1 The MFLAME project

Under co-ordination of SEXTANT, a team of European partners has been involved in two successive European Union projects, designed to study the feasibility of an airborne system capable of the remote detection of atmospheric hazards, especially wake vortices. The first project, dubbed FLAME (Future Laser Atmospheric Measurement Equipment), was completed at the end of 1995 ; the second, which started on May 1, 1996, has been completed in April 2000 and concerns a multifunction version of this system, called MFLAME (Multifunction Future Laser Atmospheric Measurement Equipment). These projects have been supported by the European Union within the scope of the Brite Euram III programme, as part of the fourth framework research programme.

The MFLAME consortium comprises two manufacturers of airborne equipment (SEXTANT and BAE SYSTEMS), a company specialised in engineering for civil aviation (SOFREAVIA), a laboratory involved in air flow simulation (CERFACS), an organisation for Lidar measurements (DLR), university laboratories for processing algorithms and LIDAR simulation (National University of Ireland, Galway), and for laser crystal research (University of Hamburg), an organisation specialised in laser and optics simulation (INESC) and a laser manufacturer (QUANTEL).

For requirement definition, demonstration on airport and future equipment aircraft installation, a “Users Club” was attached to the MFLAME project as “associated partner”. It includes airlines, aircraft manufacturers, airports and official authorities.

Wake turbulence generated by an aircraft may be a hazard for the following aircraft. To protect against this risk, regulations specify minimum separation between aircraft during both approach and landing. Work is now underway to overcome this restriction, or at least to reduce the required separation under certain conditions, which would increase traffic capacity during peak periods and reduce saturation at heavily used airports. In turn, this would enable postponing or avoiding, the construction of new airports or runways, thereby considerably decreasing costs and environmental impact.

One effective approach – ensuring the required level of safety - is to install systems onboard aircraft capable of remotely detecting wake turbulence. The FLAME and MFLAME projects have been designed to define an onboard LIDAR-based system capable of detecting wake turbulence.

In addition, an airborne remote sensing LIDAR Doppler anemometer offers the possibility of detecting other atmospheric hazards at long range, the most well-known of which is windshear.

The FLAME project has demonstrated the viability of a LIDAR wake vortex detection system, including its operational aspects. This includes the development and demonstration of the core technologies and techniques for wake detection, particularly the laser and signal/image processing technologies.

MFLAME project main results:

· Extension of the multifunction application area of the MFLAME system to include, in addition to wake vortex detection, windshear predictive detection, even in dry air, and to assess its predictive detection capabilities in areas such as clear air turbulence, volcanic ash, gust alleviation, mountain rotors and hail.

· Realisation of a demonstrator, based on a 2 µm Lidar, with original tracking method and signal processing in accordance with MFLAME future on-board equipment requirements.

· Demonstration of wake vortex detection by ground tests on Toulouse airport, so as to enable measurements in a configuration very close to an on-board detection from a follower aircraft.

· Improvement of techniques and technologies for a future cost-effective multifunction airborne system (laser/optics, signal processing).

· Severity factor definition in the MFLAME measurement configuration.

MFLAME achievements validate the concept of on-board equipment for remote detection of wake-vortices, so as to guaranty the required level of safety during approaches at reduced separation distances.

An industrial equipment for detection, warning and avoidance of wake-vortex, windshear and other atmospheric hazards could be brought to market within 5 to 8 years.

2 MFLAME concept rational

The initial vortex strength and structure depend on aircraft weight, aircraft aerodynamic design and flight configuration.

The vortex decay depends on initial vortex strength and structure and also on meteorological conditions (wind and atmospheric turbulences).

Present rules for aircraft spacing minima (ICAO…) are defined to guarantee a sufficient decay of the vortices in all weather conditions. Nevertheless some incidents are still observed.

It has been defined in FLAME with end-users, that the objective for shorter separation distances/time in approach phase should be 1 minute (» 2.5 Nm). Under this value, other limitations on the ground arise.

This is possible, taking into account meteorological conditions. The wind acts on vortex decay and blows away the vortices, and the atmospheric turbulences act also on vortex decay.

At airport level, the decision to work in shorter separation approach mode is on ATM responsibility and linked with favourable and sufficiently stable meteo conditions (Probability of dangerous wake vortex on the path < 1.10^-3 per approach). ATM end-users consider that the meteo prediction validity must be more than 20 minutes.

On the ground, measurement means like anemometers, radar, lidar, sodar, etc… are necessary (see AVOSS trials at Dallas) to observe local meteo influence on wake-vortex evolution and to confirm the stability of meteo conditions.

Initially, these measurement means are used to characterise the meteorological environment of the airport.

