). Once again, four pass
         behind manoeuvres were to be tested and the range and integrity of the VDL4 air/ground
         link evaluated. In addition, a merge behind manoeuvre was to be flown for the first time.

3.4.2    MFMS Report

         The modified company route QNQ1 had been inserted into the FMS in order to keep the
         aircraft within 135nm of Boscombe Down and therefore improve the coverage
         performance of the VDL4 data link. The pilot initialised the system in the same way as
         on the previous flights. On this occasion, however, the trajectory was activated once the
         aircraft was lined-up on the runway. At this stage, the FMS would detect that the aircraft
         was still on the ground and would only transmit guidance demands to the autopilot once
         it had detected the aircraft was airborne and climbing away from the airfield. Upon
         activation, the trajectory line on the lateral map display changed colour from light blue to
         maroon in order to indicate that the FMS had acknowledged the activation of the
         trajectory. Due the aircraft's very close proximity to the start point of the trajectory, the
         FMS was not able to consistently determine the correct 'Next Waypoint' information until
         the take off roll had begun and the aircraft had moved a few hundred metres along the
         runway. This problem was exacerbated by the fact that Boscombe Down was not only
         the departure but also the arrival airport and therefore the aircraft could also be
         determined to be at the end of the route as well as the start.

         Although the curved precision departure route was being flown for the first time, the
         primary navigation source was still set to be SBAS while data was gathered to confirm
         the GBAS performance over the course of this departure route. The lateral navigation
         demands were engaged through the autopilot as soon as possible prior to the start of the
         first turn, 2nm beyond the end of the runway. Profile guidance was also engaged as the
         aircraft entered this turn. No problems were encountered with the lateral guidance
         throughout the departure route. However, since the FMS was only generating continuous
         climbs, the aircraft had already reached about FL100 as it passed waypoint KATE and
         this placed it in close proximity to an area of controlled airspace to the south of
         Boscombe Down for which the lower altitude limit was FL95. The requirement was for
         the FMS to handle height constraints during the climb, especially for the SID procedures,
         but this was not available in this current release. Subsequent releases would provide this
         capability and would thus allow the originally defined limit of not being above FL50 at
         KATE to be met and consequently would remove any problem with the neighbouring
         controlled airspace.




QinetiQ/S&E/AVC/CR031041                                                                   Page 19 of 90
  Figure 3.4-1 GBAS Position Reports during            Figure 3.4-2 Accuracy of GBAS 3D Position
             Precision Departure                                 Data during Departure
           The performance of the GBAS while the aircraft was flying the SID was found to be
           extremely consistent for the vast majority of the route up to and beyond KATE, the end
           of SID waypoint (see Figure 3.4-1). There was a brief period of only about 5 seconds
           when the GBAS came out of differential mode. Figure 3.4-2 shows that, compared to the
           truth track data, the deviation in the GBAS information was no more than about 1m
           laterally and 3m vertically while in differential mode and the respective values when in
           stand-alone mode were 4m and 20m. This demonstrated that, for use as a data source for
           the FMS lateral guidance during the precision departure, the GBAS could easily provide
           the accuracy required in the aircraft position fix.

           The climb continued to the cruise at FL240 without any further problems. Rig testing had
           shown, however, that, with this version of software, while in the cruise, wherever the
           outbound route crossed over the inbound route, the FMS would demand a descent (the
           inbound route being part of the descent at these cross-over points). For the initial circuit
           of the route to the east of airway A25, the FMS Profile guidance was therefore
           disengaged from the autopilot in the cruise.

           The first pass-behind manoeuvre was then carried out against the simulated target
           aircraft, SIM00. For this and all subsequent flights, the standard minimum spacing
           distance that was to be used was now 6nm, but still with an additional 0.25nm
           performance buffer applied. This was to allow some flexibility from the minimum
           separation of 5nm permitted by ATC during the trials in Rome when these manoeuvres
           would be flown against a real aircraft. The geometry of this conflict situation was
           essentially identical to that experienced on the flight on the 19th February, although this
           time the BAC 1-11 was 9.8nm from the predicted intercept point when the pass-behind
           trajectory was generated. Although the minimum spacing distance had been increased by
           0.5nm, the deviation of the BAC 1-11 from its original track to waypoint FFF was
           5.5nm, about 1nm greater than that on the previous flight, and the initial track change
           was now 23.5º. This resulted in an extension of 1.8nm to the to tal route length. The
           actual minimum spacing from SIM00 achieved during the manoeuvre was 5.8nm. The
           FMS was still not utilising forecast or actual wind information in its prediction and
           consequently this could influence the spacing at the closest point of approach. However,
           the overall execution of this manoeuvre was successfully accomplished.
Page 20 of 90                                                               QinetiQ/S&E/AVC/CR031041
         With the profile guidance re-engaged once the aircraft was clear of the airway A25, the
         data for the next pass-behind manoeuvre was entered by the pilot. The simulated aircraft,
         SIM02, was flying an identical track to that on the previous flight hence the geometry of
         the conflict remained similar, although significantly, the target aircraft would now cross
         the track of the BAC 1-11 about 4.5nm ahead of the BAC 1-11 rather than 3.6nm. This
         was due to the extra 1nm required for the first pass-behind manoeuvre. A trajectory was
         generated for the pass-behind manoeuvre for a 6nm spacing, this occurring when the
         BAC 1-11 was 23.9nm from the predicted intercept between the two aircraft's tracks. The
         offset distance from the original track for this manoeuvre was 4.1nm with an initial track
         change of 17º. With this extended spacing distance, but the target aircraft now slightly
         further ahead of the BAC 1-11, the additional route length amounted to 1.3nm, which
         was comparable to the change on the previous flight.

         Activation of this pass-behind manoeuvre was successful and the aircraft started to
         follow the required track. However, a problem then occurred within the FMS itself with
         regard to the SBAS position information that it was reading from the Arinc data bus. The
         navigation data being used by the FMS froze at 1452:37 (UTC), soon after the aircraft
         had completed the initial turn off from the original track to start the pass-behind
         manoeuvre. Although the navigation data, including time, was no longer changing
         internally to FMS, other data was updating correctly and it was not immediately apparent
         that a problem had occurred. However, it was noted that the aircraft was not appearing to
         progress along the route on the map display relative to the next waypoint. The impression
         of movement was still given, though, by the progress of the target aircraft symbol across
         the map and therefore the sense that the own aircraft was no longer moving was not so
         apparent with this display configuration. Once it had been recognised that the navigation
         data was not changing, the primary navigation source was altered to be the inertial
         reference system. The aircraft was now beyond the start of the next turn, so it was
         manually flown back on to the correct track while the FMS problem was investigated. It
         was determined that the SBAS was still functioning correctly and later it was confirmed
         that the data was also updating correctly on the Arinc channel from the SBAS equipment.
         Therefore the problem was internal to the FMS, either the Arinc card had developed a
         fault reading the data from this channel or the problem existed in the access of this data
         between the Arinc card and the main FMS processor. Figure 3.4-3 shows how the time
         reference became frozen within the FMS along with the position data until the position
         data source was changed.




QinetiQ/S&E/AVC/CR031041                                                                Page 21 of 142
                  Figure 3.4-3 Freeze of SBAS Data within the FMS, 27th Feb. 03
           Once the aircraft was back on track, a new trajectory was generated and activated, but by
           now the simulated target aircraft, SIM03, for the next pass-behind manoeuvre was
           beyond the point where a sensible trajectory could be generated and therefore this was
           omitted. Consequently the system was configured for a pass-behind the simulated traffic,
           SIM04. This aircraft would cross the westerly track of the BAC 1-11 on the leg between
           waypoints JJJ and ZZZ, the track of the target aircraft being 118º to that of the BAC 1-11
           and passing from right to left. This geometry meant that the conflicting traffic had a
           reasonable component of track that was towards the BAC 1-11. This traffic was predicted
           to reach the intercept point between the two aircraft about 12.5nm ahead of the BAC 1-
           11, but with the aircraft converging towards one another, the spacing distance would still
           be reducing after the target aircraft had passed this point. Hence, a trajectory was
           generated with a minimum spacing distance set at 8nm with the BAC 1-11 only having to
           deviate by 2.7nm to the right of its original track, the generation taking place when the
           BAC 1-11 was 20nm from the original intercept point (see Figure 3.4-4). This small
           deviation in the track resulted in less than 0.5nm increase in the route length with an
           initial track change of 9.5º. The pass-behind manoeuvre was successfully flown and the
           minimum spacing that occurred between the two aircraft was approximately 8.2nm
           (comparable to the 8.25nm used by the FMS in its manoeuvre computation).




Page 22 of 90                                                             QinetiQ/S&E/AVC/CR031041
 Figure 3.4-4 Tracks of BAC 1-11 and SIM04         Figure 3.4-5 Spacing of BAC 1-11 from SIM04
         For this flight, the FMS was also tested with regard to a merge behind manoeuvre. In this
         case, the ATC controller would define the distance that the own aircraft is required to
         attain behind another aircraft at a set waypoint. At this waypoint, both aircraft should be
         flying a common lateral path and the own aircraft would maintain the required spacing
         from this point onwards, either to a defined end point or until the controller issues a
         change in instruction. The current version of the software did not perform this latter
         function, however, so the manoeuvre would be complete when the BAC 1-11 had
         reached the merge waypoint. Due to the problem in ensuring that the simulated traffic is
         in the right place for the initiation of this merge manoeuvre, a stream of five aircraft had
         been pre-defined to follow the same route, 10nm apart. These aircraft were set on a track
         from waypoint ZZZ to RRR (the merge waypoint) and then continued on the same path
         as the BAC 1-11 to SSS.

         The selected target aircraft was SIM06 and the spacing distance entered was 6nm. As
         already noted, the waypoint at which the merge was to be complete was RRR. As for the
         pass-behinds, the pilot entered this data via the ASAS pages on the MCDU. A trajectory
         was generated once the aircraft was on the leg to waypoint QQQ, effectively flying a
         parallel track at that moment to the target aircraft, SIM06. Although the new trajectory
         now took the BAC 1-11 on a direct track to waypoint RRR, there was no updated speed
         demand computed by the FMS, so the demand to the autopilot remained at 260kts CAS.
         Hence the spacing distance had not been achieved by the time the BAC 1-11 reached
         RRR.

         The primary navigation source, which had been set to the IRS since the SBAS data
         problem, was now changed to be GBAS for the remainder of the flight. A further
         trajectory was generated and activated when the BAC 1-11 was on the leg to waypoint
         TTT in order to correct the next waypoint information that was being displayed on the
         ND and the MCDU. Both the lateral and profile demands were disengaged from the
         autopilot on reaching waypoint DM001 within the simulated STAR at Boscombe Down.

         The taxi map was used after landing and provided a reasonable display of the aircraft's
         position on the airfield as it taxied back to the stand. The use of white to block in the
         taxiways as well as for the aircraft symbol did cause a problem in being able to locate the
         aircraft on the map. Although the position did give an element of situation awareness to
QinetiQ/S&E/AVC/CR031041                                                                  Page 23 of 90
           the pilot of the aircraft's location, with the weather being very clear and the pilot's
           extreme familiarity with the airfield, there is little that can be determined about the
           effectiveness of the display.


3.4.3      VDL4 Report

           In the first flight of VDL mode 4, the link had been observed to lose contact with the
           Boscombe Down ground station at an approximate range of 135nm. This loss of link was
           primarily in the uplink direction and was re-established at approximately 50nm range. As
           no explanation of the reception range difference could be discovered in the time
           available, a second route was devised whose maximum range would not exceed 135nm
           from the Base Station.


                     th
                                             Flight trajectory
                    4 Pass Behind
                     manoeuvre                                                                  51.6
                                                                 Boscombe Down

                                                                                                51.4


                                                                                                51.2

                                                                                       Curved Departure
                                                                                                 51


                                                                                                50.8
                                                                        st
                                                                       1 Pass Behind
     rd
   3 Pass Behind                                                        manoeuvre
                                                                                                50.6
    manoeuvre
                                              nd
                                             2 Pass Behind                                      50.4
                                              manoeuvre

                                                                                                50.2
           -6             -5          -4            -3            -2              -1                   0




Figure 3.4-6 ADS-B Flight Data Received by EEC
           The reporting rate of the ground station was set to provide uplink reports at specific
           intervals in the repeated sequence of 16s, 12s and 32s. The aircraft transponder reports
           were set to 4s intervals. Data recordings were made using the T5 viewer software at both
           ends of the link. The ground system was also configured to send the ADS-B data to EEC
           Bretigny. A plot of the data received at EEC is shown in Figure 3.4-6. After take-off the
           outbound performance was observed to be comparative with previous experience with
           uplink data being received at intervals not exceeding 60 seconds. Beyond 100nm range
           from the ground station gaps in the uplink data exceeding 1 minute became common.
           The first 2 minute gap in data was observed at a range of 135nm whilst in the turn at
           waypoint H. The link remained intermittent along the leg H – M - J but recovered after
           waypoint J at a range of 124nm. The link remained consistent with the first loop all the

Page 24 of 90                                                                QinetiQ/S&E/AVC/CR031041
         way around the second loop only exhibiting reliability when the range became less than
         100nm.

         The VDL4 performance was judged as being acceptable for ADS broadcasts. No
         experience has been gained yet on requirements for the Point to Point capability.

         The ground VHF voice link was not maintained to 135nm range and if needed as a data
         link backup and technical link will need further attention.




QinetiQ/S&E/AVC/CR031041                                                             Page 25 of 90
3.5        11th March 2003

3.5.1      Sortie Objectives

           This flight was to check the range and performance of the modified SAAB VDL4
           transponder, using part of the modified QNQ1 route (see Appendix A.2), then followed
           by a range check to 200nm. The first part of this flight consisted of a climb to FL240 in
           the region of VLN to join the route at waypoint F for an airway crossing to G, then to H,
           M, J, Z. This was to confirm that the revised unit will work at the shortened range of
           135nm. If this was satisfactory, it was then required to check that the VDL4 would work
           within the full box size of 200nm that had initially been allocated for the trials at Rome.

           Two hand flown 4.5º approaches from 2500ft using the GBAS guidance were also to be
           included at the end of the flight.

           The MFMS was not used for this flight.

3.5.2      VDL4 Report

           SC-TT have analysed the flight data from the 19th Feb. flight and advised that the
           transponders should be changed. SC-TT supplied 2 new transponders, both engineering
           development models, with revised hardware for improved frequency stability. The
           transponders replaced both the air and the ground systems. The estimated, but un-proved,
           range of the new transponders is 150nm. This flight was conducted to verify the reliable
           transponder range to allow a route to be designed with its coverage area.

           The transponder data was monitored using the SC-TT supplied T5viewer software.

           Pre-flight the ground transponder was configured and observed to be transmitting in a
           sequence of 4s, 4s, 8s, 4s, 4s, 8s … The airborne transponder was reporting position at
           regular 2s intervals.

           After take-off the T5 recorder was started. Ground speed was observed as being a factor
           of 10 too great, as per previous flights, both track and distance to the base-station agreed
           with the IRS and Trimble GPS systems onboard the aircraft.

           The aircraft flew around the new short route with no significant loss of data. The aircraft
           continued on a modified second loop which extended west until the air/ground link was
           lost. A plan had been prepared which allowed for the aircraft to extend to 200nm from
           Boscombe Down. However the link was lost at a base station range of 143nm. The
           aircraft continued on for a further 22nm before turning back to test where the link would
           be re-gained. Re-establishment of the link occurred at 132nm initially with sporadic
           reports being received. A regular ground report was not recovered until the aircraft came
           within 122nm from base. The aircraft returned to its standard routing and returned to
           Boscombe Down.

           On the basis of this flight a decision was made to continue with the new shortened route
           as this would enable aircraft operation with the best air/ground link available.


Page 26 of 90                                                               QinetiQ/S&E/AVC/CR031041
3.6      14th March 2003

3.6.1    Sortie Objectives

         This flight was based on the same company route, QNQ1, as flown on the 27th February
         (see Chapter 3.4.1). One significant change from the previous flights was that the FMS
         now utilised forecast meteorological data in its trajectory predictions. This data was in
         the form of the forecast air temperature, wind speed and wind direction for various
         pressure levels at different grid points covering the area traversed by the route. This
         should have provided a more accurate time profile for the aircraft’s predicted trajectory
         which would be important for merge manoeuvre in which the FMS aimed to meet a time
         constraint at the merge waypoint in order to be at the necessary spacing behind the other
         aircraft.

         Additional improvements since the last flight included the handling of height constraints
         within the SID, allowing the precision departure routing to be flown properly and not
         infringe the neighbouring controlled airspace. The problem encountered on the previous
         flight relating to the crossover points between the inbound and outbound routes, and
         resulting in premature demands for the descent being sent to the autopilot, had been
         fixed. Finally, the descent prediction had been improved to produce a steeper descent
         profile, more comparable to that flown by the BAC 1-11.

3.6.2    MFMS Report

         The prevailing winds at Boscombe Down meant that the BAC 1-11 had to depart again
         from runway 05 rather than runway 23 for which the precision departure route was
         configured. Therefore, once again a dumb-bell turn was performed directly after take-off
         so that the aircraft could be flown at 500ft above runway 23 in order to simulate a take
         off from this runway. Activation of the trajectory was performed by the pilot as the
         aircraft passed the far threshold of runway 23 allowing the FMS lateral demands to be
         engaged through the autopilot as soon as possible. Since the aircraft had already been
         cleaned up, the FMS profile demands could also be engaged soon after trajectory
         activation. The SID was flown without any problems, the FMS initially demanding a
         climb to FL50 for the duration of the SID until demanding a climb to the cruise at FL240.
         The FMS sent this demand for the climb about 1.5nm before waypoint KATE providing
         sufficient anticipation time for the autopilot to make the transition from level flight to the
         climb. Throughout the SID, the FMS was using the GBAS as its primary position source,
         however, once passed KATE, the SBAS was selected as the source for the en-route
         section of the flight. Transitions between the data from these two sources caused no
         discontinuity in the FMS guidance performance.

