Concorde Accident Analysis

FAILURE ANALYSIS Investigators in the Concorde accident undertook research to find incidents that involved tyres or landing gear on the Concorde since its entry into service. Archives from EADS, Air France, British Airways, BEA, AAIB, DGAC, CAA and Dunlop were consulted to establish a list of such incidents. The assembled list contained fifty-seven entries; all cases of tyre bursts or deflations. Thirty cases were for the Air France fleet and twenty-seven for British Airways. Of these events; •Twelve had had structural consequences on the wings and/or the fuel tank •Six led to penetration of the tanks Nineteen of the tyre bursts or deflations were caused by foreign objects •Twenty-two occurred during takeoff •One case of tank penetration by a piece of burst tyre However none of the events showed any rupture of a tank, a fire, or a significant simultaneous loss of power on two engines. It was clear that the case of tyre bursts or deflation was not a new topic as far as the history of Concorde is concerned. In this accident and in many other incidents, this material (rubber and other materials constituting the tyres) failure led to other failures on the aircraft structure.

Understanding the destruction mechanisms of the tyres and tank rupture as well as the production of the fire would arm the investigators with powerful tools in preventing future accidents. Tyre failure Experimental tests Tests were carried out by Air France’s tyre supplier, Goodyear, at their technical centre in the United States to reproduce the conditions on the day of the accident. Two new Concorde tyres were used for the test and curved metallic strips with comparable dimensions to the one found on the runway were also used.

The tyres were installed on the side of a trolley towed by a truck. The load spread out on the trolley allowed each tyre to bear a load of about twenty-five tons, equivalent to that on each main landing gear tyre on Concorde. Taking into account the test equipment and the load, the speed of the truck was around 10 km/h. The sample strips were stood on edge on a concrete surface. During the tests: •An initial positioning of the strip, done with a titanium strip, resulted in its being flattened by the tyre. In a second position, the strip remained stable on its cutting side and the tyre was cut into, •The tyre cut went right through its thickness, practically all across the width of the area in contact with the ground and in accordance with the shape of the strip, •This cut continued as tearing onto the tyre shoulders and sidewalls through a static rupture in the direction of the reinforcing material of the tyre body, •The static tear spread as far as the tyre beads, in other words slightly more deeply than the tear noted on the remains of burst tyre in the accident.

Tyre test truck Metallic strip under the tyre Tyre cut Theoretical Study of Metallic Strip Cutting Tyre In the course of the investigation, the Mechanical Industries Technical Centre (CETIM), which is specialised in the study of polymers, plastics and composites, was asked to determine the theoretical behaviour of a tyre running over an obstacle like a metallic strip standing on edge. In order to do this, the CETIM conducted a study using finite element modelling on a bias ply carcass tyre with characteristics similar to those fitted on the ill-fated F-BTSC.

The mechanical and chemical characteristics of the materials were supplied by Goodyear, the manufacturer. Those of the metallic strip corresponded to the characteristics of the one found on the runway. Two cases were considered: •A so-called “short” strip of which at least one end is inside the contact area between the tyre and the ground. •A strip that was long enough to protrude beyond the contact area. This theoretical study shows that at the ends of the strip, the damage caused was typified in both cases by separation of the different reinforcing layers and a clear perpendicular cut in the tread by the edge of the strip.

Tests carried out at the CEAT [(Centre d’Essaie Aeronautique de Toulouse) (Toulouse Aeronautical Test Centre)] The objective of the tests at the CEAT was to run the Concorde tyres over metallic strips made of titanium to establish a catalogue of the various aspects of fracture topography relative to the parameters selected. Some metallic strips similar to those found on the runway were spot-welded onto thin metal plates. These slid along two cables to be introduced between the tyre and the drum on the test rig which drove the tyre at the predetermined rotation speeds.