From FLAME and MFLAME investigations, it is clear that end-users, especially pilots and aircraft manufacturers (AIRBUS), state as necessary an on-board detection system to guarantee the required safety level by remote detection of possible residual wake-vortex on all the approach path and to enable the pilot to exercise his responsibility through a close and autonomous decision loop.

From FLAME and other studies, significant economical and environmental impact of shorter approach separation distances is expected by increasing aircraft movements of 10 to 15 % on major airports. This will induce increase in airport revenue, potential fuel saving and reduced pressure to build new runways and new airports.

The detection with a 2 µm pulsed Lidar is now the most suitable choice. This technology is currently used in the USA for airborne demonstrators. FLAME and MFLAME projects confirm the feasibility of on-board equipment that must be available for transport aircraft application in 2005-2008.

3 MFLAME requirements for wake-vortex detection

3.1 Operational basic requirements

- Alert time

From FLAME investigations and confirmed by the User-Club of MFLAME, especially AIRBUS, two levels of alert can be communicated to the pilot :

. Caution 30 s before estimated hazard encounter (distance » 1,2 Nm).

. Warning 15 s before estimated hazard encounter (distance » 0,6 Nm).

These values are sufficient for anticipation of an avoidance manoeuvre in approach configuration because of the low speed of the aircraft and the passengers have belts fastened.

- Hazard location accuracy

The hazard location, relative to the aircraft trajectory, for warning and guidance functions, is elaborated with the following accuracy:

. Distance accuracy: ± 100 m

. Angular accuracy: ± 0.5°.

- System integrity

Taking into account the occurrence probability of a wake vortex on the path (1.10^-3 per approach) and the duration of the approach phase, the general integrity requirement for an avionics equipment of 1 x 10⁹ hours of flight, is respected with a probability of non detection by the equipment without any BITE = 10^-5.

- Atmospheric conditions

. Shorter separation distances are used by clear air conditions.

. From the aircraft to the maximum detection distance, the wind gradient can be ± 3.75 m/s on the horizontal cross direction and ± 1m/s on the vertical direction.

3.2 Main functional system requirements

Taking into account the operational requirements, the dimensions and speed variations of the vortices relatively to the aircraft, and the attitude variations of the aircraft, the most significant system requirements are as follows.

- Field of view (FOV)

12° in azimuth

3° in elevation.

- Detection range – range resolution

Detection range : 0.8 to 2.375 km

Range gate length : 75 m

Range resolution : 75 m

Number of range gates : 21

- Spatial resolution

Along a line of sight (LOS) the resolution is equivalent to the range gate length: 75 m.

In a spherical co-ordinate system, which centre is the Lidar position, the spatial resolution (from one LOS to the other) is 2.7 milliradians in both vertical plan and horizontal plan.

- Measurement accuracy/resolution

Resolution of spectral width and/or velocity: < ± 2m/s.

- Scanning period

The Field Of View is completely explored by the scanner within the scanning period of 3 s.

- Scanning method – Number of LOS

The scanning method depends on optical and mechanical feasibility. A sinusoidal scan pattern with fly-back is used to cover the Field of View with the required resolution and 2500 Lidar LOS are distributed on it.

- Principal Laser characteristics

. Pulse energy : 2 mJ

. Pulse repetition frequency : 800 Hz

. Pulse length : 0,5 µseconde

. Wave length : # 2 µm (in an atmospheric window).

4 Demonstrator transceiver and scanner unit.

4.1 Laser transceiver and telescope

The optical demonstrator used for the wake vortex detection is based on a lidar transceiver, which incorporates the laser and the optical heterodyning in one device. The transceiver was manufactured by CLR Photonics Inc. in Lafayette, Colorado (USA). The master and the slave lasers used are diode-pumped Tm:LuAG lasers. They operate on a slightly longer wavelength than Tm:YAG, which better matches a good atmospheric window. To complete the optical detection system, a beam-expansion telescope and scanning devices had to be integrated. The performance parameters are summarised in Table 4.1.

Laser wavelength 2022.54 nm (Tm:LuAG)

SO pulse energy 2.0 mJ

SO pulse length (t) 400 ± 40 ns (FWHM)

SO spectral width ~ 1.2 MHz (0.5/t à single frequency)

Pulse Repetition Rate (PRF) 500 Hz

Output beam diameter 3.7 mm

LO / SO frequency offset 105 ± 3 MHz

Output polarisation linear

Total power consumption ~ 600 W

Telescope: type off-axis

magnification 20 x

clear aperture 108 mm

Table 4.1: MFLAME demonstrator, transceiver and telescope parameters

In order to achieve good measurements, the emitted laser pulse has to be of a single frequency with as low bandwidth as possible. The bandwidth of a perfect pulse is 0.4 to 0.5 (depending on the shape of the pulse) times the inverse of the pulse length. The laser used for the MFLAME demonstrator emits such “transform limited” pulses

Figure 4.1 shows the optical set-up used for the MFLAME demonstrator. The laser beam from the transceiver is coupled into an external off-axis telescope with a beam expansion of 20 times and an aperture of 10.8 cm. This resulted in an output beam of ~ 7.5 cm diameter. The two steering mirrors between the laser transceiver and the telescope were used to align the beam exactly to the telescope axis. The telescope parameters are summarised in Table 4.1.