         It was noted that upon activation of this trajectory, all the waypoint ETA values on the
         ND and the MCDU were being shown as 0314:00. This had been seen before during the
         testing with the aircraft model rig, but the exact cause of this had not been fully
         identified. Although this did not cause any problem with the FMS’s ability to generate
         the guidance demands for the autopilot in order to follow the trajectory, it was decided to
         perform an in-flight generation and activation in case the erroneous times influenced the



QinetiQ/S&E/AVC/CR031041                                                                    Page 27 of 90
           generation of the pass-behind manoeuvres. This was done once the aircraft was in cruise
           on the leg to waypoint DDD and now all the times were correctly set.

           As normal, the first pass-behind was set up with SIM00 selected as the target aircraft and
           a minimum spacing distance of 6nm entered. The generation of this trajectory failed,
           however, with the FMS reporting a problem trying to meet the time constraints that its
           manoeuvre generator function had applied to the various waypoints that it had defined
           for the pass-behind manoeuvre. The reason for the system applying time constraints to
           these inserted lateral waypoints was to try and ensure that the spacing of the two aircraft
           could be determined in relation not only to position but to time as well. The spacing
           distance was changed to 8nm, but a trajectory could still not be computed. Finally, 5nm
           was tried and this time the generation was successful. The tracks of the two aircraft prior
           to the activation of this pass-behind manoeuvre had been the same as on the previous
           flights, although the problem in generating a trajectory meant that the BAC 1-11 was
           12.2nm from the intercept point when a valid trajectory was eventually predicted. The
           target SIM00 had been predicted to pass 4.9nm ahead of the BAC 1-11. On this occasion,
           the resultant track change for the BAC 1-11 was 17° and its maximum deviation from its
           original track was 2nm. The speed demand for the first part of manoeuvre up to the
           estimated point of closest approach was 220kts CAS, while beyond this point, it reverted
           back to 250kts CAS. This change in speed was a result of the time constraints inserted by
           the manoeuvre generator and the application of the forecast meteorological data,
           resulting in variations in ground speed dependent on the air speed and track angle. It later
           transpired that there was a problem with the forecast meteorological data due to an extra
           control character having appeared at the end of each line in the file during the transfer
           process. This affected the way the MFMS interpreted the data from the file and caused a
           greater variation in the forecast values used by the trajectory prediction function than was
           actually the case. During the execution of this pass-behind manoeuvre, the actual spacing
           from SIM00 at the closest point of approach was 5.4nm.

           The second pass-behind manoeuvre, using the target aircraft SIM02, was unchanged in
           terms of the conflict geometry from the previous flight, the target being predicted to
           cross 3nm ahead of the BAC 1-11. On this occasion, a trajectory was successfully
           generated with an assigned minimum spacing distance of 6nm, resulting in the BAC 1-11
           deviating by 5.7nm from its original track with an initial track change of 28.5°. Once
           again, on trajectory activation the speed demand to the autopilot was reduced to 220kts
           CAS until passing the closest point of approach where the demand went back 250kts
           CAS. This time, the minimum spacing that was achieved was 7.6nm, compared to the
           permitted value of 6nm. This was because the ground speed predicted in the trajectory
           was over 50kts greater than that actually achieved for the demanded air speed. Although
           testing had shown that the FMS was correctly applying the forecast wind information to
           predict the effect on the aircraft's ground speed, it was later determined that there was
           problem in the derivation of true air speed by the FMS. This only existed when the FMS
           was using the forecast meteorological data and although a fix was subsequently
           produced, it was too late to implement and test prior to the aircraft departing for the trials
           in Rome. Hence this problem had an influence on the results of the various pass-behind
           and merge manoeuvres, but not to such a degree that it significantly compromised the
           manoeuvres that were being flown. This particular pass-behind manoeuvre was
           completed without problems and without any transients affecting the guidance
           performance.
Page 28 of 90                                                                QinetiQ/S&E/AVC/CR031041
         The third pass-behind with the simulated target SIM03 was also successfully
         accomplished. This was the first time that this particular manoeuvre had been flown since
         it had not been possible on the last flight due to the navigation data problems and the
         route had been updated since the flight on the 19th February. The geometry for the two
         aircraft had the target aircraft, SIM03, following a track 113° relative to that of the BAC
         1-11, passing from left to right and thus SIM03 had a small component of its track being
         towards the BAC1-11 (similar to SIM04). When the pass-behind manoeuvre was
         generated, SIM03 was predicted to cross the track of the BAC 1-11 5.7nm ahead. At the
         point of generation, the BAC 1-11 was still 18nm from this intercept point. The new
         trajectory was based on a defined minimum spacing distance of 6nm and resulted in the
         BAC 1-11 deviating 3.3nm to the left of its original track, the initial track change being
         15°. Figure 3.6-1 shows the tracks of the two aircraft, The BAC 1-11 (blue line) having
         been on a northerly heading and SIM03 (red line) on a south-easterly heading. At the
         closest point of approach (see Figure 3.6-2), the BAC 1-11 was 6.5nm from SIM03, the
         CAS demand having this time reduced to 224kts and the actual ground speed being about
         25kts less than that determined by the prediction for this air speed.




Figure 3.6-1 Tracks of BAC 1-11 and SIM03               Figure 3.6-2 Spacing of BAC 1-11 from
                                                                        SIM03
         The final pass-behind manoeuvre took place on the leg between waypoints JJJ and ZZZ.
         This configuration was as per the previous flight with the target aircraft, SIM04,
         predicted to cross the track of the BAC 1-11 about 3.1nm ahead of it. The minimum
         spacing distance was again set as 6nm and the pass-behind manoeuvre was successfully
         generated when the BAC 1-11 was 20.4nm from the predicted intercept point between
         the two tracks. The manoeuvre deviated the BAC 1-11 by 5nm to the right of its original
         track and required an initial track change of 19°. As with the other pass-behinds on this
         flight, there was a change in the speed demand from the FMS to 220kts until past the
         closest point of approach. Consequently, the actual minimum spacing distance that
         occurred during the manoeuvre was 6.8nm due to the predicted ground speed being
         greater than that achieved for the demanded air speed.

         Having completed the last pass-behind, the BAC 1-11 now followed the additional loop
         section that had been inserted for testing the merge behind manoeuvre. When the aircraft
         was on the leg to waypoint PPP, the pilot entered the selected target aircraft identity
         SIM05 into the MCDU along with the merge waypoint name RRR. A spacing distance of
         6nm was used again and once the BAC 1-11 was on track to waypoint QQQ, a trajectory
QinetiQ/S&E/AVC/CR031041                                                                 Page 29 of 90
           was predicted. At this point the target aircraft SIM05 was 19nm from the BAC 1-11 and
           therefore the predicted trajectory needed to close the distance between the two aircraft by
           increasing the speed demand for the BAC 1-11. The algorithm that was implemented for
           the merge with spacing manoeuvre applied an additional tolerance of 0.25nm to the
           spacing value entered via the MCDU. This was to create a buffer region to cater for
           errors that might arise from the prediction and guidance processes and to prevent
           unwanted alerts possibly being generated as the aircraft attempted to maintain the
           required spacing distance.

           On activation of this trajectory, the speed demand did increase to 268kts CAS, which
           resulted in a ground speed of 354kts. The aircraft SIM05 was, however, flying at a
           constant ground speed of 356kts and therefore the BAC 1-11 was only just matching the
           speed of the target without closing the distance. The reason was the same as for the
           previous pass-behind manoeuvres where the predicted ground speed was an over-
           estimation of the actual ground speed that was achieved for the demanded air speed.
           Hence, by the end of the manoeuvre, when the BAC 1-11 reached waypoint RRR, the
           aircraft SIM05 was still approximately 18nm ahead. The other aspect that was noted
           about the manoeuvre was that the transition in the aircraft’s track on to the direct path to
           waypoint RRR was too abrupt for the BAC 1-11 to follow smoothly. The prediction
           assumed a direct track from the aircraft’s current position to the merge waypoint, but
           inserted an S-bend manoeuvre to capture this track. However, with the manoeuvre
           generator configured to use 25° of bank in its lateral computations, it was not possible for
           the BAC 1-11 to transition fast enough from 25° of right bank to 25° of left bank in order
           to maintain the aircraft exactly on the lateral path. The aircraft subsequently drifted wide
           on the second section the S-turn and would then have to make a significant correction in
           order to capture back on to the track towards waypoint RRR. A smoother lateral
           transition in track, similar to that used for the pass-behind manoeuvres would have
           greatly improved this aspect of the performance, but the principal aspect of the merge is
           based on the speed change.

           An additional merge manoeuvre was attempted once the BAC 1-11 had completed its
           turn on to the track towards waypoint TTT. In this case, a stream of five target aircraft,
           10nm apart, were flying on a track from a point just north of waypoint HHH to waypoint
           ZZZ (the next waypoint on the BAC 1-11’s route after TTT). The target aircraft SIM12
           had been selected with the merge waypoint set as ZZZ and a required spacing distance of
           6nm. When the merge manoeuvre was predicted, the target aircraft, SIM12, was over
           22nm away and was also 41.5nm from waypoint ZZZ, while the BAC 1-11 was still
           61nm from ZZZ. Activation of this trajectory resulted in a similar lateral over-correction
           to capture the track to ZZZ as seen on the first merge manoeuvre. The speed demand
           increased to 267kts CAS and this gave a ground speed of 392kts compared to the 356kts
           of SIM12. This was insufficient to make up the necessary distance to achieve a 6nm
           spacing, the actual spacing being 16nm when the BAC 1-11 reached the waypoint ZZZ.

           The top of descent point that was predicted by the trajectory generator was located within
           the airway A25, which the route crossed to return towards Boscombe Down. Therefore,
           the profile guidance was deliberately disengaged while traversing the airway and re-
           engaged in the turn at waypoint CCC. The altitude demand from the FMS was zero,
           however, rather than the expected FL40. Attempts were made in the descent to regenerate
           the trajectory to correct this altitude demand but these were not successful, so the profile
Page 30 of 90                                                               QinetiQ/S&E/AVC/CR031041
         demands were disengaged from the autopilot again as the aircraft passed FL105. Lateral
         guidance was retained until the aircraft was approaching waypoint DM002, at which
         point the pilots configured the aircraft for performing the manual approaches using the
         GBAS guidance signals.

3.6.3    VDL4 Report

         This trial used the new transponders supplied by SC-TT. It was not intended to connect
         the VDL4 into the MFMS but to monitor multiple transponders and record the ground
         speed observed by the laboratory transponder during the flight. This was to be the first
         experience in MA-AFAS of data being received by another aircraft transponder and thus
         would be an indication of the data expected when operating with the ATTAS aircraft.
         Two transponders, the Base Station and a laboratory test transponder, were configured as
         follows:

         •   Old Fire Station transponder was configured as a base station. The antenna was
             mounted on the building roof;

         •   Transponder in Building 414 laboratory was configured as an airborne station but
             would remain stationary. The antenna was positioned on the ground outside the
             laboratory, as roof access was not possible. With the antenna in this position, realistic
             ground-air range was not expected. However, reported ground speed could be
             determined over a significant portion of the flight.

         The transponder data was monitored using the SC-TT supplied T5viewer software.

         During a pre-flight ground test both stationary transponders were observed on the T5
         viewer status screen. All links were maintained during the take-off and initial departure.

         The laboratory transponder link was maintained outbound to a range of 80nm from
         Boscombe, this reduced range was as expected. The base station transponder link was
         maintained throughout the flight to a range of 135nm, with only short outages at
         maximum range as the aircraft was manoeuvred. As soon as the aircraft returned to
         straight and level flight, the air/ground link was automatically re-established. During the
         return legs the laboratory transponder was re-acquired at a range of 45nm. Two passes
         down the runway were used to calibrate the ground speed observed by the laboratory
         transponder.

         As both ground transponders were stationary, a zero ground speed was recorded at all
         times on the airborne status display. However, on the ground transponder, the aircraft
         speed was reported as being 10 times too small, which would cause problems when
         ASAS manoeuvres were performed against the real ATTAS target.

         For the MA-AFAS trials SC-TT had supplied three transponders for use by both QinetiQ
         and DLR. After this flight the QinetiQ base station transponder was required for the
         Rome flight tests so no further VDL4 activity was available for the remaining Boscombe
         tests.



QinetiQ/S&E/AVC/CR031041                                                                   Page 31 of 90
3.7        19th March 2003, Sortie No. 762.

3.7.1      Sortie Objectives

           For the transit to Rome, the MFMS was to be used to provide lateral guidance for the
           majority of the route (company route BOSROM). For the initial and final parts of the
           route, where the aircraft was likely to be vectored by ATC to first join the airways
           system and later to pick up the approach into Ciampino Airport, the aircraft would be
           flown manually. The VDL4 transponder was also connected to the FMS for the first time
           during a flight since it was to be checked that the FMS could successfully communicate
           with the ground stations at Rome. It was also intended to check the range performance of
           the VDL4 data link relative to the ground stations at Rome that would be involved in the
           MA-AFAS trials.

           The flight planned route was:

           EGDM-SAM-MID-XAMAB-BAMES-RBT-PTV-NEV-LERGA-LATAM-KURIR-
           RETNO-KOLON-DIVUL-SODRI-BTA-ELBA-GILIO-MEDAL-GILET-OST-LIRA.

3.7.2      MFMS Report

           A trajectory was generated while the aircraft was still on the ground at Boscombe Down,
           the cruise altitude having been set as FL270. This trajectory was then activated once the
           aircraft had lined up on runway 23.

           The FMS now contained a full navigation data base for the European region so that, if
           changes were required to the route during the flight, these could be applied by editing the
           waypoint list on the LEGS page of the MCDU. Entering a waypoint name into the
           MCDU scratchpad and then pressing the line select key beside the point in the list where
           this waypoint was to be inserted, would accomplish this task. The FMS would extract the
           position data for the waypoint from the navigation data base and, in the case of duplicate
           waypoints existing with the same name, the pilot was presented with the options to select
           the correct one. The pilot could also easily delete waypoints from the current list.

           Additionally, a Go-Direct function was available to the pilot on the MCDU's VNAV
           page. Entering a waypoint name into the appropriate location on this page would result in
           the FMS automatically generating a trajectory direct to this waypoint from the aircraft's
           current position, providing the waypoint existed in the current route. This was the
           primary function used by the pilot during the transit flight as ATC provided occasional
           clearances to waypoints further along the route, by-passing the current route leg.

           Initially the aircraft was vectored by ATC until it was cleared to join the airways and
           proceed along its planned route. It had also been noted that the FMS was missing the data
           source from the Engine Instrumentation Unit (EIU), due to a power supply fault with this
           latter piece of equipment. Once this power fault had been rectified and the aircraft was
           established on its route, a new trajectory was generated direct to waypoint BAMES, a
           few miles after the aircraft had passed Midhurst (MID). The lateral and profile demands
           were engaged through the autopilot and the lateral guidance was maintained for
           remainder of the flight until just prior to waypoint MEDAL as the aircraft levelled at
           3800ft and 230kts CAS. At this point, the aircraft was to be vectored by the Rome
Page 32 of 90                                                              QinetiQ/S&E/AVC/CR031041
         controllers for capturing the approach to Ciampino Airport. The profile guidance had
         been disengaged from the autopilot when the aircraft reached the cruise level FL270 in
         order for the pilots to fly a different speed profile than that predicted by the FMS. It
         would normally have been possible for the pilots to update the cruise speed via the
         MCDU pages, but this facility had been disabled recently while overcoming a separate
         issue affecting the speed profile produced for the merge manoeuvres.

         During the course of the transit flight, the FMS lateral guidance, in conjunction with the
         autopilot response to these demands, had resulted in a mean Flight Technical Error value
         for the cross-track deviation of 110m (approximately 0.05nm), with a standard deviation
         of 91m. This is the measurement of lateral error from the predicted route as determined
         by the FMS itself and therefore is not the overall performance error term since it does not
         account for navigation system performance. The maximum cross-track deviation that
         occurred while using the FMS lateral demands was 815m (about 0.44nm). The position
         data source used throughout the flight by the FMS was the SBAS equipment. It should
         also be noted that the FMS was configured in preparation for the trials sortie in Rome
         and was therefore using 5nm as the default turn radius for all the waypoints. With the
         aircraft flying at 320kts CAS, the correspondingly high ground speed resulted in the FMS
         demanding its maximum permissible bank angle of 25º in order to follow the path of the
         turns and, in some cases, this was not quite sufficient. This was the cause of the
         occasional larger spike in the cross-track deviation while performing a turn. The mean
         value was also influenced by the BAC 1-11 not flying exactly in trim, generally requiring
         about 1º of left bank in order to maintain a constant track. A stable condition was
         therefore achieved when the FMS was demanding this level of bank angle. For the
         ground speed of the aircraft during the cruise, this equated to a cross-track deviation of
         around 80m.

3.7.3    VDL4 Report

         This sequence of three flights on the 19th March was arranged to test the VDL4 and voice
         communications range in the proposed Rome flight trials area. For these flight tests the
         aircraft transponder was required to be re-configured to operate on the Rome VDL4
         approved frequency of 136.95MHz. As this frequency was not approved for use in UK
         and French airspace, the transponder remained switched off until the aircraft was inside
         Italian airspace. This was the first time that the transponder had been used in flight whilst
         connected to the MFMS. The switch on process did not cause any problems to the
         MFMS and after a short start-up delay the BAC 1-11’s own transponder position, shown
         as a traffic symbol, overlaid the current aircraft position on the Navigation Display (ND)
         when “TRAFFIC” was selected on. The BICCA code of ASEQA and current ground
         speed, although 10 times too great, were also displayed next to the traffic symbol,
         together with the altitude offset (in hundreds of feet) from the host aircraft.