Various tests were carried out with a load of 2,290kN with the inflated tyre running at low speed. These tests showed that the impact speed is an important parameter for strip penetration. A tyre carcass was cut with a knife on ten of the fourteen doublers. During re-inflation, the upper edges of the cut on the tyre tread separated by about 5 mm as soon as pressure of 3 bars was reached. This shows that the metallic strip could not have remained trapped in the tyre. After inflation, the tyre was rotated. The rupture occurred at 60 m/s and the main piece of tyre released from the cut weighed about 2. kilos. Metallic Strip Dynamic Penetration Test Two tests were carried out with a tyre rotating at high speed. For these tests, after simulating a 3000 meter taxi, the wheel accelerated to simulate a takeoff run. The metallic strip was then introduced edge on between the drum and the tyre. For the first test, the mechanism was activated when the tyre was running at 60 m/s. It immediately burst. Two pieces, one of eleven the other of seven kilos, were ejected, along with a long piece of the tread. For the second test, the speed was increased to correspond to a translation speed of 75 m/s. he tyre also burst as soon as the strip was introduced, releasing several pieces with a total weight of 17. 6 kilos. The two heaviest pieces weighed 5. 9 and 5 kilos. The pieces exhibited clean cuts in the contact area with the strip and similar shapes to those seen on the tyres of the crashed Concorde. Examinations Carried out at the LRCCP The Rubber and Plastics Research and Test Laboratory (LRCCP) were ordered by those in charge of the judicial inquiry to carry out examinations on the debris of tyre No 2 of the crashed Concorde. In the first instance, reconstitution of the tyre led to the conclusion that ore than 30% was missing and that the metallic strip had been struck from its concave side. The laboratory also checked that the characteristics of the tyre were comparable to those of the other Concorde tyres examined. On the surface of the cut, the material reinforcement fibres were cut through the major part of the thickness and some of the areas of rubber were iridescent, with spacing corresponding to those of the holes on the metallic strip. Various pieces of the tyre cut during the tests conducted in the USA and at the CEAT were examined at the LRCCP.

Observation showed the resemblance of their rupture topography with that of tyre No 2. The photo below shows the positioning of three pieces coming, from top to bottom, from tyre No 2 (speed of around 85 m/s), from a tyre tested at the CEAT with impact at 65 m/s and a tyre tried in the United States with impact at 2. 5 m/s. Cuts in various Concorde tyres At this point in the investigation, the investigators decided that they new probably enough about the tyre failure and the focus was now on the fuel tank material failure. Tank rupture

The BEA report said that the rupture of the fuel tank was caused by a mechanism that had never been seen on civil aircraft before the accident and about which it is difficult to determine the precise process. Penetration of the tank by the piece of the burst tyre might have contributed to the process by creating a hydrodynamic pressure surge, described in the report as follows: On penetrating the liquid (fuel inside the tank), the projectile is rapidly slowed down. During this slowing, its kinetic energy is transferred to the liquid, and a cavity of a certain volume is created around it.

In case of confinement, or when the tank if full, the fluid, being incompressible, transmits a mechanical load dependent upon the volume of the cavity. Impact force and stress estimates One can then argue that even though the tyre had burst, if none of the tyre pieces were to penetrate the tank, the fire would not have been produces and the aircraft could have made a safe landing. If the fuel tank was impenetrable, the disaster could have been avoidable. Simple estimations of the forces and stresses experienced by the tank casing can help understand how there was a puncture on the bottom of the tank.

Investigators found tyre pieces weighing up to 4. 5 kg individually. After the tyre had burst, these pieces would have been flying around at a speed much higher than the 220 knots (113. 18 m/s) take-off speed. A part found at the site was identified as coming from the underside of tank 5. The piece had not melted but the external paint and the internal black mastic were damaged on three-quarters of their surface and the material was overheated in this area. A 40 X 10 mm hole is noticeable on the front right part of the piece.