Figure 4.1 : Optical bench, top view

4.2 Scanning device

For the high-speed scanning required by MFLAME, it was necessary to develop special scanning devices. The MFLAME image consists of a scan pattern of 3° (vertical) by 12° (horizontal). In order to achieve a sinusoidal scan pattern of 75 vertical lines in 5 s, a fast vertical scanner was developed. It consists of two counter-rotating prisms of 11 cm clear aperture with a deflection angle of 0.75° each, resulting in a ± 1.5° vertical scan. The horizontal scanner consists of a plane mirror of 20 cm diameter with a variable well-defined scan speed up to 2°/s and a return speed of 6° in 0.5 s. The vertical alignment of the scan pattern was performed with a micrometer screw on the mirror scanner, while the horizontal alignment was controlled with a PC. Both scanners were manufactured by Treffer Maschinenbau, Bruneck (Italy) as a subcontractor to DLR. Table 4.2 summarises the scanner parameters.

MFLAME image

Horizontal Scan Range 12°

Horizontal scanning time 5 s

Fly-back 0.5 s

Vertical Scan Range 3°

Vertical Lines per Image 75

Laser Shots per Image 2500 complete Scan

Horizontal Scanner

Scan Angles: Start Angle from 0° to 360° Software controlled

Stop Angle from 0° to 360° Software controlled

Min. Scan Speed < 1° per Second Laser Beam

Max. Scan Speed > 4° per Second Laser Beam

Encoder 12 Bit

Pointing Accuracy 0.18° Laser Beam

Vertical Angle -20° to 20° Laser Beam, Manual Adjustment

Vertical Scanner

Scan Range 3° fixed

Max. Scan Speed 28 Lines per second

Min. Scan Speed < 2 Lines per second

Encoder 12 Bit

Pointing Accuracy > 0.001°

Table 4.2: MFLAME demonstrator, scanner parameters

4.3 Data Acquisition

The analogue signal is amplified with a 1-GHz, 30 to 70 dB amplifier and split using a high-speed power divider. This enables the operation of the MFLAME data acquisition unit (developed by Sextant) and a DLR data acquisition and quick look unit at the same time. The data acquisition and quick look unit developed by DLR enables the operators to analyse the system and lidar performance during the measurement campaigns. It consists of a dual-processor Pentium II 350 MHz PC with a Signatec digitising card with 500MHz sampling rate.

An electronic control box acquires the scanner positions and provides a unique “LOS identification number” (LOS = Line Of Sight) for each shot to both acquisition units via a 16-bit parallel link. A LOS identification table provided by DLR identifies the horizontal and vertical scan position for each LOS number.

4.4 Integration into Mobile Shelter

The optical bench with the lidar system is integrated into a mobile shelter provided BAe. Figure 4.2 shows a photograph of the shelter at the DLR. The window above the red wind-shielded beam output is used to enable the operator to observe the landing aircraft and allows video documentation. A diagram of the lidar integration into the shelter is shown in Figure 4.3.

Figure 4.2: Photograph of mobile MFLAME shelter

Figure 4.3: Diagram of lidar integration into mobile shelter

Figures 4.4 and 4.5 show photographs of the actual lidar set-up after the integration. In Figure 4.4, the laser is visible on the right site of the optical bench, while the telescope is in the black housing on the left. In front are the steering mirrors used to couple the laser beam into the telescope. On the far corner of the optical bench are parts of the vertical and the horizontal scanner visible. In the background is the rack with the electronic control box. The MFLAME data acquisition unit is below the electronic control box (hidden by the laser). Figure 4.5 shows the same set-up from a different view. In the foreground is the reverse sight of the laser transceiver. On the very right of the optical bench is the horizontal scanner, with the mount of the 20-cm mirror clearly visible. The vertical scanner is the grey box just to the left of the horizontal scanner.

Figure 4.4: Lidar integrated into mobile shelter

with the laser transceiver in the front and the telescope on the left

Figure 4.5: Lidar integrated into mobile shelter

with the reverse of the laser in front and the scanners on the right

5 Demonstrator Signal Processing and recording unit.

The signal Processing and Recording Unit (Figure 5.1) has been designed for ground and further flight tests.