         It was noted that the ND would be too cluttered with the own-ship transponder
         information displayed and a request was made to remove this unnecessary extra traffic
         symbol. During the approach to Ciampino airport, further transponders were displayed
         successfully on the ND. These transponders equated to ground test transponders
         identified with BICCA codes of HORIA and CQAIA. At this stage the ground
         transponder installed for MA-AFAS at Ciampino was not observed.

QinetiQ/S&E/AVC/CR031041                                                                   Page 33 of 90
3.8        19th March 2003

3.8.1      Sortie Objectives

           This flight would verify the route that was to be flown the following week for the trials
           with the ATTAS aircraft. This route was referred to as QNQ5 (see Figure 3.8-1) and
           incorporated two pass behind and one merge manoeuvre within the assigned trials area.
           The cruise level for this flight was FL210. On this occasion, the ATTAS aircraft was to
           be simulated within the FMS using the same process as for the flights at Boscombe
           Down. The triggering for the start of the simulation was to be done manually, however,
           when the BAC 1-11 was 4nm before the waypoint LUNAK. The VDL4 transponder
           range would also be checked during the flight to ensure adequate coverage could be
           achieved with the ground station over the entire route.




                Figure 3.8-1: QNQ5 Route used for Test Flight in Rome, 19th March 03

3.8.2      MFMS Report

           With the departure and arrival airports entered as LIRA (Ciampino) on the MCDU and
           the company route QNQ5 selected, a trajectory was generated with the cruise altitude set
           to FL210. The SID that was used was based on the published departure route OST5A for
           runway 15. At this point the route continued to follow the airway L5 to VALMA, at
           which point the aircraft would turn left into the trials area. For the return to Ciampino, a
           STAR had been set up within the supplementary data base of the FMS in order to provide
           a routing from OST to the approach to runway 15. The trial itself would be finished by
Page 34 of 90                                                               QinetiQ/S&E/AVC/CR031041
         the time the aircraft reached OST on the return leg, however this STAR was required
         primarily to complete a trajectory profile that would lead on to the approach path.

         The primary navigation data source was again set to be the SBAS. Having activated the
         trajectory after lining up on the runway, the FMS started to transmit guidance demands
         once the aircraft had climbed through 150ft after take off. The lateral demands were
         engaged through the autopilot for the initial turn, but the profile demands were not
         engaged until the aircraft was established on track towards waypoint OST, this being a
         precaution in case there were any system problems after take off.

         The FMS guided the aircraft along the trajectory and the simulated target aircraft,
         SIM02, was started on its own route when the BAC 1-11 was 4nm before LUNAK in
         order to co-ordinate the timings around the route. This target aircraft flew a fixed route at
         a constant ground speed of 287kts. For the first pass-behind manoeuvre, the target
         identity was entered on to the ASAS page of the MCDU with a minimum spacing
         distance of 6nm and the resume waypoint specified as B3. Once the BAC 1-11 had
         completed the turn at VALMA, a trajectory was generated, the aircraft SIM02 being
         predicted to cross ahead of the BAC 1-11 by only 0.3nm on a track that was at 79º (i.e.
         almost at right-angles) to that of the BAC 1-11. With the aircraft estimated to converge
         almost exactly at the intercept point, the resultant trajectory for the pass-behind required
         the BAC 1-11 to perform significant track change of 36º at the start of the manoeuvre,
         deviating from the original track by 6.6nm. Figure 3.8-2 shows the tracks of the two
         aircraft during the pass-behind manoeuvre, the BAC 1-11 originally on a southerly track
         and SIM02 moving approximately due east.




Figure 3.8-2 Tracks of BAC 1-11 and SIM02 for       Figure 3.8-3 Spacing of BAC 1-11 from SIM02
          1st Pass-Behind Manoeuvre                      during 1st Pass-Behind Manoeuvre
         The trajectory that had been produced by the FMS was not complete, however, despite
         being presented on the ND to the pilot for activation. The trajectory only consisted of the
         first half of the pass-behind manoeuvre, the remaining points in the trajectory being
         duplicates of the same point, the trajectory containing 128 points in total, the maximum
         permissible by the system. This problem had been seen occasionally during the rig
         testing, but its exact cause was not known, although in the past it had been linked to the
         prediction being incomplete for the STAR. The trajectory was activated and the aircraft
         followed the pass-behind manoeuvre to beyond the point of closest approach, the
         minimum spacing achieved being 6.2nm (see Figure 3.8-3). To create a new trajectory
QinetiQ/S&E/AVC/CR031041                                                                   Page 35 of 90
           for the remainder of the route, the pilot successfully generated direct to the next waypoint
           B3 and activated this trajectory, as seen in Figure 3.8-2.

           The two routes then converged again on the leg between B4 and B5. Once again the
           minimum spacing distance was set to 6nm and the resume waypoint was B5. This time,
           the simulated target aircraft, SIM02, was on a track that cut that of the BAC 1-11 at an
           angle of around 50º. When the pass-behind trajectory was generated, SIM02 was
           estimated to cross 3.1nm ahead of the BAC 1-11. This trajectory, activated by the pilot,
           had an initial track change of 24º and the deviation from the original leg to B5 was 6.3nm
           (see Figure 3.8-4). The FMS speed demand to the autopilot increased to 267kts CAS, in
           accordance with the prediction. As seen on the previous flight trial at Boscombe Down,
           however, this predicted air speed was greater than was actually required for the desired
           ground speed. Hence the minimum spacing distance that was achieved during this
           manoeuvre was 5.7nm compared with the originally defined value of 6nm, as shown in
           Figure 3.8-5. The application of a speed change during the lateral pass behind was not
           considered as an ideal solution for this type of manoeuvre and it was only really intended
           if there was an additional time constraint that had been imposed further along the route.




Figure 3.8-4 Tracks of BAC 1-11 and SIM02 for        Figure 3.8-5 Spacing of BAC 1-11 from SIM02
          2nd Pass-Behind Manoeuvre                       during 2nd Pass-Behind Manoeuvre
           A merge manoeuvre was also performed with the target aircraft SIM02. The simulated
           target followed a route from waypoint A5 to MMP and then to waypoint A6, while the
           planned route for the BAC 1-11 was initially on a track from B7 to B8 that paralleled the
           track of SIM02 from A5 to MMP. After B8, the BAC 1-11's route went to MMP and
           then, like SIM02, to A6. Once the turn at B7 was completed, a trajectory for the merge
           manoeuvre was generated, the merge waypoint being MMP and the required spacing was
           set to 20nm. Although a trajectory was successfully predicted, upon activation the speed
           demand to the autopilot remained at 267kts CAS, despite the trajectory having a
           predicted speed of 272kts CAS for the duration of the merge manoeuvre. The lateral
           element of the trajectory was carried out with the aircraft turning on to a track towards
           MMP, this change in track being about 10º. The trajectory still consisted of an S-shaped
           turn for capturing this direct track and therefore there was an initial overshoot of the
           second part of the turn, as seen in the trials at Boscombe Down. The spacing between the
           two aircraft at the start of manoeuvre had been 31nm, with the BAC 1-11 56nm from
           waypoint MMP. By completion of the initial part of the merge manoeuvre, i.e. when the
           BAC 1-11 passed the merge waypoint MMP, the spacing from SIM02 was around 21nm.
Page 36 of 90                                                               QinetiQ/S&E/AVC/CR031041
         The speed demand was still unchanged, however, and at this stage the aircraft in trail, i.e.
         the BAC 1-11, should now have adjusted its speed demand to match the ground speed of
         the lead aircraft, i.e. SIM02. Therefore, despite being about 1nm from the required
         spacing from SIM02, this had not really been achieved directly from the trajectory
         prediction, but was more a result of the BAC 1-11 already flying at a suitable ground
         speed for closing the distance on the target aircraft.

         The MFMS guidance was disengaged at waypoint MMP since the aircraft was then
         flown towards upper left corner of the trials area in order to verify the VDL4 coverage in
         this region where the DLR's ATTAS aircraft was to pass through on its planned route.

         Once the aircraft had landed back at Ciampino, the taxi map was selected on the ND.
         This revealed an offset in the FMS position data and the reference of the taxi map. The
         position data source being used by the FMS was the SBAS equipment. The FMS had
         automatically selected the Ciampino airport map, but the map showed the aircraft to be
         displaced about 200m to the left of its position on runway 15. Similarly, by the time the
         aircraft had reached its stand position on the apron, the position displayed on the map
         indicated that the aircraft was on the runway. The inference was that there was a position
         reference problem with the map display co-ordinates.

3.8.3    VDL4 Report

         On power up, the Ciampino ground transponder was observed, located at N41° 48.2050’
         E012° 35.0136’ 131.06m and identified with BICCA code JKAAA. It was transmitting
         in the interval sequence 4s, 4s, 8s, 4s, 4s, 8s, etc. and it was noted during this initial test
         that the ground transponder in use was an old type. AMS had experienced problems with
         the new transponder, transferred from the Boscombe Down ground station site. The
         aircraft transponder status was monitored using the T5 monitor software rather than
         connecting to the MFMS. This would allow the MFMS to conduct ASAS manoeuvres
         with simulated targets and provide a more reliable record of VDL4 performance in the
         trials area.

         The VDL4 link was maintained throughout three MFMS ASAS manoeuvres to the
         maximum route range of 108nm from Ciampino as shown in Figure 3.8-6. The aircraft
         route also explored the area where the ATTAS aircraft would be operating. At a
         maximum range of 90nm the VDL4 link was still reliable. However, it was noted that the
         voice communication at the southerly and north-western ends of the route had become
         broken and at times unreadable.




QinetiQ/S&E/AVC/CR031041                                                                     Page 37 of 90
                  st
                 1 Pass Behind
                  Manoeuvre                                                          Ciampino


                                                                                VDL mode 4
                                                                               Ground stations



                                                                nd
                                                               2 Pass Behind
                                                                Manoeuvre


                Part of ATTAS route




                            Start of Merge
                             Manoeuvre



                 Figure 3.8-6 Plot of ADS Position Data during Rome Test Flight
           The result of this flight declared the VDL4 ground/air link range performance to be
           acceptable for the Rome trials. Unfortunately, due to network problems, the air/ground
           data could not be received at the ENAV Shadow control workstation.




Page 38 of 90                                                           QinetiQ/S&E/AVC/CR031041
3.9      19th March 2003

3.9.1    Sortie Objectives

         For the transit back to Boscombe Down, the VDL4 transponder was again connected to
         the FMS in order to ensure that the data exchange with the transponder was functioning
         correctly and that no data errors could occur that might affect the FMS performance. As
         with the outbound transit flight, the FMS was to be used to provide lateral guidance to
         the autopilot during the majority of the flight.

         The flight planned route was:

         LIRA-OST-MEDAL-GILIO-ELBA-DOBIM-AKUTI-PIGOS-BARSO-OKTET-GIPNO-
         BULOL-ARDOL-CHABY-LAULY-BRY-CLM-KOPOR-KOMEL-GUBAR-GURLU-
         SAM-EGDM.

3.9.2    MFMS Report

         With the MFMS lateral guidance demands being used for nearly 2 hours of the flight, the
         mean cross-track deviation during this period was 108m with a standard deviation value
         of 269m. The reason for this larger standard deviation compared with the transit flight
         out to Rome is primarily due to the way two trajectories were generated and activated.
         For instance, ATC cleared the aircraft to go-direct to waypoint GIPNO just as it was
         approaching the turn at PIGOS. A trajectory was generated and activated about 4 seconds
         later. The trajectory predictor, however, would generate direct from the aircraft's current
         position to the new waypoint and therefore did not compensate for the turn required to
         get the aircraft on to the direct track. With a track change of just under 40º and a ground
         speed of over 460kts, the result was that, for a maximum bank angle of 25º, the aircraft
         deviated from the new track by a maximum of 2900m. before recovering back on to new
         route leg. This deviation was the principal cause of the larger standard deviation. With a
         better estimate of the route transition from the previous track to the new one, there would
         have been little difference in the data from the outbound flight.

3.9.3    VDL4 Report

         On power up for the third flight test in this sequence the ground based targets (BICCA
         codes of CH9SK, CQAIA, HORIA and JKAAA) were observed on the Navigation
         Display (ND) as being similar to the inbound test flight. The transponder was again used
         with the MFMS with no problems observed. The aircraft transponder was switched off
         before leaving Italian airspace.

         It was noted that the latest version of transponder software had not uploaded correctly.
         This was remedied for the Rome trials, moving the software version to 1.4.5.




QinetiQ/S&E/AVC/CR031041                                                                 Page 39 of 90
3.10       20th March 2003

3.10.1     Sortie Objectives

           Updated software was available for this flight in order to restrict the need for speed
           changes during the pass-behind manoeuvres when the forecast meteorological data was
           being used and also to improve the determination and execution of the speed changes for
           the merge manoeuvres. The flight profile itself was a repeat of that carried out on the 14th
           March using the company route QNQ1 and engaging the FMS demands as soon as
           possible after take off in order to follow the precision departure route, SID23.

           A total of four pass behind and two merge behind manoeuvres were to be performed
           during the course of this flight using simulated traffic generated in the same way as on
           the previous trials flights from Boscombe Down.

3.10.2     MFMS Report

           As usual, the FMS was initialised while the aircraft was still at the stand and 4D
           trajectory predicted from take off to the final approach path. With the position data
           source set to the GBAS equipment, the aircraft taxied out to the holding point for runway
           23 with the taxi map selected on the ND. As a situational awareness aid, the map worked
           reasonably well with the aircraft position being shown correctly for the taxiway that it
           was on at any time, but with a displacement of the order of 20 metres from the aircraft's
           exact location.

           A further trajectory generation was performed at the holding point and this was activated
           once the aircraft was on the runway. Following take off, the lateral guidance demands
           from the FMS were engaged through the autopilot to ensure a smooth tracking of the
           precision departure route. Once the aircraft was clean, the profile demands were also
           engaged, the aircraft levelling at FL50 as required until completion of the departure route
           at waypoint KATE. No problems were encountered during this part of the profile. The
           FMS was still using the GBAS equipment as the primary position source and this was
           maintained throughout the departure route. Figure 3.10-1 shows that for about 15
           seconds, while the aircraft was on the leg between waypoints WOLFE and INGLE, the
           GBAS dropped out of differential mode, probably due to masking from the ground
           station situated near the touchdown point of runway 23. A comparison of the GBAS data
           to the GPS truth track system during the departure segment of the flight (see Figure
           3.10-2) reveals that, while taxiing the lateral error in the GBAS data was less than 2.5m.
           However, once airborne, this figure reduced to being about 0.75m when the GBAS was
           in differential mode. For the short period when the GBAS was in stand-alone mode (no
           differential corrections), the lateral error only increased to 2m. Similarly, the error in the
           vertical axis was less than 2.5m while GBAS was in differential mode, increasing to 16m
           for the stand-alone condition. This level of accuracy would provide a significant
           contribution towards meeting requirements for achieving the capability of an RNP of
           0.5nm.




Page 40 of 90                                                                QinetiQ/S&E/AVC/CR031041
Figure 3.10-1 Precision Departure Route With      Figure 3.10-2 Accuracy of GBAS Position Fix
   Indication of GBAS Operational Mode                   Referenced to GPS Truth Track
         Once the aircraft had passed waypoint KATE and initiated its climb to the cruise altitude
         of FL240, the primary position source was reverted to SBAS for the en-route section. A
         smooth transition occurred when the change in position source was made with no
         significant lateral adjustment taking place. During the climb, ATC instructed that the
         climb should stop at FL220, so the autopilot was disengaged from the FMS profile
         demands. After being re-cleared to FL240 by ATC, the profile demands were re-engaged
         and the FMS demanded a continuation of the climb to FL240, the FMS recognising that
         the planned climb had not yet been completed.

         As per normal on these sorties now, the pilot entered the conditions for the first pass-
         behind manoeuvre into the MCDU, the target aircraft being SIM00, the minimum cleared
         spacing distance was 6nm and the resume waypoint was FFF. When the trajectory for the
         pass-behind manoeuvre was generated, the aircraft SIM00 was predicted to cross the
         track of the BAC 1-11 4.8nm ahead. The resultant trajectory deviated the BAC 1-11 by a
         maximum of 3.3nm from its original track with an initial required track change of 23º,
         extending the route by 0.6nm. The FMS speed demand remained unchanged at 250kts
         CAS. The minimum spacing that was actually encountered during the manoeuvre was
         6.4nm. The manoeuvre was completed successfully by waypoint FFF.

         The second pass-behind was set-up using the target aircraft SIM02, still using a
         minimum permitted spacing of 6nm, resuming the original route at waypoint HHH.
         Generating after the turn at GGG, the target aircraft, SIM02, was predicted to cross the
         track of the BAC 1-11 in 3.3 minutes time, at which point it would be 3.7nm ahead of the
         BAC 1-11. The resultant trajectory required an initial change of 25º in the track of the
         BAC 1-11, causing a maximum deviation of 5.1nm from the original track and an
         extension to the route distance of 1.9nm (see Figure 3.10-3). At the closest point of
         approach during the manoeuvre, the spacing between the two aircraft was 5.9nm (see
         Figure 3.10-4). In this case, the point of closest approach was almost exactly on the
         minimum spacing ring around the target aircraft. Thus, as the BAC 1-11 began to turn
         directly after passing this point, the spacing distance between the two aircraft remained
         relatively constant for about 30 seconds at around 5.9 to 6.0nm until the BAC 1-11 was
         on its new track for its recovery route back towards HHH.

QinetiQ/S&E/AVC/CR031041                                                               Page 41 of 90
   Figure 3.10-3 Tracks of the BAC 1-11 and        Figure 3.10-4 Spacing of BAC 1-11 from SIM02
                    SIM02
           For the third pass-behind manoeuvre, the selected target aircraft was SIM03, with a
           minimum cleared spacing of 6nm, to resume the original route at waypoint JJJ. When the
           trajectory was generated, the aircraft SIM03 was 2.5 minutes from intersecting the track
           of the BAC 1-11, at which point it was predicted to be 4.9nm ahead of the BAC 1-11.
           This therefore required a smaller track change of only 15º as compared to the previous
           manoeuvre, while the maximum deviation from the original track for the BAC 1-11 was
           3.3nm., which extended the route by 1.3nm. Once again, there was no change in the FMS
           speed demand in association with this predicted manoeuvre and the minimum spacing
           that was achieved relative to SIM03, during execution of the pass-behind, was 6.1nm.