Examination thereof revealed the following details: •The impact occurred from the outside towards the inside of the tank, from the left to the right and more or less from the rear towards the front. •The puncture showed clear petal-shaped structure, implying a high-energy penetration, which appears to indicate that it was not due to the final impact. Analysis was unable to provide details on the makeup of the penetrating object. Its probable trajectory shows that it could have come from the area of the left main landing gear. For argument sake, let’s say that a 4. kg piece of tyre was travelling at the take-off speed of 113. 18 m/s and had to come to a halt within a fraction of a second, say on millisecond, the change in momentum would exert a force of: If this force, which is acting on the outer surface of the tank casing was to have been applied on a surface area equal to that of the puncture on the bottom of the tank, i. e. : 10 mm by 40 mm or 0. 0004 square metres, the axial stress on that are would have been of: Or roughly 1. 273 GPa Note:The above calculations are by no means accurate.

They are simply estimates and a lot more factors could have contributed in puncturing the tank. The fuel tanks of the Concorde were made from a copper based aluminium alloy known in Britain as RR58 and in France as AU2GN, this material was chosen for its creep resistance and strength, but having a look at the table below, of typical yield strength of material used in engineering, we can see why the stress exerted on the tank casing could have been too much even for the special chosen material to withstand. MaterialTypical yield strength (GPa)

Steel0. 2 to0. 7 Cast iron0. 2 to 0. 5 Aluminium alloys0. 02 to 0. 1 Tank Rupture Mechanism Examination of the piece of the ruptured tank allowed investigators to exclude the possibility that the destruction of this part of the tank resulted from a direct puncture by a large object or by tearing off of the piece as a result of a puncture. To explain the rupture from the inside towards the outside of the underside of the panel, a lot of theoretical and practical work was undertaken. Based on available information, two scenarios ere considered: a) Impact of a piece of tyre On a self-stiffened panel a shock leads to: •In the impact area, deformation in the direction of the impact (direct mode); •In neighbouring areas, deformation in the opposite direction by continuity effect on the structural elements (indirect mode). When the box contains liquid, a secondary effect can appear which contributes to the indirect mode, an effect due to: •The wave of pressure that is propagated in the liquid at the speed of sound, that is to say at about 1,400 m/s.

This wave diminished rapidly and after an initial pressure of two hundred bars, it was only about ten bars in the area where the indirect mode was expected; •The successive displacements of the liquid itself, at a speed of a few dozen metres a second. Because of the incompressibility of liquids, and in as much as the tank is full”, that is to say there is no free surface too near the impact area that disturbs the phenomenon, this displacement tends to push the tank structure towards the outside, first of all in the nearest areas. b) Rupture by hydrodynamic pressure surge:

Methods used in the military field have shown that the puncture of a tank by a high-speed projectile can have catastrophic consequences through generation of what is known as a pressure surge: on penetrating the liquid, the projectile is rapidly slowed down. During this slowing, its kinetic energy is transferred to the liquid, and a cavity of a certain volume is created around it. In case of confinement, that is to say when the tank is full, the fluid, being incompressible, transmits to the structure a mechanical load dependant upon the volume of the cavity.

Note: a backshock can also be generated when the cavity collapses. The investigation therefore tried to determine if these scenarios could be applied to the case of the Concorde accident and explain the damage to tank 5. Rupture by Tyre Impact The principle The initial shock, by pushing the walls, displaced a certain amount of fuel, which caused a displacement movement within the liquid. It was this displacement that pushed out the surfaces neighbouring on those on which the impact occurred.

It might be the neighbouring areas on the underside or the vertical walls, depending on the local geometry and the location of the impact. To effectively reach the level of rupture: •The zone where the indirect mode can appear must be an area of thin skin; it must be surrounded by an area notably more rigid to withstand the initial shock and to limit the possibility of deformation beyond the area where the indirect mode can appear. •Displacement of the fluid must be partially channelled in a particular direction due to a lateral wall, for example. Very local variations in geometry such as in the stiffener fillets are potential incipient rupture zones, through concentration of stresses. Tests In the context of the investigation and also for the work performed to return the aircraft to service, a series of tests to damage a tank with heavy projectiles was carried out at the CEAT in the first half of 2001. During these tests, pieces of tyre were fired at high speed at test boxes. So as to be as representative as possible, the box used for the last firing was made out of a panel from tank 5 taken from a Concorde.