This unit is linked to laser and interferometer for acquisition of atmospheric signal, associated with a reference signal, for each shot. Other links with the scanner enable the localisation of each line of sight in the field of view.

In case of airborne utilisation, links with the aircraft system allow attitude compensation for image computing.

A DAT tape recorder is used to transfer data files from the hard disk for possible post-processing.

The main difficulty for hardware structure and software has been the real time processing constraints for data acquisition, storage, and image presentation to the operator (Figure 5.2).

Figure 5.1: Signal processing and recording unit

Figure 5.2: Operator interface

6 Demonstrator Image processing algorithms and detection simulations.

In a Doppler lidar, only the component of the air velocity in the direction of propagation of the lidar beam causes a Doppler shift in the returned signal. Hence, it is not possible to directly detect the rotational velocity component of a wake vortex while viewing it axially. Consequently, detection of wake vortices generated by a leading aircraft with an on-board Doppler system relies on the presence of identifiable axial Doppler signatures. Such signatures may be axial velocity components within or near to the vortex core, or increased Doppler spectral width close to the vortex core due to turbulent effects.

In order to investigate the feasibility of axial detection, Large Eddy Simulations (LES) techniques were used to simulate the evolution of wake vortices in various turbulence conditions. The three dimensional flow fields from the LES were applied to a time domain simulation of a Doppler lidar system. Other inputs to the Doppler lidar simulation system are the lidar performance parameters, the atmospheric parameters, the scanning pattern and the measurement geometry. An overview of the simulation approach shown in experiments is given in Figure 6.1

The simulation experiments were used to define the parameters of Doppler lidar system capable of detecting wake vortices from an axial point of view, and then, to develop the processing algorithms which could be used with such a system. Simulation results indicated that lidar pulse length of about 400 ns and a scanning system with a transverse spatial resolution of better than 6 meters would be required in order to achieve remote axial detection out to 2000 meters.

A two stage processing approach was developed. The first stage, known as the signal processing stage, involves estimating the mean radial velocity and the spectral width at various ranges along each line of sight. The second stage, known as image processing, is applied to the results of signal processing and aims to eliminate signal processing noise while preserving wake vortex Doppler signatures in the range gate images.

Figure 6.1: Overview of Simulation Approach

7 MFLAME ground tests definition

7.1 Phases of tests

Three phases have been planed to progress with optics and signal processing from good wind measurements to useful detection of wake-vortices and other atmospheric hazards, to demonstrate the validity of an airborne detection system.

- PHASE 0: December 1998 at DLR/Oberpfaffenhofen

· Definition :

Optimised German ODIN system with existing scanner and data unit.

· Results :

Single-shot wind measurements with encouraging results up to ranges > 2 km.

· Observation :

The 2 µm ODIN system has a pulse length of 1 µs and a pulse repetition frequency of 100 Hz.
For wake-vortex measurements, the pulse length lead to too long range gates for an optimised detection and the PRF is too low for the expected scanning period.

For this reason DLR decided to buy a CTI transceiver and to put it at MFLAME’s disposal for the next test phases.

- PHASE 1 : 31 January 2000/24 February 2000 at Toulouse-Blagnac

· Definition:

. Complete MFLAME system: Signal processing, 2 µm pulsed transceiver and scanner.

. Comparison with DLR 10.6 µm Lidar.

. Help of sensors for atmospheric environment (Airport meteo and Sodar/Rass).

· Results:

. System optimisation and calibration.

. First raw data acquisitions for MFLAME signal processing evaluation.

. Raw data exploitation demonstrates clearly wake vortex detection capability of MFLAME.

- PHASE 2 : 06-17 March 2000 at Toulouse-Blagnac

· Definition :

. Improved MFLAME demonstrator

. Comparison with DLR 10.6 µm Lidar

. Help of sensors for atmospheric environment (airport meteo and Sodar/Rass).

· Results :

. Large collection of raw data for different aircraft types

. Complete MFLAME efficiency evaluation for different measurement conditions

. Multifunction capability assessment

7.2 Measurement method at Toulouse airport

The demonstrator has the same functional characteristics as required for the future MFLAME equipment. The only difference is the scanning period of 5 sec instead of 3 sec due to the laser PRF of 500 Hz.

It has been installed at the extremity of a runway on Toulouse airport so as to enable measurements in a configuration very close to an on-board detection from a follower aircraft (see Figures 7.1 and 7.2).

The observation direction is at the opposite of the on-board application but the measurement aspect angle is very similar.