           The pilot entered details for the final pass-behind on this flight using the MCDU,
           selecting the simulated aircraft SIM04 and a 6nm minimum spacing for the manoeuvre,
           intending to resume the planned route again at waypoint ZZZ. On this occasion, when the
           pass-behind trajectory was generated, the FMS predicted that the conflicting aircraft
           SIM04 would cross the track of the BAC 1-11 in 3 minute's time and at that point it
           would be 2.8nm ahead of the BAC 1-11. For this manoeuvre, the predicted trajectory
           required a track change of 19º and resulted in maximum deviation of 5nm from original
           route. The extension in the route length was of the order of 1.1nm. No speed change was
           predicted and as the BAC 1-11 followed the new trajectory, at the closest point of
           approach, the spacing from SIM04 was again approximately 5.9nm. As with the other 3
           pass-behind manoeuvres on this flight, there were no problems encountered during the
           execution of the manoeuvre and the lateral transitions were all implemented smoothly
           through the autopilot.

           The two merge manoeuvres were then attempted using the FMS during the second loop,
           west of the airway A25. As the BAC 1-11 approached waypoint PPP, the pilot selected
           the aircraft SIM09 and then entered the details into the MCDU for the merge. The
           requirement was to be 12nm behind SIM09 by the time the BAC 1-11 reached waypoint
           RRR. A trajectory was generated first time, after the BAC 1-11 had completed its turn at
           PPP and was on track towards QQQ. At this point, the SIM09 was 24.9nm from the BAC
           1-11 on a bearing of 28º relative to the BAC 1-11's current track to QQQ. A trajectory
           was generated direct to waypoint RRR to merge behind SIM09, this trajectory being
           activated within 5 seconds of generation. The FMS speed demand increased from 250kts
Page 42 of 90                                                            QinetiQ/S&E/AVC/CR031041
         CAS to 300kts CAS, in accordance with the predicted trajectory. Although 300kts CAS
         was the maximum value that the trajectory predictor was permitted to use for the cruise
         phase, the resultant trajectory estimated the BAC 1-11 would arrive within 3 seconds of
         the time constraint applied to waypoint RRR to ensure a spacing of 12nm from SIM09.
         To allow some flexibility in the prediction process, a time window of ±10 seconds was
         set-up around the time constraint and if the prediction was inside this window, then the
         constraint was assumed to have been met.

         For this manoeuvre, the aircraft SIM09 was 33.4nm from waypoint RRR and was
         predicted to get there at a time of 1611:57 (time at start of prediction was 1606:20). The
         system then predicted the time it would take SIM09 to travel a further 12nm (the required
         spacing distance) and this was added to the previous time in order to obtain the time at
         which the BAC 1-11 needed to arrive at RRR. It later transpired that the calculation of
         this additional time was incorrectly using the CAS value for the target aircraft rather than
         its ground speed and thus, in this case, the estimated time at which SIM09 would be
         12nm beyond RRR was 1614:48 rather than 1613:58. Over these extra 50 seconds, the
         target aircraft SIM09 would cover an extra 5nm. The trajectory that was subsequently
         predicted for the BAC 1-11 estimated that it would reach RRR at 1614:45. The speed
         profile was designed so that, when it reached RRR, it should be at the correct spacing
         while its ground speed should match that of SIM09. Therefore, a waypoint was inserted
         along the route just prior to RRR where the BAC 1-11 would start to change speed to
         match that of the target aircraft. Although the trajectory was successfully generated to
         meet the assigned time constraints, there was a flaw in the derivation of the true air speed
         from the computed air speed and consequently this affected the predicted ground speed.
         The result was the actual ground speed achieved for CAS demand of 300kts was not
         408kts as predicted, but was actually 442kts. The BAC 1-11 consequently arrived about
         33 seconds earlier than predicted at waypoint RRR. Along with the prediction being 3
         seconds earlier than the original constraint, the effect was that the aircraft SIM09 was
         only about 14 seconds ahead of where it needed to be for a spacing of 12nm. This extra
         distance was therefore of the order of 1.4nm and, given that the FMS assumed a point-to-
         point route for the target aircraft, the final spacing distance between them was actually
         determined to be 13.1nm.

         With the FMS speed demand returning to 250kts CAS when the BAC 1-11 reached RRR,
         the aircraft continued along the route and the system set-up for the second merge
         manoeuvre. For this one, the selected target aircraft was SIM10 and the merge waypoint
         was to be ZZZ. The required spacing distance at this merge waypoint was to be 18nm. A
         trajectory was generated once the BAC 1-11 had completed the turn at SSS and was on
         track to waypoint TTT. SIM10 was currently 23.5nm from the BAC 1-11 on a relative
         bearing of 30º from the BAC 1-11's current track. The direct distance from the BAC 1-11
         to the waypoint ZZZ was 63.1nm. A trajectory was generated first time and this predicted
         a CAS of 298kts to meet the defined time constraints at ZZZ for achieving the required
         spacing. As previously, this time constraint had been incorrectly derived with regard to
         the time it would take SIM10 to fly the extra 18nm beyond ZZZ. The result was a time
         constraint to be met by the BAC 1-11at ZZZ of 1630:54, which should have been
         1630:21.

         The predicted trajectory for this second merge manoeuvre did not entirely conform with
         the expected speed profile in which the speed changed to that of the target aircraft at the
QinetiQ/S&E/AVC/CR031041                                                                  Page 43 of 137
           inserted manoeuvre waypoint prior to ZZZ. In this case, the change in speed was not
           predicted to occur until the waypoint ZZZ itself and the CAS value estimated for the
           target aircraft from its ground speed data was 303kts, which was significantly in excess
           of the correct value. This should have been in the range of 260 to 270ktsCAS. It is
           possible that data from the first merge behind manoeuvre was influencing the prediction
           of the second or a limit had been reached within the FMS with regard to the total number
           of new subphases that could be created throughout a flight.

           The FMS guidance function successfully followed this new trajectory following
           activation of it. As before, the actual ground speed for a CAS of 298kts was greater
           (398kts) than the predicted value (354kts). In this case, the difference was 44kts and the
           BAC 1-11 arrived at the merge waypoint not only 1 minute earlier than predicted, but
           also 27 seconds before the simulated aircraft SIM10 had reached a distance of 18nm
           from ZZZ on the track to the next waypoint KKK. In 27 seconds, SIM10, at ground
           speed of 357kts, would cover about 2.8nm and this agrees with the actual distance
           between the positions of the two aircraft at this point being 15.2nm (compared with the
           planned spacing of 18nm). Once passed ZZZ, the FMS reverted to its normal cruise
           speed demand of 250kts CAS.




 Figure 3.10-5 Approach Route with Indication          Figure 3.10-6 Accuracy of GBAS Position Fix
          of GBAS Operational Mode                             Relative to GPS Truth Track
           For this flight, the FMS lateral guidance was maintained through the autopilot until the
           aircraft was actually in the turn between waypoints DM002 and DM003, bringing the
           aircraft on to the final approach path. This was partly to determine how well the
           waypoints had been defined to be in line with the localiser path that was being generated
           by the GBAS equipment (the GBAS was already known to be aligned with the airfields
           ILS localiser). Earlier, with the aircraft already in the STAR, the primary position source
           for the FMS had been changed from the SBAS to the GBAS. The FMS was able to
           maintain the aircraft on the defined lateral path throughout this final turn, but the position
           of DM003 was determined to be a few hundred metres to the right of the localiser centre
           line. Therefore, as the aircraft passed through the centre line, the pilots disengaged the
           system to ensure a suitable capture was made of the approach. Figure 3.10-5 shows the
           approach route up to the end of the FMS lateral guidance at DM003, along with regions
           during this phase where the GBAS was operating in differential mode (highlighted in
           green). Figure 3.10-6 indicates the accuracy of the GBAS position data (relative to a GPS
Page 44 of 90                                                                QinetiQ/S&E/AVC/CR031041
         truth track system) during the period that the FMS was using it as the primary position
         source. It can be seen that laterally, when in differential mode, the error was a fraction of
         a metre and vertically it was 1 to 2m.The actual lateral guidance performance throughout
         the flight is indicated in Figure 3.10-7, which shows the cross-track error derived by the
         MFMS. This demonstrates that the flight technical error for the lateral guidance of the
         MFMS was typically no more than about 150m (or 0.08nm) when controlling to the route
         and for all of the pass behind manoeuvres. The exact figures for this flight were a mean
         error of 70m with a standard deviation of 154m. The spikes that can be seen in the
         deviation are related to the go-direct function used for the merge behind manoeuvres and,
         as explained previously, result from the aircraft being unable to react sufficiently quickly
         to follow the second part of the predicted S-shaped path. A smoother predicted track
         transition, similar to that employed for the pass behinds would resolve this and should
         mean that the same level of performance would be achievable throughout the flight.




Figure 3.10-7: MFMS Lateral Guidance Performance




QinetiQ/S&E/AVC/CR031041                                                                   Page 45 of 90
3.11       Summary of GBAS Approach Guidance Presented on the Primary Flight Display.

           In preparation for the auto-coupled approach trials of the MFMS, GBAS guidance
           information was displayed on the primary flight display (PFD) on the left-hand side of
           the cockpit. The aim of this was to give the flight crew experience of the anticipated
           auto-coupled approaches and the steeper approach angle (4.5°) that was to be used during
           some of the later auto-coupled approach work. A summary of the flights on which GBAS
           information was used for manual approaches is given in Table 3.11-1:

                       Date         Number of Approaches          Glideslope
                       7 Feb                   2               1 @ 3º, 1 @ 4.5º
                      19 Feb                   1                       3º
                      27 Feb                   2                       3º
                      11 Mar                   2                      4.5º
                      14 Mar                   2                      4.5º
                      20 Mar                   1                       3º
                                                           Table 3.11-1: GBAS Guided Approaches
           The quality of the GBAS information will be assessed on a flight-by-flight basis and will
           be presented in terms of the percentage full-scale deflection (FSD) of the guidance
           deviations. Also shown on the graphs will be the equivalent FSD of the guidance
           deviations derived from the post-processed GPS truth source.

           The approaches were conducted after following representative STAR procedures, an
           example of which is given in Figure 3.11-1.




                                          Figure 3.11-1: Representative STAR for Boscombe Down

Page 46 of 90                                                                QinetiQ/S&E/AVC/CR031041
3.11.1   7th February

         This flight was divided into two separate circuits with a full stop landing and FMS reset
         in between. At the end of the first circuit a 3º approach was flown, with the GBAS
         providing guidance. At the end of the second circuit the approach was conducted at 4.5º.

         During both of these approaches the GBAS coverage remained constant with no drop-
         outs observed and the guidance displayed to the pilot was seen to be good.

         It can be seen in Figure 3.11-1 that there is a small divergence between the GBAS and
         truth glide slope data with approximately half a mile to go to touchdown during the first
         approach. This divergence was found to be approximately 1m in both the glide slope and
         localiser deviations. The reason for the start of the glide slope trace being at –100% FSD
         is due to the fact that the aircraft was flying level below the glide slope prior to capture at
         approximately 8nm to go.




                               Figure 3.11-1: 7th February GBAS Approach 1 using 3º Glide Slope
         The data for the second approach (see Figure 3.11-2) shows a similar pattern to the first
         approach, the only difference being that the divergence in the glide slope is more gradual
         from about one and a half miles to go. The corresponding difference between the GBAS
         and truth glide slope deviation was found to increase by 1.5m.




QinetiQ/S&E/AVC/CR031041                                                                     Page 47 of 90
                              Figure 3.11-2: 7th February GBAS Approach 1 using 4.5º Glide Slope
3.11.2     19th February

           During this flight, a problem developed with the GBAS equipment during the downwind
           leg prior to the approach. The reacquisition of the GBAS VHF broadcast had been
           achieved during the descent and, as the aircraft passed the AGIBS waypoint, the
           equipment reported a non-differential operating mode. At first it was thought that this
           was just caused by masking of the VHF transmission as had been seen before and that the
           transmission would be reacquired in due course. As the flight progressed it appeared
           more and more apparent that the signal would not be reacquired, and eventually the
           GBAS approach was aborted. It can be seen from Figure 3.11-1 that the difference in
           guidance between the post-processed GPS truth track and the non-differential GBAS is
           most marked in the glide slope due to the nature of the uncorrected errors in the stand
           alone GPS system.

           Analysis of the logged GBAS data showed that there had been no differential messages
           logged from the aircraft VHF receiver after the AGIBS waypoint. The reason for this is
           unknown, but could be caused by a communications breakdown between the GBAS
           processing unit and VHF receiver or by an integrity failure of one of the components of
           the ground system.




Page 48 of 90                                                           QinetiQ/S&E/AVC/CR031041
                              Figure 3.11-1: 19th February GBAS Approach using 3º Glide Slope
3.11.3   27th February

         The GBAS equipment performed in a similar fashion to that seen before, with patchy
         coverage on the downwind leg, probably due to masking of the VHF transmission. Once
         again there is a small divergence between the GBAS and truth glide slope guidance data
         evident on both approaches corresponding to a difference in linear deviation of
         approximately 1m.

         During the first approach the pilots commented on an observed two dot jump in the
         displayed guidance, but a distinct jump is not evident on the trace of the internally
         recorded data.

         The second approach was conducted with good guidance and no data anomalies were
         seen.

3.11.4   11th March

         During this flight the GBAS coverage was again patchy during the downwind leg, with
         several outages observed in the VHF transmission. During this flight the first approach
         had to be aborted due to problems with the display of GBAS guidance data. This problem
         appeared to be related to the analogue guidance signal provided by the equipment for
         display on the PFD. During the aborted approach, guidance computations were occurring
         and appropriate numerical deviations were being displayed on the GBAS equipment in
         the aircraft cabin. The guidance bugs on the PFD (and repeater in the cabin), which are
         driven by analogue outputs from the GBAS equipment, were seen to remain centred and
         red. This approach was broken off with 5nm to go and the GBAS equipment reset while

QinetiQ/S&E/AVC/CR031041                                                              Page 49 of 90
           the aircraft repositioned for further approaches. The cause of the fault was not isolated,
           but resetting the GBAS equipment cleared the error.

           The first successful approach of the sortie was achieved after the GBAS equipment had
           been restarted. The fault observed previously where the guidance bugs did not move was
           not re-observed and the approach was conducted to a low overshoot. Good agreement
           between the GBAS and truth-derived guidance was observed.

3.11.5     14th March

           During this flight the GBAS coverage was again patchy during the downwind leg and
           also for part of the base turn, where outages were observed in the VHF transmission.
           During this flight the first approach again had to be aborted due to problems with the
           analogue guidance data preventing the bugs from becoming active on the PFD. While the
           aircraft repositioned for another approach the GBAS equipment was again reset to
           recycle the hardware.

           The first approach after the reset was successful with good guidance provided to the
           pilots and no observed dropouts.

           The second approach was also successfully conducted with good agreement between the
           GBAS and truth data.

3.11.6     20th March

           After the problems observed during the previous approaches, the GBAS equipment was
           restarted before the aircraft passed the AGIBS waypoint. No problems were encountered
           during the approach with the guidance bugs on the PFD.

           The guidance supplied to the pilot was generally good, although when the aircraft was
           within 4nm of touchdown there were some jumps in the differences between the GBAS
           and truth glide slope guidance of the order of 1 to 1.5m. These differences were
           essentially undetectable on the guidance displayed on the PFD.

3.11.7     Summary

           The guidance information provided by the GBAS equipment was, on the whole, reliable
           and of sufficient accuracy to enable the pilots to follow the beam bars on both a 3.0º and
           4.5º glide slope approaches.

           The source of the problems with the analogue guidance was independent of the GBAS
           signal-in-space, and most probably caused by an intermittent hardware fault within the
           receiver or somewhere else in the signal path between the GBAS equipment and the
           aircraft displays.

           The loss of the GBAS link during the downwind leg should not have an impact on the
           operational deployment of the system. It is envisaged that the coverage of airport GBAS
           installations should be similar to that of an ILS and not be based on an omni-directional
           antenna similar to that used in the experimental installation at Boscombe Down.

Page 50 of 90                                                             QinetiQ/S&E/AVC/CR031041
3.11.8   Summary of MA-AFAS Glide Slope Guidance Compared to GBAS

         For the later phases of the MA-AFAS project where precision curved approaches were
         planned, the MFMS would have to provide glide slope guidance commands to the
         autopilot during the initial curved section of the approach, the autopilot transitioning to
         GBAS-derived approach guidance once the aircraft had captured the localiser. To test the
         computation of these guidance commands, comparisons were made between the GBAS
         guidance data and the MFMS glide slope guidance data. At this stage of the development
         of the MFMS precision approach function, there was still some refinement to be done to
         the algorithms, but this test was to verify that the MFMS was capable of using actual
         GBAS position solution information to determine a reasonable glide slope deviation
         value.




                            Figure 3.11-1: Glide Slope Comparisons for Approaches on 14 March
         It can be seen in Figure 3.11-1 that there is a difference between the two deviation traces
         (the GBAS-derived deviation is shown red whilst the MFMS-derived value is in blue),
         which on closer inspection appears to be constant.




QinetiQ/S&E/AVC/CR031041                                                                 Page 51 of 90
                                       Figure 3.11-2: GBAS and MA-AFAS Glide Slope Differences
           It can be seen from Figure 3.11-2 that the offset appears constant at about 5% FSD
           (equivalent to 0.00875 DDM) when the guidance bugs are active. Close comparison of
           the two traces also shows that the MA-AFAS-derived information is noisier than the
           GBAS. The exact reason for this is unknown but was possibly related to the way the
           Arinc data from the GBAS was handled by the MFMS, which consequently could
           influence the derivation of the distance to go to the touchdown point. With the position
           information from the GBAS being split between coarse and fine components over the
           Arinc data bus, if these components were not exactly matched on each iteration, then this
           might result in some background noise affecting the solution.