However, the exact shape of the tank walls, their size and internal equipment could not be represented precisely. The boxes were filled with a liquid whose mechanical characteristics and viscosity were similar to those of kerosene. They were equipped with load sensors and pressure sensors. The major limitations on the tests due to existing equipment were as follows: •Maximum projection energy imposed by the weight and speed of the projectile (4. 8 kg – 106 m/s). •Horizontal firing. •Limited size of boxes. •Limited number of firings and boxes.

Bearing in mind the large number of parameters enabling the impact to be defined and the limitations of the available test equipment, it was not possible to reproduce the rupture noted at the time of the accident. Nevertheless, the overall result of the tests performed enabled the scenario to be developed – the indirect mode certainly existed – and to confirm the theoretical models used to quantify this phenomenon. Calculations Theoretical studies were undertaken on the basis of the overall tank 5 structure-fuel model using the RADIOSS software programme.

This code, still called the “crash” code, is recognised as the state of the art in dealing with rapid dynamic phenomena and fluid/structure interconnections at the same time. The computer models were based on Concorde’s tank 5 and on the boxes defined and manufactured for the ratification tests. The procedure was carried out in two stages: •Identification of the most sensitive areas in the structure. •Detailed modelling of these areas with sample backup tests to adjust the parameters. The rupture criteria were the subject of a specific study.

The results of the calculations were in accordance with the facts and measurements taken during the study, under the conditions in which they were carried out, that is to say below the energy level required to bring about a rupture. Possible Sources of Energy Taking into account the preceding analysis and the known accident conditions, the level of energy locally necessary to cause the rupture can be calculated through the impact of a piece of tyre of around 4. 5 kg with a speed of around 140 m/s.

On the basis of the calculations made, this piece of tyre could have reached this speed through a combination of effects resulting from rotation of the tyre and the tyre burst. However, it cannot be ruled out that the level of energy necessary could have been reached through the added effect of other phenomena such as: •The impact of one or more other pieces of tyre. •Greater concentration of the energy in the fillets. This can be achieved by special impact conditions in terms of position, attitude and perhaps rotation speed of debris.

The movement of the fuel and its interaction with the internal structure of the tank may also influence this. •The previous weakening of the structure in the rupture initiation area. Rupture by Hydrodynamic Pressure Surge ONERA (the National Aerospace Study and Research Office) developed a method for numerical analysis of the pressure surge phenomenon in the context of tank punctures via high-speed projectiles, and the BEA asked them to study the relevance of this scenario in the case of the Concorde accident. The objectives of the study were: To determine if the hydrodynamic pressure surge phenomenon can occur at relatively low speeds (in comparison with the speed of a bullet which is about 1,000 m/s). •To determine if the hydrodynamic pressure surge phenomenon can be the cause of an “indirect mode” rupture of the tank structure. •In case of tank rupture, to determine if it starts from the puncture location. Method Employed ONERA did not model the puncture process on the lower skin, the simulation beginning after the projectile entered the fluid. The finite element calculation code was the same as that used by EADS, that’s to say the RADIOSS code.

The theoretical characteristics of a characteristic projectile, in accordance with the characteristics of the hole found on the piece of tank discovered at the site, correspond to a small cylinder with a weight of forty-five grams. Its speed in the fluid was fixed at 120 m/s. Finally, its point of impact was chosen as the location of the puncture observed on the piece of tank 5 found at the Gonesse site, which corresponds to a skin thickness of 1. 6 mm. It should, however, be noted that some of the trajectory characteristics chosen are not entirely compatible with observations made on the piece of the tank.