Figure 7.1: Ground tests measurement method – side view

Figure 7.2: Ground tests measurement method – top view

7.3 Installation and instrumentation on Toulouse airport

The MFLAME demonstrator and the DLR 10.6 µm CW Lidar (LDA) have been installed at the extremity of runway 33 L on Toulouse airport (Figure 7.3).

In addition to the airport meteo means, a SODAR/RASS system was used to know wind speed and direction from 20m to 1000m altitude by steps of 15 or 30m with a time resolution of 15 minutes, and also to know the temperature with the same altitude and time resolution.

Figure 7.3: Location on Toulouse airport – Runway 33L

8 MFLAME Ground tests realisation and recordings on Toulouse airport

8.1 General conditions

At the end of ground tests phase 1 (after calibration) and during the two weeks of phase 2, favourable weather conditions (landing on the 33L runway most of the time) enabled the MFLAME team to make numerous measurements. 93 landing aircraft of various types have been observed and measurement data recorded on the MFLAME Signal Processing and Recording Unit.

The data files were transferred regularly on DAT tapes and sent to Galway University (NUIG) for post-processing (Figure 8.1).

Note: The processing methods used by NUIG remain compatible with real time exploitation in on-board equipment.

A specific signal-processing unit has been developed by DLR for optical unit development tests and calibration and was particularly useful at Toulouse for assessment of long range detection capabilities of MFLAME for multifunction applications (wind-shear).

The DLR LDA CW Lidar was of great interest for location and characterisation of the wake-vortices.

Figure 8.1: Ground tests recording organisation

8.2 Meteorological data

For each test, the following meteo data are available:

- Control tower wind
Average value : 340°/ 10 kts

- Visibility
Generally > 8 km

- Weather type (sunny, etc…)

- Wind profile from SODAR

- Temperature profile from RASS.

8.3 Tests configuration and calibration

The FOV axis elevation was generally 3°, but set at 2° for 12 tests and at 4° for 26 tests.

The FOV axis azimuth was always the same as the runway axis.

The measurement distance was 800 to 2350 m for 19 tests and 400 to 1950 m for all others.

Before each measurement sequence a test was done without any aircraft for wind and noise reference.

For each aircraft test, the elapsed time between the beginning of recording and the passing of the aircraft over MFLAME was noted. This time allows the evaluation of aircraft distance during the test.

8.4 Type of aircraft and test distribution

The observed aircraft have been divided into three categories:

- Light (<30t): FOKKER 100, BAC 146, EMBRAER 120, FALCON 20.

- Medium (30t to 100t): MD 80, MD 90, AIRBUS 319 and 320, BOEING 737, DC9.

- Heavy (>100t): AIRBUS 330 AND 340, BELUGA.

The distribution is as follows:

- Light: 21 tests

- Medium: 58 tests

- Heavy: 14 tests.

9 MFLAME Ground tests exploitation

9.1 Experimental set-up

The objective of the Ground tests Phases 1 and 2 was to demonstrate the feasibility of the MFLAME Demonstrator for wake-vortex detection under very small aspect angles, as it is the case for a forward-looking airborne sensor. In such way, the ground simulation of the airborne MFLAME system could be approached by the Ground tests.

Ground test Phase 1 from 31 January to 24 February 2000 was mainly dedicated to system installation and tests, to optimisation of measurement strategy, and to first wake-vortex measurements. During Ground test Phase 2 from 06 to 17 March 2000, many wake-vortex measurements could be achieved under favourable weather conditions.

The lidar instrumentation consisted of two complementary systems, the 2 µm pulsed lidar as MFLAME Demonstrator unit and the 10 µm cw Laser Doppler Anemometer (LDA). Optimum performance for wake-vortex detection could be expected from the synergy of the first time used 2 µm lidar and the well-established 10 µm system.

The lidar systems were installed below the glide slope of runway 33L approximately 530 m in front of the runway threshold. As sketched in Figure 9.1, the aircraft are flying towards the lidar systems passing the sensing area at an altitude of 50 - 80 m. In this configuration, the angle between the vortex axes and the lidar line-of-sight (LOS), the so-called aspect angle, is rather small - similar to the airborne case. The 2D sensing area of the LDA azimuth scan covers a 80° wide arc segment in a slightly inclined plane, whereas the 3D sensing volume of the Demonstrator field-of-view (FOV) covers a 3° x 12° wide box of 1575 m length.

Figure 9.1: Comparison of the sensing volumes covered by the LDA azimuth scan and the Demonstrator field-of-view.