Page 52 of 90                                                             QinetiQ/S&E/AVC/CR031041
3.12     Summary of Boscombe Results

         The flight trials that took place at Boscombe Down and the single check flight at Rome
         achieved the aim of developing the MFMS to a sufficient level of performance that it
         could be used in the subsequent trials at Rome with two live aircraft. None of the flights
         had to be aborted due to system failures and the MFMS was demonstrated to be capable
         of meeting the general objectives of each flight within any limitations imposed by its
         developmental progress at that stage. This also vindicated the process of ground testing
         and validation that was being followed during the trials period before system
         improvements were tested in the airborne environment. In the majority of situations, the
         type of performance experienced with the MFMS during these flight trials was
         comparable to that which had been determined from the ground testing.

         The initial couple of flights had primarily proved the compatibility of the MFMS with the
         various other avionics systems to which it was interfaced onboard the BAC 1-11. The
         key system in this respect was the autopilot, which was demonstrated to operate without
         any serious problems when using the demands from the MFMS. These first flights were
         also important in verifying the basic FMS functionality of the system for predicting a 4D
         trajectory for the aircraft and generating the necessary autopilot demands for guiding the
         aircraft around the planned flight profile. There had been added emphasis on this aspect
         due to the project having been forced into producing a complete FMS from scratch,
         which had not been the original plan.

         With the main baseline FMS functionality in place, it was possible to start flight testing
         those components of the ASAS delegated manoeuvres that had been implemented. These
         essentially consisted of the pass behind and merge behind manoeuvres. All these flight
         trials used simulated aircraft targets within the MFMS against which these ASAS
         manoeuvres could be generated.

         The main route used for the Boscombe Down trials incorporated four pass behind
         situations, with different configurations in the relative tracks of the two aircraft where
         they crossed. These included a shallower intercept where the two aircraft were flying
         comparable tracks and two situations where the simulated aircraft had a component of
         track that was in direct opposition to that of the BAC 1-11. In practically all of these
         tests, the simulated aircraft was timed to reach the intercept point ahead of the BAC 1-11
         in order to replicate the condition that would require the BAC 1-11 to be the selected
         aircraft to manoeuvre around the simulated one. Apart from when problems arose with
         the position data being used by the MFMS, the trajectory predictor was always able to
         generate a lateral pass behind manoeuvre. The one exception to this had been the first
         pass behind during the check flight in Rome, when a separate problem prevented the
         predictor from creating a complete trajectory.

         A total of 17 pass behind manoeuvres were flown against simulated target aircraft. The
         geometry of the route meant that typically, the BAC 1-11 was less than 3.5 minutes from
         the intercept point when the pass behind was predicted. Assuming a standard separation
         distance of 5nm, then this separation would likely have been compromised anything up
         to 1 minute before the BAC 1-11 had reached the track intercept point. Although it
         should be noted that the pilots had been requested to activate the trajectory as soon as
         they were satisfied with it on the lateral map display, the time between the pilot
QinetiQ/S&E/AVC/CR031041                                                                Page 53 of 90
           requesting the MFMS to generate the manoeuvre and then activating it was often within
           5 seconds. The MFMS was allowing a 15-second period over which the aircraft was
           assumed to be maintaining its currently planned profile before any new turn was inserted,
           so this comfortably covered the response times that were being encountered.

           All the lateral manoeuvres were predicted within the standard performance limits used by
           the MFMS for normal generations of en-route flight profiles. For the situations where a
           6nm minimum spacing was defined, the maximum deviation of the BAC 1-11 from its
           originally planned track was 6.6nm, although the situations were that the target aircraft
           was always passing ahead of the BAC 1-11. Although there was insufficient data to make
           a definitive statement, the magnitude of this deviation from the original track tended to
           be dependent more on the relative time difference between the two aircraft passing the
           intercept point in their tracks, rather than the relative difference in the track angles
           themselves.

           The flight trials from Boscombe Down were also used to test the ability of the MFMS to
           merge the BAC 1-11 to be a specified distance behind a simulated target aircraft. More
           problems were encountered with this function than with the lateral pass-behind. This was
           primarily due to the need to add in additional speed change subphases, the
           implementation of which was not initially being handled correctly. It was not really until
           the flight on the 20th March that the insertion of the speed change subphases was
           functioning more as expected and the true behaviour of the merge behind process itself
           could be investigated. The results on the 20th March revealed that, although the system
           was producing speed demands to modify the spacing distance between the two aircraft,
           the magnitude of these demands was not quite correct to achieve the specified distance. It
           was clear that the trials were demonstrating the feasibility of the concept, but not
           necessarily an error-free implementation. However, with the equipment needing to be
           packaged ready for transportation to Rome for the live aircraft trials directly following
           this flight, further evaluation of the system and the testing of additional software
           modifications could not be carried out prior to the start of the Rome trials themselves.

           The flight trials had shown that the MFMS was able to guide the aircraft to the predicted
           flight profile for both the pass behind and merge behind manoeuvres. Since the MFMS
           used the standard FMS trajectory prediction functions in the determination of these
           manoeuvres, the level of guidance performance was essentially unchanged from that
           achieved under normal en-route conditions.

           The en-route navigation performance was predominantly dependent on the use of the
           SBAS as the primary position source. When compared against a GPS truth track, the
           SBAS was shown to be in error laterally by only 1.3m. With the MFMS itself exhibiting
           a flight technical error of the order of 70m (data from the flight on 20th March), then this
           would demonstrate that such a system could provide the capability of meeting RNP 1
           performance. The trials, however, also highlighted the need for back-up systems to
           provide a means of cross-checking the integrity of the position solution.

           Some performance data was obtained for the GBAS equipment to consider its ability to
           support precision departure and arrival operations using the MFMS. Relative to the truth
           track system, the GBAS was seen to be capable of providing a position solution with a
           lateral error of less than 1m and a vertical error within 3m while in differential mode.
Page 54 of 90                                                               QinetiQ/S&E/AVC/CR031041
         FMS guidance could be maintained through the departure and arrival routes, although
         system development was not sufficient to perform in-flight precision approach trials with
         the FMS providing initial glide path guidance to the autopilot.

         During these flight trials at Boscombe Down, the VDL4 data link communications tests
         were concentrating on range and integrity performance of the system. The initial
         transponders that were used were found to lose the communication link with the ground
         station at the extremities of the original trials route (a range of around 150nm). This
         resulted in the update to the route that ensured the aircraft remained within reliable link
         coverage range of 135nm from the ground station at Boscombe Down. Revised versions
         of the transponders also improved the performance of the system, not in range but with
         regard to the speed at which the link could be re-established in situations where data link
         communications were briefly lost. Only the ADS-B component could be tested during
         these trials while the point-to-point communications for CPDLC continued to be ground
         tested. The trials also indicated that the air-to-ground link performed more reliably than
         the ground-to-air link, when the VDL4 ground station at Boscombe Down was in use.
         Additionally, the operation of a second airborne transmitter on the ground during one
         flight had provided a level of confidence that the air-to-air ADS-B reports would be
         successfully exchanged between the two live aircraft during the flight trials in Rome. It
         had also been possible to connect the transponder at the ground station to the air/ground
         router and to the Broadcast Application Ground Server (BAGS) in order to relay ADS-B
         data to a simulated ATC position at the Eurocontrol Experimental Centre (EEC) at
         Bretigny, France. There were some issues with the validity of some of the recorded ADS-
         B data (altitude and speed) which were subsequently investigated in the lab but were not
         fully resolved.




QinetiQ/S&E/AVC/CR031041                                                                 Page 55 of 90
4          Rome Trials

4.1        24th March 2003

4.1.1      Sortie Objectives

           This was the transit flight by the QinetiQ BAC 1-11 to Ciampino Airport in Rome in
           preparation for the MA-AFAS trials that would be taking place here during this week. As
           with the transit flights to and from Rome during the previous week for the practice sortie,
           the MFMS was to be used to provide lateral guidance demands to the autopilot for the
           majority of this flight.

           The flight planned route was:

           EGDM-SAM-MID-XAMAB-BAMES-RBT-PTV-NEV-LERGA-LATAM-KURIR-
           RETNO-KOLON-DIVUL-SODRI-BTA-ELBA-GILIO-MEDAL-GILET-OST-URB-
           LIRA.

           The FMS was also to be coupled to the VDL4 transponder with the intention of
           investigating whether air-to-air broadcast data could be successfully received from DLR's
           ATTAS aircraft, which was due to arrive at Rome shortly before the BAC 1-11. The
           airborne transponder had been reset to the Rome trials frequency and therefore was not
           cleared to be powered outside of the Italian FIR.

4.1.2      MFMS Report

           Having initialised the FMS on the ground as usual, the aircraft took off and once ATC
           had vectored the aircraft to join the airways system, a new trajectory was generated by
           performing a go-direct to waypoint HAWKE, this trajectory then being activated. The
           FMS lateral guidance was subsequently engaged through the autopilot.

           While transiting across France, ATC deviated the aircraft from its original pre-planned
           route using airway UM728 on to a neighbouring airway UL612. This occurred about
           12nm before reaching Pithiviers (PTV) and consequently the FMS lateral guidance was
           disengaged from the autopilot while the pilot was receiving heading instructions from the
           controller. After some negotiation with the controller, the aircraft was permitted to return
           to its original route, rejoining on the leg from Nevers (NEV) to the waypoint LERGA.
           Although FMS lateral guidance was re-established, it appears from the post-flight
           analysis that the FMS source had been changed from the MFMS to the secondary lateral
           guidance system onboard the BAC 1-11, which was being used as a backup to the MA-
           AFAS equipment. This was corrected, but not until much later in the flight and therefore
           there is only about 40 minutes of this flight for which cross-track deviation data can be
           considered as valid. From this data, the flight technical error related to the cross-track
           deviation term shows a mean value of 40m with a standard deviation of 29m and a
           maximum offset of 206m.

           Once in Italian airspace, the VDL4 transponder was activated. The ADS-B reports from
           the ATTAS were successfully received by the VDL4 transponder and this data passed to
Page 56 of 90                                                               QinetiQ/S&E/AVC/CR031041
         the FMS. An aircraft symbol was displayed on the lateral map of the ND to indicate the
         position and track of the ATTAS, the CDTI overlay having been selected by the pilot
         using the trackball and the soft keys on the display. The identity code for the ATTAS
         was also shown alongside the symbol with the ground speed of the ATTAS and its height
         offset (in hundreds of feet) relative to the BAC 1-11. The identity code was indicated as
         IOEAA while the height offset appeared to be valid as both aircraft were descending
         towards Rome. The ground speed data for the ATTAS was, however, about 10 times too
         large (displayed as 2430kts), consistent with the results that had been seen on the 19th
         March concerning the ADS-B data for the BAC 1-11. The other observation from
         monitoring the CDTI data on the ND was that there were occasional instances of the
         ground speed and track angle data being zero. Along with the magnitude of the ground
         speed data, these zero values would need to be corrected, or at least filtered out, before
         the main trials began at Rome, in case these affected the trajectory prediction process for
         the pass-behind and merge manoeuvres.




QinetiQ/S&E/AVC/CR031041                                                                 Page 57 of 90
4.2        25th March 2003

4.2.1      Sortie Objectives

           This was to be the first proper test of the MA-AFAS system in which two live aircraft,
           the BAC 1-11 and the VFW-614 ATTAS, would perform delegated ASAS manoeuvres
           with air-to-air broadcast data being provided via the VDL4 data link. The routes flown
           by the two aircraft in the assigned trials area are shown in Appendix A.3. Both aircraft
           were at FL210, with the ATTAS cruising at 210kts CAS and the BAC 1-11 at 250kts
           CAS, although this could be modified to co-ordinate the timings during the flight. The
           BAC 1-11 was to take-off 10 minutes after ATTAS aircraft in order to arrive in the area
           of the first pass behind manoeuvre at the right time. A total of two pass behind and one
           merge behind manoeuvres were to be tested. The BAC 1-11 would carry out these
           manoeuvres while the ATTAS maintained its planned track. The trials scenarios were
           planned to not reduce aircraft separation below the standard 5nm. A dedicated radio
           communications link was provided on P134.900, S129.000 in order to allow co-
           ordination of the trial between the various participants.

           Updates had been made to the FMS surveillance database software with regard to the
           handling of the ADS-B reports received from the VDL4 transponder. The ground speed
           data was now divided by an additional power of 10 to ensure that the manoeuvre
           prediction process used the correct speed. The reports were also filtered to remove any
           that contained zero track or ground speed values, while those reports relating to the own
           aircraft were ignored so that this data would not be displayed on the CDTI overlay. It had
           been noted that the identity information had also been decoded incorrectly, so this had
           been revised as well, the new codes that were expected are listed in Table 4.2-1.

                       Aircraft/Ground Station          VDL4 Identity
                              BAC 1-11                    AAEQJ
                               ATTAS                      AAEEH
                              Ciampino                    AAAJI
                               Table 4.2-1: Updated VDL4 Identity Codes
           The route that had been used on the practice flight on the 19th March had also been
           modified to allow a little extra route distance beyond the merge waypoint (in this case
           MMP). The FMS company route that was selected by the pilot was now QNQ6. The
           STAR that was selected related to an approach to runway 33 at Ciampino, although it
           was expected to be runway 15 that would be used. This was an initial precaution in case
           the short STAR used for runway 15 had caused some of the manoeuvre prediction
           problems encountered on the 19th March.

           As the ground tests of the VDL4 communications network in Rome were not able to
           produce a reliable Point to Point capability, the fall back position of using voice
           communication for the ATC control of the delegated ASAS manoeuvres was adopted.

4.2.2      MFMS Report

           With the flight profile data entered into the FMS, a trajectory was generated while the
           BAC 1-11 was still on the ground at Ciampino, the aircraft taking off, as planned, 10
Page 58 of 90                                                             QinetiQ/S&E/AVC/CR031041
         minutes after the ATTAS aircraft had departed. With the trajectory having been activated
         while the BAC 1-11 lined up on the runway, the FMS guidance demands started
         automatically once the aircraft was airborne and lateral guidance could be engaged
         through the autopilot prior to the start of the first turn at SOT15. FMS profile guidance
         demands were not engaged through the autopilot until the aircraft was on track towards
         Ostia (OST) at about FL60. The aircraft climbed under FMS control to the cruise at
         FL210, accelerating as planned from 230kts to 250kts CAS as it reached FL100.
         Meanwhile, the ATTAS aircraft was also at FL210, but flying at 210kts CAS.

         Once the BAC 1-11 had entered the trials area and passed waypoint B2, a manoeuvre
         was generated to pass-behind the ATTAS aircraft (code AAEEH) with a minimum
         allowed spacing of 6nm, resuming the original route at waypoint B3. This was in
         response to the instructions via R/T from the trials ATC controller to select this target
         aircraft, which had to be first confirmed by the pilot, before the further instruction was
         received defining the requirements for the manoeuvre itself. The current tracks of the two
         aircraft would intercept one another at waypoint P1, the ATTAS crossing at an angle of
         80º to the track of the BAC 1-11. At the point when the pass-behind manoeuvre was
         generated at 0914:25, the ATTAS was predicted to reach P1 at 0917:00, approximately
         30 seconds before the BAC 1-11. With the ground speed of the BAC 1-11 being 42kts
         greater than that of the ATTAS aircraft, the estimated separation by the time the BAC 1-
         11 was at P1 would have been 2.4nm.




                           Figure 4.2-1 Spacing during first Pass-Behind Manoeuvre, 25th Mar. 03


QinetiQ/S&E/AVC/CR031041                                                                Page 59 of 90
           The resultant trajectory for the first pass-behind required an initial track change for the
           BAC 1-11 of 26º, the aircraft deviating by 4.9nm from its original track before
           recovering back to waypoint B3. The duration between the pilot triggering the trajectory
           prediction and then activating it was around 5 seconds. During the course of the
           manoeuvre, the actual minimum spacing that was achieved relative to the ATTAS was
           6.4nm (see Figure 4.2-1).

           Having cleared the ATTAS aircraft, the pilots increased the speed of the BAC 1-11 to
           280kts for about 3.5 minutes to try and reduce the time difference between the
           predictions of when the two aircraft would arrive at waypoint P2 for the second pass-
           behind manoeuvre. The ATTAS would cross the path of the BAC 1-11 at P2 when the 1-
           11 was on the leg from B4 to B5, the ATTAS being on a track that was 50º to that of the
           BAC 1-11. At 0923:51, a trajectory was generated at the first attempt with the FMS
           predicting that the ATTAS would reach waypoint P2 at about 0927:04, 25 seconds before
           the BAC 1-11. The BAC 1-11 would be approximately 2.2nm from P2 when the ATTAS
           reached this waypoint. The pass-behind manoeuvre generated by the FMS required an
           initial track change of 27º and would deviate the BAC 1-11 by 7.3nm from its original
           track in order to ensure a 6nm minimum spacing. Once again, there was a 5-second
           period between the start of the prediction process and the trajectory being activated. The
           manoeuvre was carried out successfully and the BAC 1-11 resumed its previous route at
           waypoint B5, having passed the ATTAS at a minimum distance of 6.1nm.