Note: the speed of 120 m/s is an estimated maximum speed, consistent with: •The linear speed of the aircraft at the time of the tyre burst (85 m/s). •The increase in speed imparted to the debris by the tyre destruction mechanism. •The loss of speed due to the puncture. Based on knowledge acquired in the military field, it was also hypothesized that the projectile had an initial slope angle in the fluid of 30° in relation to the skin it struck and that it was turning round during the first moments of its passage.

It has been established that this type of configuration can generate a hydrodynamic pressure surge on the skin underside, the latter being even greater when the turn occurs near the skin. This is the most onerous case known. Several calculations were made, always with the tank fully filled, using various material laws, with or without rupture criteria, as well as different projectile turn kinematics. It should be noted that the phenomenon described diminishes very rapidly, or even disappears, if a free surfaces is located near the puncture area.

The Results The significant results of the particular case studied were as follows: •The calculations for each simulation took place normally, without any accumulation of energy errors or numerical instability, which shows that the method was reliable; •A hydrodynamic pressure surge phenomenon was observed following penetration and turning of the projectile in the tank; •The loads transmitted to the structure did not lead to a rupture in the area affected by the pressure surge.

However, they can lead to structural damage in the connection areas : the shock wave created overpressure that loaded the rib laterally and the resulting bending could initiate a local rupture at the base of the rib; •The crack did not initiate in the puncture area itself. Position of the rupture areas on the lower skin Fire production On the basis of the known facts and based on the known properties of turbulent flames, three points were studied: •The stabilisation of a quasi-stationary turbulent flame under the wing of the Concorde during the takeoff run and flight. Estimation of the fuel flow coming from the leak under the wing of the Concorde. •The mechanisms that may have led to the ignition then the propagation of the flame under the aircraft’s wing. Flame stabilisation and retention When an obstacle is placed in an airflow, the development of turbulence is observed with re-circulation zones. In these zones, the flow can move in the opposite direction to that of the main flow in some areas. This re-circulation zone allows a flame front to stabilise through two mechanisms: •The re-circulation generates an area of low speeds. The re-circulation zone contains burnt gases and acts as a reservoir for hot gases that contribute to the ignition, slightly downstream, of the fuel-air mixture. These mechanisms may explain the stabilisation of the flame in the left landing gear bay, as can be seen on photos of the aircraft on takeoff. Indications of stabilisation of the flame are not therefore necessarily apparent on the gear leg, partly because the flame is slightly stabilised downstream and in part because the leg is continuously cooled by the flow from upstream. Re-circulation zone

Estimation of fuel flow Based on photos and videos of the accident flight, the estimation of the average fuel flow was carried out using three approaches, which give similar results. The first uses the Magnussen model, a simple model developed to describe the reaction rate of non-pre-mixed turbulent flames, that is to say where the reactive elements are injected into the reaction zone separately. Taking the hypothesis of a flame three metres in diameter, fifty metres long and ten centimetres thick, modelling leads to fuel consumption close to 60 kilograms per second.

In the second method applied, the coherent flame model equates the flame with a surface, and the reaction rate becomes the product of this surface and a surface reaction rate estimated according to a laminar flame model. According to this method, and in relation to the parameters selected for the size of the surface, the fuel consumption varies between 20 and 130 kilograms per second, with a peak in probabilities (corresponding to average and realistic values of the size of the flame) of around sixty kilograms per second. This model thus confirms the overall rate established with the first odel. The third estimate was made from the quantity of fuel remaining in tank 5. The quantity loaded was 7. 2 tons and the gauge indicated two tons after the accident. The flight time between the estimated rupture of the tank and impact was around eighty-one seconds. The estimated fuel flow rate, apart from the leak due to the small puncture and (the) consumption by engines 1 and 2 (around 350 kg) was therefore around 60 kilograms per second. In conclusion, the overall flow rate of the leak is several dozen kilograms per second, thus about ten times greater than in the Washington event.