9.2 Wake-vortex measurements using the Laser Doppler Anemometer

The DLR Laser Doppler Anemometer (LDA) is based on a cw CO₂ laser, a 30 cm diameter telescope and a flexible scanning device. It is well established for basic short-range (100 - 200 m) wake-vortex measurements. In the MFLAME Project, it has been used to support the measurements of the Demonstrator unit.

A step-by-step approach has been chosen to approach the MFLAME objective. At first, elevation scans perpendicular to the glide-slope direction have been carried out to measure in well-known configuration the radial velocity field of the vortices. For example, the signatures of a MD83 vortex are shown in Figure 9.2. The colour-coded intensity plot represents frequency spectra versus spectrum number, respectively LOS velocities versus time. Around 18 s after aircraft passage, the typical pattern of the radial velocity field can be observed, consisting of the velocity increase towards the vortex core near spectrum no. 1433, the change in velocity sign inside the core, the second maximum on the opposite side of the vortex core, and the velocity decrease outside the core. The lower part of Figure 9.2 shows, for example, the frequency spectrum no. 1439 with a LOS pointing through the vortex in core vicinity. It consists of several peaks whereby the outer edge of the last peak near channel 120 is representing the highest velocity for this LOS of more than 10 m/s.

Figure 9.2: LDA signatures of the radial velocity field of a MD83 wake vortex.

The next step was to point the laser beam parallel to the glide-slope direction at constant elevation angle (9°). In this way, the first signatures of the velocity field under small aspect angles could be acquired. That led to the final configuration for wake-vortex detection by the LDA system, sketched in Figure 9.1 as “LDA azimuth scan“: a continuous azimuth scan at small elevation angle resulting in slightly inclined sections through the vortex pair.

An example of wake-vortex detection by LDA azimuth scans is given in Figure 9.3 for an A330 vortex pair. The upper trace shows the beam orientation (LOS) during 27 continuous azimuth scans between 100° and 180° at constant elevation angle of 10° and constant range of 150 m. The lower part shows a frequency spectrum measured before the aircraft passage (indicated by the position of the vertical cursor in the intensity plot). It is characterised by a pronounced wind peak near channel 45, corresponding to a LOS velocity component of 3.8 m/s. The intensity plot shown in the middle of the figure consists of 5120 colour-coded frequency spectra measured during a period of 130 s with intervals of 26 ms. Before the aircraft passage at t = 0 s we observe a regular arc-shaped pattern of the wind peak which arises from the continuous oscillations of the LOS between the wind direction and the direction perpendicular to the wind. After aircraft passage, this regular pattern is overlaid by small-scale velocity components arising from the vortex pair descending through the LDA sensing plane. From that, it takes more then 75 s until the regular signatures of the undisturbed wind field are appearing again.

Figure 9.3: LDA measurement of an A330 vortex pair.

- Upper trace: beam orientation (LOS) during 27 continuous azimuth scans between 100° and 180° at constant elevation angle of 10° and constant range of 150 m.

- Middle: intensity plot consisting of 5.120 colour-coded frequency spectra measured during a period of 130 s with intervals of 26 ms.

- Lower part: frequency spectrum measured before aircraft passage.

A closer look to the A330 signatures from Figure 9.3 is given in Figures 9.4 and 9.5, by ten expanded sections of azimuth scans each reaching from 100° to 180°. They are marked in red colour in the upper trace of Figure 9.3. The vertical cursor is indicating the direction of the approaching aircraft from direction 150°. Only 2 s after aircraft passage we can observe the first superposition of the normal wind pattern by the signatures of the descending wake vortex. This signature becomes clearly the shape of a pair of vortices (7 s) showing increasing vortex diameters with time (17 s, 26 s, 36 s). These organised flow patterns are turning into turbulent components within the next three pictures (50 s, 60 s, 69 s). After that, the pattern of the undisturbed wind field (84 s) as it was before (8 s before aircraft passage) is established again. The descent of the vortex pair is overlaid by a lateral transport from the direction of the aircraft passage (150°) to the left side of the picture, indicated by smaller azimuth angles. This movement is due to the cross component of the surrounding wind field of roughly 1.8 m/s.

In summary, the LDA measurements have clearly demonstrated that the wake-vortex signatures are detectable at short ranges even under small aspect angles. This has given an optimistic basis for the long-range vortex detection to be carried out by the MFLAME demonstrator.

8 sec before

aircraft passage

2 sec after

7 sec after

17 sec after

26 sec after

100° azimuth scan à 180°

Figure 9.4: LDA signatures of the A330 vortex pair shown in Figure 9.3,

for ten expanded sections of azimuth scans from 100° to 180°. The vertical cursor indicates the direction of the approaching aircraft from 150°.