                         Figure 4.2-2 Spacing during second Pass-Behind Manoeuvre, 25th Mar. 03

Page 60 of 90                                                              QinetiQ/S&E/AVC/CR031041
         For the merge manoeuvre, a spacing distance of 20nm was entered into the MCDU and
         the merge waypoint was defined as MMP. No software modifications had been
         implemented since the flight on the 20th March. This was due to there having been no
         available time to identify the problems, apply the necessary software modification and
         ground test using the aircraft model rig before all the equipment had had to be loaded on
         to the aircraft in preparation for the detachment to Rome. Thus the same basic
         performance problems still existed as previously described for the flight on the 20th
         March at Boscombe Down. However, the principle of the merge manoeuvre could be
         tested. With the system configured for the merge behind, a trajectory was predicted at
         0941:59. At this point, the BAC 1-11 was currently 32.9nm from the ATTAS while the
         waypoint MMP was 66.9nm on a direct track from BAC 1-11's current position. A
         trajectory was generated and activated about 4 seconds after the pilot initiated the
         prediction. Although the aircraft turned on to the direct track to MMP, there was no
         change in the speed demand from the FMS. Post-flight analysis revealed that there had
         been a problem with part of the ADS-B surveillance data used by the FMS to predict the
         flight path of the target aircraft, i.e. the ATTAS. The manoeuvre generator process used
         the height rate from the ADS-B report for the target aircraft to estimate the aircraft state
         conditions for this aircraft at the merge waypoint, MMP. The height rate data was not
         providing any sensible information, however, the value being set to 9999m/s, and this
         caused incorrect data to be derived for the speed and time of the ATTAS at MMP. Since
         the manoeuvres were all being executed in level flight, this parameter was set to always
         be zero in the surveillance database and thus allow this process to function better.

         Throughout the flight, the ADS-B reports had been received from the ATTAS via the
         VDL4 transponders and processed by the FMS giving sufficiently continuous updates
         that allowed the position of the ATTAS to be monitored successfully on the lateral map
         of the ND. This information remained on the display and it was not noted that it
         disappeared at any time. The ATTAS, however, had had problems obtaining the ADS-B
         reports from the BAC 1-11 when the two aircraft were at the same altitude.




QinetiQ/S&E/AVC/CR031041                                                                  Page 61 of 90
4.3        25th March 2003

4.3.1      Sortie Objectives

           In the afternoon, the BAC 1-11 and the ATTAS repeated their flights from the morning,
           but with the FMS surveillance database now always setting the height rate within the
           target aircraft reports to be zero. Otherwise the FMS was set up identically to the
           morning's flight, although with the latest meteorological forecast file loaded.

4.3.2      MFMS Report

           The departure of the BAC 1-11 was delayed by about 2 minutes due to another aircraft
           being on the approach at the 10-minute mark after the ATTAS had taken off. The
           ATTAS was therefore requested over the trials R/T frequency to delay its progress to
           allow the BAC 1-11 to catch up. Consequently, the ATTAS slowed down in order to co-
           ordinate the arrival times of the aircraft at waypoint P1.

           With the trajectory activated after lining up on the runway, the output of the FMS
           guidance demands was triggered after take off and the lateral demands engaged through
           the autopilot before the first turn. The profile demands were also engaged earlier than on
           the previous flight, the aircraft being at about 2500ft. As the aircraft reached the cruise at
           FL210, degradation in the performance of the SBAS started to occur. It is believed that
           there was some source of interference in the area that was affecting the reception of the
           GPS signals. With the SBAS being unable to maintain lock on to the signals from
           sufficient GPS satellites, it was decided to change the primary position source for the
           FMS to be the IRS. This was done without any problem to the flight itself.

           By the time the first pass-behind manoeuvre was to be carried out, both aircraft were at
           their originally planned cruise speeds of 250kts CAS for the BAC 1-11 and 210kts CAS
           for the ATTAS. The position source for the FMS on the BAC 1-11 had also been
           reverted back to the SBAS after clearing the area of GPS interference. The configuration
           of the pass-behind was unchanged from the previous flight, the minimum permitted
           spacing distance being 6nm. A trajectory was predicted at 1419:30, the FMS estimating
           that the ATTAS would reach waypoint P1 at 1421:39, approximately one and a quarter
           minutes before the BAC 1-11, so that by the time the BAC 1-11 had reached P1, the
           distance to the ATTAS would be 6.1nm. The trajectory that was produced for the pass-
           behind manoeuvre only required a 7º change in track for the BAC 1-11, which resulted in
           a total deviation from the original track of 2.1nm. The time between initiation of the
           prediction and activation of the trajectory was just 4 seconds and the spacing between the
           two aircraft at the closest point of approach during the manoeuvre was 6.4nm.

           Once clear of the ATTAS, the speed of the BAC 1-11 was increased to try and make up
           some of the earlier delay and to co-ordinate the ETAs of the two aircraft at waypoint P2
           for the second pass-behind manoeuvre. An additional region of GPS interference was
           encountered shortly before carrying out this pass-behind so the IRS was selected as
           position source until completion of the manoeuvre. With the BAC 1-11 back to 250kts
           CAS, the next pass-behind was generated at 1429:20, the FMS predicting that the
           ATTAS would arrive at P2 at 1431:38, 38 seconds ahead of the BAC 1-11, indicating a
           separation of 3nm when the BAC 1-11 had reached P2. This pass-behind manoeuvre

Page 62 of 90                                                                QinetiQ/S&E/AVC/CR031041
         required an initial track change of 26º for the BAC 1-11, the deviation from the aircraft's
         originally planned track being 5.9nm. The time from triggering the trajectory prediction
         to activating this trajectory was 4 seconds again. While following this trajectory, the
         achieved minimum spacing of the BAC 1-11 from the ATTAS aircraft was 6.3nm.

         Once the BAC 1-11 had completed the turn at waypoint B7, the merge manoeuvre was
         calculated with the merge waypoint defined as MMP and the required spacing behind the
         ATTAS set to be 15nm. At 1446:13, when this new trajectory was predicted, the ATTAS
         was 31.6nm from the BAC 1-11 and had a ground speed of 295kts. The direct distance
         from the BAC 1-11 to the merge waypoint MMP was 67nm, while the ATTAS was
         currently 36.3nm from MMP and already on a track towards it. The trajectory for this
         merge behind manoeuvre was activated 6 seconds after the prediction process had
         originally been triggered. A new speed demand of 286kts was output by the FMS to the
         autopilot. This provided a reasonable closure rate for the BAC 1-11 relative to the
         ATTAS, although by the time the BAC 1-11 had reached the waypoint MMP, it was
         actually 14.2nm behind the ATTAS. Prior to MMP, the FMS speed demand changed to
         229kts CAS, which was intended to match the speed of the target aircraft, i.e. the
         ATTAS, which was actually flying at around 210kts CAS. The report on the flight on the
         20th March details why there were problems with the prediction of the speeds for merge
         manoeuvre. The predicted wind for this flight in Rome was 189º/8kts, while that actually
         encountered during execution of the merge manoeuvre was 187º/18kts. Since this was
         very much a tail wind for this part of the flight, it would have resulted in the ground
         speed of the BAC 1-11 being about 9kts greater than predicted. However, the difference
         between the predicted ground speed from the trajectory and the actual ground speed,
         when flying at 286kts CAS, was of the order of 44kts, so the problem of the conversion
         to true air speed was resulting in an error of around 35kts. In this case, a CAS demand of
         around 275kts may have been more suitable for meeting the set spacing distance of
         15nm.

         Having passed MMP, the CAS demand reverted to the standard cruise value of 250kts.
         The aircraft then departed the allocated trials area after passing waypoint A6 and the
         control was taken over by the pilot in order to follow ATC instructions for returning to
         Ciampino airport.




QinetiQ/S&E/AVC/CR031041                                                                 Page 63 of 90
4.4        26th March 2003

4.4.1      Sortie Objectives

           The plan for this sortie was the same as that for the previous day with the BAC 1-11
           following the QNQ6 route and the ATTAS acting as the target aircraft for two pass
           behind and one merge manoeuvre to be performed by the BAC 1-11. The FMS remained
           in the same state as for the previous flight, but with the latest meteorological forecast file
           having been loaded. The co-ordination between the two aircraft had worked well on the
           day before and therefore the timings were left unchanged.

4.4.2      MFMS Report

           Initialisation of the flight details in the FMS was carried out by the pilot and the BAC 1-
           11 took off at 0900, 10 minutes after the ATTAS. No problems were encountered after
           the FMS guidance demands were engaged through the autopilot as soon as practicable
           after take off. Towards the top of the climb, the interference that affected the GPS
           systems was experienced again and consequently the pilot selected the IRS as the
           primary position source for the FMS until the aircraft had cleared this region of
           interference. A couple of hundred feet before the BAC 1-11 reached the cruise altitude of
           FL210, the thrust demand from the FMS reduced to flight idle, causing the aircraft to
           level a fraction early and start to descend. The profile guidance demands were
           disengaged and the autopilot was used directly to complete the climb to FL210. The
           cause of this was that the FMS had detected that the conditions were correct for the
           transition to the cruise mode, for which there is no requirement for a thrust demand (only
           height and speed demands). On this occasion, the transition point determined by the FMS
           was prior to the autopilot starting its flare to the cruise level and therefore the autopilot
           continued to react to any changes in the FMS thrust demand. After the flight, an
           adjustment to the flare to level anticipation parameter used by the FMS ensured that this
           would not reoccur on any subsequent flights.

           The first pass-behind manoeuvre was performed as planned when the BAC 1-11 had
           passed waypoint B2, the pilot responding to the R/T instructions from the trials ATC
           station for a minimum permitted spacing of 6nm. The FMS trajectory was generated at
           0916:04 with the ATTAS predicted by the FMS to reach the intercept point P2 at
           0918:30, 53 seconds before the BAC 1-11, and resulting in a predicted separation of
           4.5nm at P1. This new trajectory required a track change of 16º for the BAC 1-11 and
           produced a maximum offset distance from its original route of 3.2nm. Activation of the
           new trajectory was selected by the pilot 3 seconds after the prediction was made. At the
           point of closest approach during the manoeuvre, the spacing between the BAC 1-11 and
           the ATTAS aircraft was 6.6nm.

           The interference that had been affecting the reception of the GPS signals for the SBAS
           equipment had disappeared earlier on during the pass-behind manoeuvre and so the
           SBAS was reselected as the primary navigation data source for the remainder of the
           flight. The second pass-behind also took place without any problems. At the time of the
           trajectory prediction (0925:29), the ATTAS was being predicted by the FMS to arrive at
           P2 at 0928:16, which was now 62 seconds prior to the BAC 1-11 reaching P2. However,
           with the ground speed of the BAC 1-11 being 65kts greater than that of the ATTAS at
Page 64 of 90                                                                QinetiQ/S&E/AVC/CR031041
         this point, the separation distance would have been down to 5nm by the time the BAC 1-
         11 was at P2. The trajectory to pass-behind the ATTAS required a track change of only
         7º and consequently, the BAC 1-11 would only deviate by 2.5nm to the left of its
         previously planned track. The time between prediction and activation of this trajectory by
         the pilot was again 4 seconds, while this time the spacing at the point of closest approach
         was 6.3nm.

         The conditions input for the merge behind manoeuvre were to be a distance of 15nm
         behind the ATTAS on reaching waypoint MMP. A trajectory was computed at 0943:24
         when the BAC 1-11 was on track towards waypoint B8. At this point, the ATTAS
         aircraft was 30.2nm from the BAC 1-11 with 37.2nm to run before reaching the merge
         waypoint MMP. Meanwhile, on a direct bearing, the BAC 1-11 was a distance of 66.5nm
         from MMP. Activation of the predicted trajectory was within 3 seconds of the prediction
         and the new speed demand from the FMS was 272kts. The BAC 1-11 reached the
         waypoint MMP at 0954:28, by which point it was 15.4nm behind the ATTAS (see Figure
         4.4-1). Prior to MMP, the FMS speed demand had reduced to 228kts CAS in order to
         match the predicted ground speed of the ATTAS. The subsequent determination of the
         problem with the conversion to true air speed with the forecast meteorological data
         showed that there was a combination of effects that resulted in the BAC 1-11 being close
         to the required spacing. However, as shown by the 228kts CAS demand, which should
         have been 210kts CAS to match the ATTAS, the predicted speed profile from the FMS
         was not deriving the correct values, although the sense of the speed changes was correct.




                           Figure 4.4-1: Spacing Distance during Merge Manoeuvre, 26th Mar. 03
         Profile guidance was disengaged at MMP in order to follow the height and speed
         clearances from Rome ATC. The pilot changed the STAR in the FMS to be appropriate
         to runway 15 rather than runway 33 and generated a new trajectory, which was activated
QinetiQ/S&E/AVC/CR031041                                                                 Page 65 of 90
           successfully. Lateral guidance was maintained until reaching Ostia (OST), whereupon
           ATC heading instructions required the disengagement of the FMS bank demands.




Page 66 of 90                                                        QinetiQ/S&E/AVC/CR031041
4.5      26th March 2003

4.5.1    Sortie Objectives

         A further trial flight was performed in the afternoon on the 26th March with the system
         remaining unchanged from the morning apart from including the latest forecast
         meteorological data and the modification to the FMS flare to level anticipation parameter
         (to overcome the problem seen on the morning flight). The planned sortie profile
         remained the same as that flown in the morning.

4.5.2    MFMS Report

         Initialisation of the FMS and trajectory prediction was carried out as per the earlier
         flights, the BAC 1-11 taking off approximately 10 minutes after the ATTAS aircraft.
         Lateral guidance demands from the FMS were engaged through the autopilot as soon
         after take off as possible, with the profile demands engaged as the aircraft passed about
         2000ft.

         During the climb, ATC restricted the BAC 1-11 to FL160 for a while before clearing it
         all the way to the cruise level of FL210. This resulted in the BAC 1-11 being partly
         delayed and was thus further behind the ATTAS than had been planned. The interference
         of the GPS signals was encountered in the same place again, so the IRS was selected as
         the primary navigation data source until completion of the first pass-behind manoeuvre.

         This first pass-behind was executed successfully, with the trajectory being predicted at
         1416:06, after the BAC 1-11 had completed the turn at waypoint B2. A minimum
         spacing distance was still being used (this being the value passed by the trials ATC
         controller), although the FMS predicted that the ATTAS was going to reach P1 66
         seconds (about 6.3nm) ahead of the BAC 1-11. The actual point of closest approach
         would be approximately when the BAC 1-11 reached P2, the separation at this point
         being down to around 5.6nm. Consequently, the pass-behind manoeuvre that was
         generated consisted of only a small offset, the initial track change being just over 9º with
         a maximum deviation distance of 1.9nm from the aircraft's previous track. The time
         between requesting prediction of the trajectory and its activation by the pilot was 5
         seconds, similar to the other flights. During the manoeuvre, the closest distance that was
         actually achieved was just under 6.5nm.

         As on the previous flight, the speed of the BAC 1-11 was adjusted, after having
         completed the first pass behind, in order to co-ordinate the time difference between the
         two aircraft at the next intercept point P2. The navigation data source was also changed
         back to SBAS for the second pass-behind manoeuvre. The minimum permitted spacing
         was again 6nm and, when the pass-behind trajectory was generated at 1424:35, the FMS
         predicted that the ATTAS would reach P2 at 1427:31, which was 47 seconds prior to the
         BAC 1-11. By the time the BAC 1-11 had reached P2, it was estimated that the two
         aircraft would be 3.9nm apart. Hence the resultant trajectory required a track change of
         8º, but this time the deviation distance from the original track was 3.9nm. It took less
         than 5 seconds from initiating the trajectory prediction to activating it and at the closest
         point of approach during the manoeuvre, the two aircraft were 6.4nm apart.


QinetiQ/S&E/AVC/CR031041                                                                  Page 67 of 90
           Generation of the merge manoeuvre took place at 1442:00, after the BAC 1-11 was
           established on track to waypoint B8, the intention being to achieve a spacing of 15nm
           behind the ATTAS by the time MMP had been reached. The ATTAS was currently
           30.4nm from the BAC 1-11 and 37.1nm from MMP. At the point when the trajectory for
           the merge manoeuvre was generated, the direct distance from the BAC 1-11 to the MMP
           waypoint was 66.7nm. The pilot quickly activated this trajectory (4 seconds after the
           generation had been initiated) and the new CAS demand derived from this trajectory was
           266kts. By the time (1453:32) that the BAC 1-11 arrived at the merge waypoint MMP, it
           was still approximately 18nm behind the ATTAS. This is partly because the actual
           ground speed of the BAC 1-11 was only 16kts greater than the predicted value for the
           section when it was flying at 266kts CAS. Therefore, although the BAC 1-11 reached
           MMP about 35 seconds earlier than predicted, this had not compensated sufficiently for
           the error in the predicted time for the ATTAS to fly the extra 15nm, the error in this term
           being of the order of 55 seconds. The difference of 20 seconds would account for about
           1.5nm in the spacing distance from the ATTAS. In the calculation of the merge
           manoeuvre for the BAC 1-11, the FMS estimated a time of about 1449:59 for arrival of
           the ATTAS at MMP and this proved to be very close to the actual arrival time of
           1449:57. Consequently, in this respect, the forward estimation of the target's movement
           was very accurate. The merge manoeuvre was regarded as complete once the BAC 1-11
           had passed MMP.




Page 68 of 90                                                              QinetiQ/S&E/AVC/CR031041
4.6      28th March 2003

4.6.1    Sortie Objectives

         This was the final trials flight to be carried out from Rome. For this flight, an update had
         been incorporated into the software to correct for the determination of time for the target
         aircraft to fly the additional spacing distance beyond the merge waypoint (previously it
         had been based on the target's derived CAS value rather than its ground speed). A new
         taxi map had also been received for testing. This was to overcome the displacement
         between the aircraft's position on the map and its actual location on the airport that had
         been seen with the previous version. No other changes had been implemented and the
         route remained the same as for the other trials flights.

4.6.2    MFMS Report

         With the latest meteorological forecast file loaded into the FMS, the FMS was initialised
         as normal and the trajectory predicted prior to departure from the stand. The taxi map
         now gave a more representative indication of the aircraft's position on the airport,
         although the aircraft was shown towards the edge of the taxiway rather than following
         the centre line.