The high rate of flow from this leak contributed to the ignition of the fuel since it led to a fuel/oxidizer mixture, which was almost a stoechiometric mixture, thus perfectly flammable. Ignition and Propagation of the Flame Various potential sources of ignition of the fuel were identified in the course of the investigation. Three were selected and were the subject of extensive study: •An engine surge, •An electric arc, •Contact with the hot sections of the engine and/or reheat. No evidence was found of previous ignition of a hydraulic leak.

No trace of any hydraulic leak was found at any stage of the investigation. Engine Surge Ingestion of solid or liquid elements by an Olympus 593 engine can cause a surge in the high-pressure compressor, which would generate a wave of pressure towards the front of the engine. This phenomenon can lead to the appearance of a flame spreading toward the auxiliary air intake then the main air inlet. Fuel ingestion tests carried out by Rolls Royce confirmed the appearance of such a flame with duration of eighty to a hundred milliseconds.

Other tests conducted by BAE Systems showed that a flame coming from the auxiliary air intake can propagate forward in the turbulent airflow located downstream from the left landing gear and attach itself on it. Nevertheless, this hypothesis was rejected, since the appearance of the fire preceded the surges, as shown by the chronology of events (pool of unburned kerosene and traces of soot on the runway) and the nature of the surges identified (ingestion of hot gases and not of liquid fuel). Electric Arc

A study conducted at the CEAT showed that it was possible to generate an electric arc by a short-circuit on an electric harness situate in the area of the main landing gear and that the energy produced was compatible with igniting vaporised kerosene. The tests simulated a short-circuit in the case of damage by crushing, tearing or cutting through the insulators of the electric line supplying the brake ventilators (3-phase 115 V, 400Hz). During the tests, the circuit breakers never tripped, apparently because the phenomenon was of too short a duration for them to detect it.

The successive sparks had an energy estimated at twenty-seven joules, clearly above that required to ignite the vaporised kerosene, including in turbulent air conditions. Tests carried out in Great Britain confirmed that the immediate ignition of vaporised kerosene was possible in the area of the gear well with an electric spark of three joules. The flame then attached and stabilised directly on contact with the landing gear bay, in the re-circulation zones.

Although the electric cables are partially protected by the gear leg, possible damage due to the destruction of tyre No 2 cannot be entirely ruled out. It should, however, be noted that after the modifications carried out following the Washington event, no further cases of damage to these cables has been reported by the operators. Ignition after spark Contact with the Hot Sections of the Engine After the rupture of the tank, kerosene ingestion through the nacelle/engine assembly could have occurred through: •The auxiliary air intake and/or the ventilation door, The air conditioning air bleeds exchangers. In red, the lateral air conditioning air bleed door used for the air conditioning The kerosene ingested could have ignited on contact with the hot walls of the engine or on contact with the gas coming from the reheat, at the level of the thrust nozzle. In this area, many obstacles allow the development of re-circulation zones and ensure retention of the flame in the rear part of the engine. It should, however, be noted that no traces of fire were discovered during the examination of the engines.

For this hypothesis on ignition to be applied to the 25 July 2000 accident, it is necessary to explain how the flame could then have “propagated forward” to get to and attach itself behind the landing gear well. A study conducted in the context of the investigation by two CNRS researchers shows that two routes are possible: via the outside of the inside of the nacelle. The airflow speeds inside the nacelle, of around 20 m/s, would allow the flame to flow back quickly enough so as not to cause engine damage. No trace of any fire was in fact brought to light during examination of the engines.

The forward propagation of the flame could not possibly have occurred through the air conditioning circuit, whose exchanger mesh is too fine. It is possible, however, in the direction of the second secondary air bleed, which would take the flame to the area of the re-circulation zone that develops behind the gear leg. The tests in Great Britain showed that by igniting the main air flow at the level of the first secondary air bleed, thus about one metre upstream of the second, a flame was created that flowed back rapidly to attach itself to the gear well.