36 sec after

50 sec after

60 sec after

69 sec after

84 sec after

100° azimuth scan à 180°

Figure 9.5 : Continuation of Figure 9.4

for the period from 36 s to 84 s after passage of the A330 aircraft.

9.3 MFLAME multifunction capability

In order to evaluate the performance of the lidar system, wind measurements at long ranges were performed with the DLR specific 2 µm signal processing unit. Data were analysed as single-shot measurements and as 5-shots accumulation. The measurements were taken parallel to the glide path (3° elevation), but with stationary scanners.

An example of single-shot measurements is shown in Figure 9.6, where the line-of-sight (LOS) component of the wind vector is drawn in the upper part of the figure for one single shot. The values below 0.5 km are not representative, since they might be disturbed by the outgoing laser pulse. Between 0.5 km and 4.0 km, the slope of the wind component is smoothly developing around a value of – 3 m/s. The negative sign indicates wind components directed away from the lidar system. Beyond a range of 4.0 km, several stray values due to insufficient signal-to-noise ratio are occurring. The same behaviour can be observed in the intensity plot in the lower part of the figure, where 200 single-shot LOS profiles are drawn. Here, the stray values, indicated by the solitary dots, are showing increasing frequency beyond 3 to 4 km range.

Figure 9.6: Single-shot wind measurement with 9 km maximum range.

In Figure 9.7, the results of 5-shots accumulation of the data shown in Figure 9.6 are presented. For example, no stray values are occurring in the chosen accumulated profile, as shown in the upper part of the figure. Moreover, the small number of stray values in the intensity plot (lower part) indicates the benefit of the accumulation of single-shot measurements. At least in the atmospheric boundary layer, maximum ranges of more than 10 km can be achieved by accumulation of few single-shot measurements. This is an important conclusion with respect to the multi-function objective of the MFLAME project

Figure 9.7: Wind measurement from Figure 9.6 with 5 shots accumulated.

9.4 MFLAME wake-vortex detection capability

The archived data from the phase 1 and phase 2 of the ground tests were transferred to NUI, Galway for processing and analysis. Prior to application of the MFLAME processing algorithms the lidar signal were corrected for gain difference caused by the automatic gain control of the recording system and for frequency shifts caused by jitter in the frequency of the outgoing lidar pulse.

The MFLAME processing approach involves two stages: signal processing and image processing. The aim of the signal processing stage is to estimate the mean radial velocity and the spectral width at various ranges along each line-of-sight. For this application, mean radial velocity estimation is carried out over 75 meter range gates, while spectral width estimation is carried out over 225 meter range gates.

In the image processing stage, mean radial velocity and spectral width estimates from all range gates that are equidistant from the lidar are used to generate transverse, range gate images. If a rectilinear scanning scheme was employed the construction of these images would be straight forward, as each mean velocity and spectral width estimate could be mapped to each pixel in mean velocity and spectral width range gate image.

A sinusoidal scanning pattern, however, results in a highly non-linear spatial distribution of estimates. Such a distribution does not facilitate image processing or image display. Therefore, a dedicated approach has been developed in order to convert the sinusoidally distributed velocity and spectral width estimates into a set of rectilinearly distributed velocity and spectral width estimates, while simultaneously allowing the implementation of noise reduction algorithms and preserving wake vortex Doppler signatures in the range gate images.

The off-line processing approach used at NUI, Galway is similar to that implemented in the MFLAME real time signal processing system. However, in the real time system, only the mean radial velocity is available after signal and image processing. It is intended to add the spectral width signal processing and image processing stages in the next version of the demonstrator software.

Airbus A340 vortices in the FOV

An example of the results of image processing is shown in Figures 9.8 and 9.9. This case involves an Airbus 340 landing at 13:38 on the 9^th of March 2000. The local wind information provided by the control tower indicated that wind was from 330° at 3 knots. The elevation of the central axis of the FOV was set to 4°.

Figure 9.8 shows the average top view of the mean radial velocity and a series of range gate images from scan 5 of this case. Figure 9.9 presents the spectral width results for the same scan in a similar format. A pair of spectral width signatures is clearly visible in Figure 9.9, in both the average top view image and in the individual range gate images between ranges 400 to 1400 meters. Beyond 1400 meters one of the vortices seems to have dropped out of the FOV and beyond 1800 meters both vortices seem to have dropped out of the FOV.

Using the vortex positions identified in Figure 9.9, axial velocities associated with the vortices can be identified in the range gate images of Figure 9.8. However, other turbulence structures are present in the radial velocity images, so axial velocity signatures are not as identifiable as the spectral width signatures.