         The BAC 1-11 was held by ATC prior to take-off due to landing traffic and therefore its
         actual departure time was about 13 minutes after the ATTAS aircraft, rather than the 10
         minutes that had been planned. In order to allow the BAC 1-11 to catch up, the ATTAS
         slowed down during the initial part of its cruise. Similarly, once the BAC1-11 had also
         reached the cruise altitude, its speed was increased to reduce the difference in the ETAs
         of the two aircraft at P1, the intercept point for the first pass-behind manoeuvre.
         Although this had been achieved to certain extent, when the trajectory was generated
         using a minimum spacing distance of 6nm, the FMS predicted that no lateral offset was
         required. At this stage, the FMS predicted that the ATTAS aircraft would reach P1 at
         0919:54 (actual time of arrival was 0919:54), around 98 seconds before the BAC 1-11.
         By the time the BAC 1-11 was expected to have passed P1, the ATTAS would be 7.9nm
         away. A spacing distance of 8nm was used instead and this time a lateral route change
         was predicted with a track deviation of 14º and lateral displacement of 3nm from the
         original route. Due to the delay in having to generate a second trajectory with this
         different spacing value, the BAC 1-11 was only 15nm from P1 at the time the trajectory
         was activated. During the manoeuvre itself, the two aircraft were 8.5nm apart at the
         closest point of approach.

         A further speed adjustment was carried out to try and further reduce the time difference
         at the next intercept waypoint P2. When the pass-behind trajectory was generated
         0927:16, the MFMS was predicting that the ATTAS aircraft would be at P2 by 0929:54
         (actual arrival time was 0929:53), while the ETA for the BAC 1-11 was some 62 seconds
         later. The estimated separation at the time when the BAC 1-11 was predicted to reach P2
         was 4.9nm. This meant that a trajectory could be generated for the pass-behind
         manoeuvre using the normal 6nm minimum spacing distance. With an initial track
         change of 8º, the BAC 1-11 was required to deviate by a cross-track distance of 3.3nm



QinetiQ/S&E/AVC/CR031041                                                                  Page 69 of 90
           from its original route. The actual minimum spacing from the ATTAS aircraft
           encountered by the BAC 1-11 during the manoeuvre was 6nm.




                Figure 4.6-1: Spacing of BAC 1-11 from ATTAS during Merge Behind, 28th Mar. 03
           With this software having been modified for determining the correct time at which the
           BAC 1-11 should arrive at the merge waypoint to achieve the required spacing (15nm
           again), a trajectory was generated once the BAC 1-11 was on track towards waypoint B8.
           This occurred at 0944:01, when the ATTAS was 29.1nm from the BAC 1-11 and 38.6nm
           from reaching MMP, the merge waypoint. The trajectory was activated within 4 seconds
           of the generation being triggered. When computing the conditions for the merge
           manoeuvre, the FMS determined that the ATTAS would pass MMP at 0951:54 (actually
           passed this waypoint at 0951:51) and consequently would take a further 184 seconds to
           cover the additional 15nm. The prediction for the BAC 1-11 to merge behind the ATTAS
           determined that the BAC 1-11 would reach MMP about 8 seconds ahead of the time
           when the spacing would be exactly 15nm. This 8 seconds would equate to just over
           0.6nm at the ground speed of the ATTAS aircraft. The predicted CAS for the manoeuvre
           was 300kts and the BAC 1-11 actually arrived at MMP approximately 46 seconds earlier
           than predicted. This was a direct result of the error in the true air speed calculation that
           manifested itself in an under-estimated ground speed value for the selected CAS.
           Consequently, on reaching the merge waypoint, the spacing of the BAC 1-11 behind the
           ATTAS aircraft was 11nm (see Figure 4.6-1), which can be related to the distance
           covered by the ATTAS in 46 seconds being of the order of 3.5nm. In order to have
           achieved a more appropriate ground speed for the manoeuvre, a CAS of between 280 and
           285kts would probably have been better suited.


Page 70 of 90                                                               QinetiQ/S&E/AVC/CR031041
         As encountered on the previous trials flights from Rome, interference had once again
         affected the SBAS and GBAS equipment towards the top of the climb to FL210. On this
         occasion, the FMS had detected the degradation in the SBAS performance status and it
         had automatically reverted to using the IRS as the position data source. After region of
         interference had been passed and the SBAS started to indicate that it had a reliable fix
         again, the FMS automatically re-selected it as the position source. This transition gave no
         problems to the FMS guidance function.




QinetiQ/S&E/AVC/CR031041                                                                 Page 71 of 90
4.7        28th March 2003, Sortie No. 774.

4.7.1      Sortie Objectives

           Airways transit from Ciampino back to Boscombe Down. As for the other transit flights,
           the MFMS was to be used to provide lateral guidance and, where possible, profile
           guidance as well. The system was in the same state as for the trials flight in the morning
           apart from updating the default cruise speed to be 320kts CAS. The MA-AFAS company
           route for this flight, ROMBOS, was selected by the pilot via the MCDU and the cruise
           altitude was set to be FL260. The VDL4 transponder was to be operated for monitoring
           purposes while the aircraft remained in the ROME FIR

           The flight planned route was:

           LIRA-OST-MEDAL-GILIO-ELBA-DOBIM-AKUTI-PIGOS-BARSO-OKTET-GIPNO-
           BULOL-ARDOL-CHABY-LAULY-BRY-CLM-KOPOR-ABUDA -GUBAR-GURLU-
           SAM-EGDM.

4.7.2      MFMS Report

           After departure, the aircraft was initially vectored by ATC until it was cleared to go
           direct to waypoint MEDAL. At this point, an in-flight generation was carried out and the
           resultant trajectory activated, allowing the lateral and profile guidance demands from the
           FMS to be engaged through the autopilot. This guidance was maintained for the majority
           of the remainder of the flight until the profile demands were disengaged prior to the top
           of descent. Lateral guidance was re-instated for periods after this, dependent on the
           clearances that were currently being issued by ATC for the recovery into Boscombe
           Down.

           The lateral guidance demands from the FMS were used for nearly 1.75 hours of the
           return flight to Boscombe Down. Over the course of its use, in terms of the flight
           technical error determined by the FMS, the mean cross-track deviation that was achieved
           was 84m with a standard deviation of 77m. The maximum magnitude of deviation was
           410m and this, as mentioned before, was principally due to the tight default turn radii
           (5nm) that the FMS was set to use. With the high ground speed of the aircraft when
           operating at 320kts CAS, the FMS was demanding the maximum available bank angle of
           25º in order to complete the various turns and this was not always sufficient to track the
           lateral path exactly. Similarly, the transitions into and out of the turns resulted in peaks in
           the cross-track deviation value caused by the need to apply an anticipation factor to the
           demand to cater for the roll response performance of the autopilot. Normally, the FMS
           would have been configured with a larger default turn radius for en-route waypoints and
           this would have reduced the peaks in the cross-track error. The use of the go-direct
           function could also result in a localised increase in the cross-track deviation until the new
           track had been captured.

           These values of cross-track error that were experienced while using the MFMS were
           essentially, however, very small and demonstrated an accurate capability for tracking a
           defined lateral path, suitable for the reduced values of RNP that are intended for future
           operations.

Page 72 of 90                                                                 QinetiQ/S&E/AVC/CR031041
4.8      Summary of Rome Results

         The Rome flight trials demonstrated that ASAS delegated manoeuvres could be
         performed between two live aircraft using the functionality developed for the MFMS and
         the aircraft state data received via the ADS-B reports using the VDL4 communications
         system. When the two aircraft arrived in Rome at the start of the trials week, this was the
         first time that a full check of the ADS-B report data could be carried out using another
         actual aircraft as the data source. The air-to-air broadcast provided by the VDL4 system
         proved to be robust and neither the transponders nor the MFMS Communications
         Management Unit (CMU) needed to be reset during any of the flights. With the VDL4
         system operating successfully for Air to Air broadcast data throughout these airborne
         trials, the ATTAS was therefore able to act as the target aircraft about which the BAC 1-
         11 performed the ASAS manoeuvres. This required the VDL4 transponder to be
         connected to the MFMS for all these flights. This configuration restricted the type of data
         logging that could be achieved on the aircraft and therefore it was not possible to directly
         determine performance figures for the data link.

         MFMS performance would be dependent on the integrity of the data that it received from
         the ADS-B reports and, once a scaling factor for the ground speed data had been revised,
         this proved to be suitable for the requirements of the ASAS manoeuvres. With the
         occasional report being filtered out that contained values of zero for ground speed or
         track angle, the frequency of valid updates being received by the MFMS on the BAC 1-
         11 from the ATTAS aircraft proved high enough for these trials. For each manoeuvre
         that was performed, the MFMS computed a trajectory without failure and the ADS-B
         data available was typically less than 3 seconds old (the update rate of the ADS-B reports
         being around 4 seconds). This also supported the requirements of the Cockpit Display of
         Traffic Information (CDTI), allowing the execution of the manoeuvre to be monitored by
         the pilot via the ND. The Ciampino ground station had been able to monitor the ADS-B
         reports that were received from the two aircraft as well.

         The MFMS was operated successfully throughout the trials without any system failures
         occurring during the flights. It was able to predict trajectories from take-off to the final
         approach point and provided guidance to follow the Standard Instrument Departure (SID)
         Ostia 5A once the aircraft was airborne. With both aircraft operating at the same cruise
         flight level, two pass behind and one merge manoeuvre were tested on each of the five
         flights that were undertaken. Unlike the flight trials at Boscombe Down, where the
         simulated target aircraft had comparable ground speeds to the BAC 1-11, for the trials in
         Rome, the ATTAS aircraft was cruising with a CAS 40kts below that of the BAC 1-11.
         This meant that there was a significant variance in the ground speeds of the two aircraft
         for these manoeuvres.

         The pilots were able to readily enter the relevant data into the MCDU for the pass behind
         manoeuvres before generating and executing a trajectory via the MFMS, this process
         being generally comparable to setting up an FMS for more standard types of operations.
         The required data was provided over R/T by the trials air traffic controller. The pilots and
         controllers noted, however, that some of the phraseology that was used might need to be
         revised to improve the clarity and efficiency in the issuing of certain instructions. There
         was also a reasonable confidence from the pilots that the MFMS was producing sensible
         trajectories for these manoeuvres due to the fact that the typical time between selecting
QinetiQ/S&E/AVC/CR031041                                                                  Page 73 of 90
           prediction of a trajectory and the activation of it was around 5 seconds. The estimation
           for the duration of the prediction process itself is of the order of 1 to 1.5 seconds.

           Apart from the 1st pass behind manoeuvre on the flight on the 28th March, which used
           an 8nm minimum spacing distance, all the others employed a 6nm minimum spacing,
           with the MFMS adding a 0.25nm tolerance to this in its trajectory computation. In none
           of these situations was the minimum specified spacing distance ever compromised at the
           closest point of approach (CPA) between the two aircraft. For the seven pass behind
           manoeuvres using a 6nm minimum spacing, the mean value at the CPA was 6.3nm with
           a standard deviation in the results of 0.2nm. On the lone occasion that an 8nm minimum
           spacing was used, the actual value at the CPA was 8.5nm. On average the BAC 1-11 was
           about 3.3 minutes from reaching the intercept point between the tracks of the two aircraft
           when the pass behind manoeuvre was predicted. Therefore, given the relative positions
           and speeds of the two aircraft, the minimum specified spacing distance would likely have
           been compromised approximately 3 minutes ahead without any action being taken.

           There is insufficient data to make too many significant conclusions on the type of lateral
           deviation required to resolve the conflict. Various factors are involved, but generally, the
           second pass behind on each flight required a slightly greater deviation from the original
           route due to the ATTAS flying a track with a shallower cut angle relative to that of the
           BAC 1-11. However, as would be expected, the difference between their predicted times
           of arrival at the track intercept point was also important in terms of the magnitude of the
           lateral deviation. Along with this term, the remaining time before the BAC 1-11 reached
           the intercept point would tend to determine the change in track angle that was required.
           For the predicted conflict conditions experienced between the two aircraft in these trials,
           the resultant pass behind manoeuvres required the BAC 1-11 to perform an initial track
           change of no more than 27° and in the majority of cases it was within 15°. Typically, the
           lateral deviation was within the normal ±5nm width of a current airway with all but one
           of the pass behind manoeuvres using a spacing of 6nm. However, the ATTAS was
           always ahead of the BAC 1-11 in time and in many cases it was predicted to be of the
           order of 1 minute ahead at the intercept point in the tracks of the two aircraft, hence
           limiting the magnitude of the lateral deviation..

           Similar to the final trial flight at Boscombe Down (on the 20th March), the merge
           manoeuvre that was tested at Rome demonstrated the concept and application of this
           function, but was lacking the complete software modifications to allow it to operate to
           the required precision. The MFMS was always able to compute a trajectory to meet the
           timing constraints determined by the manoeuvre generator function, but as already
           mentioned, the conversion of computed air speed to true air speed was under-estimating
           this value. Hence, the demanded CAS typically resulted in a ground speed that was too
           high for required merge manoeuvre. With the ATTAS flying at a constant CAS of 210kts
           and a fixed track towards the merge waypoint, the results indicate that the MFMS on the
           BAC 1-11 was making a reasonably accurate estimate for the arrival time of the ATTAS
           at this waypoint. With a fairly stable wind system over the period of the manoeuvre, the
           estimate could be within 2 or 3 seconds of the aircraft's actual ETA. In most cases, the
           ATTAS was 35 to 40nm from the merge waypoint and the spacing distance would
           normally be 15nm. For the ground speed of the ATTAS, this would equate to about 11
           minutes of flying time. Meanwhile the BAC 1-11 would be around 67nm from the merge
           waypoint and consequently was aiming to close the distance to the ATTAS by
Page 74 of 90                                                               QinetiQ/S&E/AVC/CR031041
         approximately 15nm in this time frame. The MFMS was therefore aiming for a ground
         speed for the BAC 1-11 that was of the order of 80kts greater than that of the ATTAS
         aircraft. With the ATTAS flying at 210kts CAS, then to achieve this closure rate required
         a speed demand for the BAC 1-11 from the MFMS that was typically around 270kts
         CAS.

         During the course of these flights, the application of these pass behind and merge behind
         manoeuvres were generally regarded as relatively simple procedures by the pilots and
         controllers. Having become familiar with the method of operation involved with these
         manoeuvres, the overall process placed no undue pressure on the participants and it was
         typically considered in a similar context to any standard ATC instruction. The execution
         of the manoeuvre was also fairly benign from the user’s perspective.

         During the course of the trials, it was found that there were regions of interference in
         which GPS equipment onboard the aircraft were being affected. This was important to
         the MFMS because it was configured to use the SBAS as its primary source of position
         and ground velocity data. This interference was located between waypoints LUNAK and
         VALMA and in the vicinity of waypoint B3, this latter situation influencing the aircraft
         between not only B3 and B4, but also as it approached MMP. The impact of this
         interference was to reduce the Signal-to-Noise Ratio (SNR) for the GPS satellites and to
         cause tracking of the geostationary (GEO) satellite to be lost briefly. Once this
         interference had been encountered on one flight, the tendency was to manually select the
         IRS as the primary position source for the MFMS on later flights whenever the SNR
         figures started to fall. It was noted, however, that the MFMS was capable of switching
         automatically to the IRS if there was a loss of valid data from the SBAS. Additionally,
         the results show a drop in the SNR values of about 8dBHz soon after 0900 UTC on each
         day, although the exact reason for this is unknown.




QinetiQ/S&E/AVC/CR031041                                                               Page 75 of 90
5          Overall Summary of MA-AFAS Flight Trials

The ASAS delegated manoeuvres were based on single-shot predictions using current state data
obtained for the target aircraft via either real ADS-B reports from the VDL4 data link or from
simulated ADS-B reports generated within the MFMS itself. For the conditions under which these
flight trials took place, this aircraft state data proved sufficient for the MFMS to predict the future
flight path of the target over the time period required for the manoeuvre. This was probably more
true for the pass behind manoeuvre, which was typically looking at the resolution of a possible
conflict 3 or 4 minutes ahead, compared with the longer term (around 10 minutes) needed to
complete the merge manoeuvre.

The flight trials with the two live aircraft revealed that knowledge of the current state data of the
other aircraft was sufficient to produce a short-term prediction for that aircraft that could be
accurate to within about 3 seconds. Clearly, the wind conditions during the Rome trials were fairly
stable which assisted this process. If conditions were more variable, then the single-shot approach to
the prediction would need to be backed up by a monitoring process and modification of the
trajectory values if it is determined that the specified spacing will be compromised. This process
had been designed into the MFMS, but it had not reached a sufficient level of development that it
could be used during the flight trials themselves.

The flight trials at Boscombe Down helped to develop the MFMS and prove the functionality of the
system in flight. These trials were all performed with simulated target aircraft and, as such, allowed
a systematic approach to expand the capability of the MFMS on the aircraft itself. With various
updates to the system being received during the period of the trials at Boscombe Down, there was
never really a fully stable configuration that could be assessed over a series of flights. The flights
did demonstrate, however, that the implementation of the lateral pass behind could achieve the
required spacing within the normal guidance limitations used by the FMS. A similar situation
existed with the development of the merge manoeuvre, although this encountered a few more
problems related to the derivation of the required speed demand to achieve the defined spacing.

With the Rome flight trials occurring directly at the end of the period of testing at Boscombe Down,
the MFMS was essentially in the same configuration used for the final Boscombe flight. There were
therefore known limitations in the performance, notably with regard to the merge manoeuvre
needing further refinement. Since, when the aircraft arrived at Rome, this was the first time that the
air-to-air broadcast of the VDL4 data link could be tested for real, it was satisfying to find that this
was working reliably. With the MFMS status remaining essentially unchanged during the Rome
trials, it could be demonstrated that the prediction and execution of these delegated ASAS
manoeuvres was repeatable for the same situation on each flight.