The aircraft over-flew the MFLAME shelter approximately 2 seconds after this scan was completed. Assuming an aircraft speed of 75 m/s, the vortices in Figure 9.8 and 9.9 can be calculated to be approximately 4 seconds old in range gate 1, 5 seconds old in range gate 2, 6 seconds in range gate 3, etc. (as consecutive range gates are 75 meters apart).

The tick marks along the axes of the range gate image, which are separated by 20 meters, can be used to estimate the vortex signature separation. Between range gates 4 and 14 the separation of spectral width signatures seems to be stable at approximately 80 meters. The approximate vortex ages corresponding to these ranges can be calculated to be 7 and 15 seconds respectively. This estimated separation is slightly larger than the wingspan of an Airbus 340. However, it should be noted that the spectral width signature has not been shown to coincide with the cores of the vortices, so conclusions about core separation should not be drawn without further investigation.

Figure 9.8: Radial velocity images with A340 wake vortices in the FOV.

The left panel shows the average top view representation of the estimated mean radial velocity, while the right panel shows individual range gate images. The mean radial velocity of each range gate images has been removed. The ‘+’ symbols on the top view image mark the centres of the range gate images shown. Tick marks on the range gate images are separated by 20 meters on each axis of each image..

Figure 9.9: Spectral width images with A340 wake vortices in the FOV.

The left panel shows the average top view representation of the estimated spectral width, while the right panel shows individual range gate images. The ‘+’ symbols on the top view image mark the centres of the range gate images shown. Tick marks on the range gate images are separated by 20 meters on each axis of each image.

Airbus A320 vortices in the FOV

A second example involves an Airbus A320 in the approach to land at 13:01pm on the 9^th March. Figure 9.10 shows the average top view of the mean radial velocity and a series of range gate images from the 5^th scan of the FOV. As in the previous case, the average radial velocity of each range gate image has been subtracted in order to reduce the range of velocity in the image set. Figure 9.11 presents the spectral width results for the same scan in a similar format. Spectral width signatures are clearly visible in Figure 9.11 in both the average top view image, although unlike the A340 case, separate spectral width signatures are not visible for two vortices.

As in the previous case, the vortices’ location can be identified via the spectral width images, and then axial velocities associated with the vortices can be identified in some of the range gate images of Figure 9.10 However, other turbulence structures are again present in the radial velocity images. The spectral width signatures are clearly more useful for detecting wake vortices.

The aircraft flew over the lidar system shelter approximately 9 seconds before this scan began. Assuming an aircraft speed of 75 m/s, the vortices age in Figure 9.10 and Figure 9.11 can be calculated to be approximately 16 seconds old in range gate 2, 18 seconds old in range gate 4, etc.

Vortices from other aircraft types in the FOV

Detection of many other types aircraft has been demonstrated, including other medium sized aircraft (e.g. B737 and MD80) and some smaller aircraft (e.g. FOKKER 100 and BAC 146).

9.5 Synthesis of ground test issues

It can be stated that the MFLAME Demonstrator was successfully operated on ground in a configuration similar to a forward-looking airborne sensor. The analysis of the results has demonstrated the feasibility of the system for wake-vortex detection, even for the detection of vortices generated by small size aircraft.

By accumulation of several lidar shots, long-range wind measurements can be carried out, which is important for multi-function applications, like wind-shear detection.

The acquired results are appropriate for system scaling with respect to laser energy, pulse repetition rate, optics diameter, scanner performance, frequency estimator, and so on.

To define an operational equipment, well integrated in the aircraft system and able of reliable hazard detection without false alarms, further work is expected to obtain a complete statistical analysis of the MFLAME recordings and to design an automatic wake-vortex pattern recognition.

Figure 9.10: Radial velocity images with A320 wake vortices in the FOV.

Figure 9.11: Spectral width images with A320 wake vortices in the FOV.

10 Conclusion

After collection of user requirements for airborne detection of wake-vortices and other atmospheric hazards, a future equipment has been defined and a demonstrator has been built.

The demonstrator has been installed at the extremity of a runway on Toulouse airport so as to enable measurements in a configuration very close to an on-board detection from a follower aircraft.

This “airborne-like” detection has been successful. The tracking method as well as the signal processing constitutes a world premiere that has demonstrated the feasibility of an airborne equipment.

Industrial equipment for detection, warning and avoidance of wake-vortex, windshear and other atmospheric hazards could be brought to market within 5 to 8 years.

To achieve this goal, it is important to progress in the system integration of the MFLAME concept. This global approach should be the subject of further work with two major objectives:

- On board system integration with study of detection, warning and avoidance aspects, including avoidance manoeuvre trials in a flight simulator and flight tests.

- ATC and ground system integration, taking into account links with board system, to finally demonstrate how the operational capacity and safety of aircraft can be improved.