The HMI that had been developed to support these ASAS manoeuvres worked as principally
intended and did not prove burdensome to operate for the pilot. For these flight trials, the pilot had
to enter the conditional data for the manoeuvres via the MCDU, this data being specified by the
controller over R/T. The use of data link to automatically supply this data to the MFMS could not
be tested and although this might reduce the work required of the pilot, it may also raise new issues
concerning the display of information for the pilot's situation awareness. There were areas where
the HMI could be improved. For the pass behind manoeuvre, since this was solely a lateral change,
then the graphical display on the ND was generally sufficient to view the revised trajectory and
Page 76 of 90                                                                QinetiQ/S&E/AVC/CR031041
monitor progress. It may have been useful to also have a readout of the predicted minimum spacing
between the two aircraft, which could be continually updated. This would help, along with the ring
drawn around the target aircraft, to confirm that the system was achieving the required spacing. The
merge manoeuvre involved not only a lateral modification, but a speed change as well. The pilots
were interested in having the updated air speed being more predominantly displayed when the
trajectory had been predicted so that it was not necessary to access other MCDU pages to locate it.
This would provide confidence that the FMS was not going to demand an excess speed to meet the
spacing requirements (the MFMS did contain speed limits for each flight phase and so it would not
be expected to exceed these). Another area that was commented upon concerned the feedback from
the system if it could not meet the specified requirements in order to be able to clearly identify
where any limitation might exist. This actually requires quite a bit of development to perfect and
was not a key issue within the project at this stage.

When flown on the BAC 1-11, the MFMS was typically used with the SBAS equipment as the
primary position source, although for the flights at Boscombe Down, the GBAS was used in the
SID and the STAR. The SBAS itself was capable of providing a lateral fix to a mean accuracy of
1.3m with a standard deviation of 0.7m, while the vertical fix had a mean of -0.9m and a standard
deviation of 2.5m. These accuracy figures were determined relative to a GPS truth track system that
can be considered to have an accuracy of 0.5m. The typical flight technical error for the lateral
guidance that was achieved by the MFMS during the flights on the BAC 1-11 was of the order 90 to
100m with a standard deviation that was just over 100m. These levels of performance would
support the type of accuracy required in the execution of the ASAS manoeuvres and any
implementation of RNP 1 for en-route navigation.




QinetiQ/S&E/AVC/CR031041                                                                 Page 77 of 90
6          Conclusions and Recommendations

The MA-AFAS flight trials with the QinetiQ BAC 1-11 were able to demonstrate feasibility of
incorporating into a modern FMS the functionality for performing delegated ASAS manoeuvres.
This was achieved not only with simulations of other aircraft within the same airspace as the BAC
1-11, but also with another live aircraft (DLR's ATTAS) during the trials in Rome. The ASAS
manoeuvres that were flown consisted of a lateral pass behind and a merge behind with distance
and, during the Rome trials, it was shown that these had a consistent performance and could be
applied successfully.

The approach taken to the development and testing of the MFMS functionality leading to the Rome
trials was also proved to be a successful method in terms of the efficient use of the flights. By using
accurate ground-based simulation to determine the expected overall system performance prior to
each flight meant that no flights were actually aborted due significant system problems being
encountered while airborne. It also allowed confidence in the behaviour of the system to be built up
in the limited time that became available for completing these flights. This level of confidence that
was achieved in the expected performance of the MFMS was also vital in ensuring that all of the
participants in the Rome trials were satisfied with operating these two live aircraft at the same
altitude and on conflicting tracks. This was especially so, since the first trials flight in Rome was
also the first time that the complete MA-AFAS system environment had been operated as a
complete entity.

An additional key component to the success of these flight trials was the implementation of ADS-B
via VDL4. This proved capable of satisfying the air-to-air broadcast requirements for computing
these ASAS manoeuvres by providing regular position and velocity reports for the two aircraft.
There were still issues to be resolved concerning the data communication between the VDL4
transponder and the MFMS. However, the proportion of valid position reports received by the
MFMS was sufficiently high to not cause any disruption to the manoeuvre generation process and
to allow the pilot to monitor progress via the CDTI option on the ND.

Although the use of the current aircraft state data proved adequate for these manoeuvres, it would
tend to limit the application to those situations where the motion of the target was relatively
constant (track, ground speed and height rate) for the duration of the manoeuvre. This was probably
more reasonable for the pass behind manoeuvres in which the look-ahead time was of the order of 3
to 4 minutes, but for the merge behind, the duration of manoeuvre could be around 10 minutes.
Regular monitoring by the FMS of the relative progress of the two aircraft would therefore improve
the performance. Similarly, the inclusion of additional intent data in the ADS-B report would
increase the accuracy of the longer-term prediction by having greater knowledge of the target
aircraft's planned profile. Both of these capabilities had been incorporated within MA-AFAS, but
neither was actually operational during these particular trials, although developments continued to
increase the availability of this functionality for DLR's later flight trials (see [4]).

The operation of the systems provided to the pilot and the controller, to support the execution of
these ASAS manoeuvres, did not prove to be a significant distraction from their normal tasks. The
trials indicated that the pilots and controllers found the application of the pass behind and merge
behind functions to be relatively simple to perform, similar in respect to normal ATC procedures.
Aspects of the HMI still need to be improved, though, before complete user acceptability is
Page 78 of 90                                                               QinetiQ/S&E/AVC/CR031041
achieved. The implication, however, is that the integration of such procedures into the operational
environment should not require a major transition. There is further work to be done to resolve the
responsibility and course of action for the various failure states when executing these types of
ASAS manoeuvres, but this would be typical of any new procedure.

The following recommendations are made to further the work and validation of the MA-AFAS
system:
• Further development of the ASAS manoeuvres in terms of types and efficiency – the project
    proposed ASAS Lateral, Vertical and Longitudinal manoeuvres. In each of these categories, a
    variety of executions were designed. In the event, due to development time, only Pass Behind
    and Merge could be flight tested in this phase. The further manoeuvres should also be tested and
    verified to provide a complete capability (part of this was achieved in the flight trials at DLR,
    see [4]).
• The MFMS used an instantaneous interrogation of the target state vector to determine the extent
    of the ASAS manoeuvre – this was found to work for the conditions experienced during the
    trials and proved effective against an aircraft not fully equipped. However, to provide a more
    effective, reliable and flexible approach to obtaining the target information, aircraft intent data
    should be included in the data link messages.
• Further flight testing of the VDL4 sub-system to verify air/ground operations – the VDL4
    system was delivered to the flight test phase in a state where the Air/Ground Point to Point link
    was unproven. Considerable effort was expended during ground testing to prove the capability
    but this was not capable of being used during these trials. Further work is required before this
    element of the communications chain can be implemented in an efficient manner.
• Further improvements to the HMI – although the HMI was considered, by the pilots, as being
    adequate for the job a number of improvement areas were highlighted. For instance in the area
    of required speed to perform the Merge manoeuvre. At present, there is no immediate indication
    to the pilot of the revised speed profile for the Merge manoeuvre and of the predicted spacing at
    the completion of the Merge.
• The phraseology used for the ASAS instructions needs to be revised to make them as clear and
    concise as possible and avoiding unnecessary repetition.




QinetiQ/S&E/AVC/CR031041                                                                    Page 79 of 90
7          References


           1.   The More Autonomous – Aircraft in the Future Air Traffic Management System
                Simulation and Flight Test Plan – D32 of 29 Nov 2000. Report Number
                QinetiQ/FST/TR025809-D32.
           2.   Aeronautical Co-ordination Notice Activity No. 2003-01-0011 of 23 December
                2002. Aeronautical Utilisation Services, CAA, Kingsway, London, UK.
           3.   The More Autonomous – Aircraft in the Future Air Traffic Management System
                Air/Ground Validation Report – D37 of 30 June 2003. Report Number
                QINETIQ/FST/CR032536-D37.
           4.   MA-AFAS D39 Annex A - Taxi trials and Flight Test Report from Braunschweig.
                DLR, Germany, June 2003
           5.   MA-AFAS D39 Annex B - Stanford Plot Evaluation of Ground and Space Based
                Augmentation Systems Flight Trial Data. NATS, UK, June 2003
           6.   D40: Results of Pilot-in-the-Loop Simulator Trials for ASAS Spacing. NLR, The
                Netherlands, June 2003




Page 80 of 90                                                        QinetiQ/S&E/AVC/CR031041
8        List of abbreviations

A/A          Air to Air
ADS-B        Automatic Dependant Surveillance – Broadcast
ADS-C        Automatic Dependant Surveillance – Contract
A/G          Air to Ground
AGP          AOC Ground Platform
AOC          Airline Operations Control
A/P          AutoPilot
ASAS         Airborne Separation Assurance System
ATTAS        Advanced Technologies Testing Aircraft System
ATC          Air Traffic Control
ATM          Air Traffic Management
ATN          Aeronautical Tele-communications Network
AvP          Avionics Package
CAS          Computed Air Speed
CCD          Cursor Control Device
CDTI         Cockpit Display of Traffic Information
CM           Context Management
CMU          Communications Management Unit
CNS          Communication Navigation Surveillance
CPDLC        Controller Pilot Data Link Communication
DADC         Digital Air Data Computer
EFIS         Electronic Flight Instrumentation System
EIU          Engine Instrumentation Unit
FIR          Flight Information Region
FIS-B        Flight Information System – Broadcast
FMS          Flight Management System
GBAS         Ground-Based Augmentation System
HMI          Human-Machine Interface
ICCS         Integrated Civil Cockpit Simulator
IHTP         In-House Test Platform
ILS          Instrument Landing System
IP           Internet Protocol
IRS          Inertial Reference System
ISDN         Integrated Services Digital Network
LCD          Liquid Crystal Display
MA-AFAS      More Autonomous Aircraft in the Future Air traffic management System
MCDU         Multi-function Control and Display Unit
METAR        Meteorological Aerodrome Report
MFMS         MA-AFAS Flight Management System
MTA          Managed Terminal Area
ND           Navigation Display
OOOI         Out, Off, On, In
PAD          Precision Approach and Departure
RFS          Research Flight Simulator
R/T          Radio Telephony
QinetiQ/S&E/AVC/CR031041                                                            Page 81 of 90
RTAVS           Real Time All Vehicle Simulator
SBAS            Space-Based Augmentation System
SID             Standard Instrument Departure
SIGMET          Significant Meteorological Report
SMGCS           Surface Movement Guidance Control System
STAR            Standard Arrival
TAF             Terminal Area Forecast
TCAS            Traffic alert and Collision Avoidance System
TIS-B           Traffic Information System – Broadcast
UP              User Platform
UTC             Universal Time Co-ordinated
VDLM4           VHF Data Link Mode 4
WP              Work Package




Page 82 of 90                                                  QinetiQ/S&E/AVC/CR031041
A         Appendix

A.1       Original QNQ1 UK Trials Route



Waypoint list:
Position or          Lat/Long
Waypoint             (WGS84)
S/S                  N5058.0 w001 47.0
A                    N5053.0 W02 55.0
B                    N51 05.0 W003 00.0
C                    N51 23.0 W003 00.0
D                    N51 23.0 W002 28.0
E                    N51 13.0 W002 12.0
F                    N50 45.0 W 003 10.0
G                    N50 47.0 W 003 47.0
H                    N50 26.0 W005 02.0
I                    N50 45.0 W005 38.0
J                    N51 23.0 W005 08.0
K                    N51 25.0 W003 32.0
C                    N51 23.0 W003 00.0
A                    N50 53.0 W002 55.0
S/S                  N5058.0 W001 47.0




QinetiQ/S&E/AVC/CR031041                   Page 83 of 90
                        BOSCOMBE MA-AFAS TRIAL - ROUTE                                     ANNEX A TO
                                                                                           ACN 03-01-0011
                                                                                           DATED 23 DEC 02


                                                                                         Route Order:
                                                                                         BD - S/S - A - B - C - D
                                                                                         -E-*-F-G-*-H-I-
                                                                                         * - J - * - K - C - A - S/S
                                                                                         - BD

                    J
                                                                                         *‘Avoidance
                                                                                                Manoeuvring

                                                          K

                                                                               C
                                                                                       D


 I
                                                                                            E
                                                                       B

                                                                                                                  BD
                                        G                                  A

                                                                                                            S/S
                H                                             F

                                                                                   Trial Route
                                                                                   Bi-directional Portion
                                                                                   Manoeuvring Area
Page 84 of 90                               QinetiQ/S&E/AVC/CR031041
A.2       UK route QNQ1 (revised)

Waypoint list:
Position or         Lat/Long               Position or   Lat/Long
Waypoint            (WGS84)                Waypoint      (WGS84)
EGDM                N51 09.13 W001 44.84   PPP           N51 13.0 W003 38.0
ESPIN               N51 07.3 W00148.2      QQQ           N50 24.5 W004 22.0
WOLF                N51 04.4 W001 47.1     RRR           N50 25.0 W004 39.5
INGL                N51 03.2 W001 42.3     HHH           N50 26.00 W005 01.80
KATE                N50 58.6 W001 43.1     SSS           N50 42.5 W005 11.0
XXX                 N50 53.63 W002 50.39   TTT           N51 20.0 W004 16.0
BBB                 N51 04.90 W003 00.00   ZZZ           N51 24.39 W003 48.26
YYY                 N51 23.25 W002 55.21   KKK           N51 24.80 W003 32.80
DDD                 N51 23.00 W002 28.30   CCC           N51 23.30 W003 00.00
EEE                 N51 12.50 W002 11.90   AAA           N50 53.30 W002 55.10
FFF                 N50 44.60 W003 10.00   AGIBS         N50 55.00 W002 20.00
GGG                 N50 46.60 W003 46.80   DM005         N50 59.87 W001 42.75
HHH                 N50 26.00 W005 01.80   DM001         N51 06.17 W001 30.74
MMM                 N50 50.0 W005 13.0     DM002         N51 13.10 W001 29.09
JJJ                 N51 23.10 W005 07.80   BD10E         N51 16.21 W001 32.30
ZZZ                 N51 24.39 W003 48.26   EGDM          N5109.13 W001 44.84




QinetiQ/S&E/AVC/CR031041                                                       Page 85 of 90
                                MA-AFAS ROUTE REVISED 26/02/03                                          Route Order:
                                                                                                        EGDM-SID-KATE-X-
                                                                                                        B-Y-D-E*F- G*H-
                                                                                                        M*J*Z-P*Q*R-H-
                                                                                                        S*T*Z-K-C-A-AGIBS-
                                                                                                        DM005-STAR-EGDM

                                                                                                        *       ASAS
                                                                                                                Manoeuvring

                            J

                                                                 Z
                                                                       K
                                                T

                                                                                         C   Y
                                                              P                                         D


I                   M
                                                                                                            E
                                                                                     B
                                                                                                                           EGDM
                        S                                                                                                         DM
                                                      G                                                                           002
                                                                                     A                                   KATE
                                                                                         X
                                                                                                 AGIBS
                    H       R                                              F                      Trial Route
                                   Q
                                                                                                  MERGE Route
                                                                                                  SID            STAR
                                                                                                  Old route not in use
    Page 86 of 90                                         QinetiQ/S&E/AVC/CR031041
                                                                                                  Manoeuvring Area
A.3      Rome Route QNQ6

AREA 1 boundaries
 Position or Waypoint       Lat/Long   (WGS84)                  Remarks
          A1A              N41 25.48   E011 21.50      OST 239/45, TAQ 198/50
          A1B              N41 06.54   E010 16.56   OST 24/97, TAQ 224/93
         A1B1              N40 44.21   E011 27.12   PNZ 260/69, OST 208/73
          A1C              N39 46.28   E012 00.26   PNZ 221/81, SOR 245/117
          A1D              N40 05.32   E013 04.46   CAR 242/65, PNZ 172/49
MA-AFAS QNQ6 Route Waypoints
 Position or Waypoint      Lat/Long (WGS84)                     Remarks
         OST               N41 48.2 E012 14.3
        LUNAK              N41 42.2 E011 52.3
        VALMA              N41 34.6 E011 25.3
           B2              N41 20.0 E011 26.0
           P1              N41 00.5 E011 29.0
           B3              N40 40.0 E011 31.0
           B4              N40 31.0 E011 36.0
           P2              N40 28.0 E012 10.5
           B5              N40 26.0 E012 37.0
           B6              N40 02.5 E012 43.0
           B7              N39 51.0 E012 02.5
           B8              N40 42.0 E011 21.5       Will be bypassed during execution
                                                    of merge behind manoeuvre.
         MMP                N40 57.0 E011 24.5
          A6                N41 17.5 E011 28.5
        ESINO               N41 23.1 E011 47.7
        TORLI               N41 35.8 E012 01.1
         OST                N41 48.2 E012 14.3



QinetiQ/S&E/AVC/CR031041                                                                Page 87 of 90
                                           BAC 1-11 route
                                           ATTAS route
                                           Merge manoeuvre legs never flown
                                           Both aircraft after merge
                                           Cleared area boundary




Page 88 of 90   QinetiQ/S&E/AVC/CR031041
      Report documentation page


       1. Originator's report number:

       2. Originator's Name and Location:                     I Mansfeld 414/1 MoD Boscombe Down


       3. MOD Contract number and period covered:             GRD1-2000-022840 months

       4. MOD Sponsor's Name and Location:


       5. Report Classification and Caveats in use:           6. Date written:    Pagination:                 References:

                                                              jUNE 2003           ix + 90

       7a. Report Title:                                      FLIGHT TEST VALIDATION REPORT - D39



       7b. Translation / Conference details (if translation give foreign title / if part of conference then give conference
       particulars):



       7c. Title classification:

       8. Authors:                                            IMANSFELD

       9. Descriptors / Key words:                            TRIALS, AIR TRAFFIC MANAGEMENT,
                                                              FUTURE,

       10a. Abstract. (An abstract should aim to give an informative and concise summary of the report in up to 300 words).




       10b. Abstract classification:                                                FORM MEETS DRIC 1000 ISSUE 5




Page 89 of 90                                                                                   QinetiQ/S&E/AVC/CR031041
                           This page is intentionally blank




QinetiQ/S&E/AVC/CR031041                                      Page 90 of 90

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:10
posted:7/28/2012
language:
pages:100