Nevertheless, it must be underlined that it is not easy for the flame to come out of the nacelle at the level of this air bleed. Note: The hypothesis on kerosene ingestion through the air conditioning air bleed and its ignition on contact with hot gases had been studied by the manufacturers after the Washington event. The result was that the risk of ignition was real but that the flame could not propagate against the airstreams because of the exchanger mesh.

The absence of a fire and the low flow rates noted explain why this hypothesis was not developed further. Forward propagation of the flame via the outside of the nacelle meets a theoretical obstacle: the propagation speed of a turbulent flame can barely exceed a few metres per second whereas the airstreams under the wing of the aircraft is about 100 m/s. It is, however, sufficient for the flame to encounter locally, at a given moment, airflow that is sufficiently slow for it to be able to flow back.

The complex geometry of Concorde’s lower wing, in particular the presence of a fairing between the nacelle and the wing, the disturbance to the airstreams by the presence of the flame itself and the wake from the landing gear are three elements which make it possible to envisage sufficiently low speeds to be born by the flame Note: Because of the chaotic nature of the turbulent combustion, a numerical simulation would be too inconclusive since the results would be too dependent on the model and the hypotheses selected.

Flame forward propagation from the rear of the aircraft could not be produced during the tests conducted in Great Britain, but it was not possible to reproduce the exact conditions of the accident. CONCLUSION Findings •The aircraft possessed a valid certificate of airworthiness. •Repeating the calculations for the flight preparation showed that the estimated weight of the aircraft on departure was in accordance with operational limits. •Taking into account the fuel not consumed during taxiing, the aircraft’s takeoff weight in fact exceeded the maximum weight by about one ton.

Any effect on takeoff performance from this excess weight was negligible. •During takeoff, the tyre on wheel No 2 was cut by a metallic strip present on the runway. •The metallic strip came from the thrust reverser cowl door of engine 3 on a DC 10 that had taken off five minutes before the Concorde. •A piece of the tyre from wheel No 2 weighing 4. 5 kg was found on the runway, near the metallic strip. Other pieces of this tyre and a few light pieces from the aircraft were also found. •Rubber marks from the damaged tyre on wheel No 2 then appeared. •A large part of the underside of tank 5 was found on the runway.

It bore no signs of impact and had been ripped away from the inside towards the outside. •Another part of the underside of tank 5 was found at the accident site. It had a puncture ten millimetres wide and forty millimetres long. •Research showed that a projectile penetrating tank 5 could have generated a hydrodynamic pressure surge but that this could not have caused the ripping out of the piece of the tank found on the runway. •A large kerosene mark was found on the runway, immediately after the piece of the tank. •The fuel that was leaking was ignited; a flame and large quantities of smoke appeared behind and to the left of the aircraft. After the aircraft’s passage over the metallic strip, the rupture of tank 5 and the ignition of the leak, engines 1 and 2 suffered simultaneous surges leading to slight loss of thrust on engine 1 and a severe loss on engine 2. •Because of incomplete opening of the left main landing gear door or the absence of detection of opening of these doors, the crew was unable to retract the landing gear. •Because of the lack of thrust and the impossibility of retracting the landing gear, the aircraft was in a flight configuration which made it impossible to climb or to gain speed. The aircraft then adopted a very pronounced angle of attack and roll attitude. •The aircraft crashed practically flat, destroying a building and was immediately consumed by a violent fire. •Many pieces of the aircraft found along the track indicate that severe damage to the aircraft’s structure was caused in flight by the fire. •Even with the engines operating normally, the significant damage caused to the aircraft’s structure would have led to the loss of the aircraft. Probable Causes The accident was due to the following causes: High-speed passage of a tyre over a part lost by an aircraft that had taken off five minutes earlier and the destruction of the tyre. •The ripping out of a large piece of tank in a complex process of transmission of the energy produced by the impact of a piece of tyre at another point on the tank, this transmission associating deformation of the tank skin and the movement of the fuel, with perhaps the contributory effect of other more minor shocks and /or a hydrodynamic pressure surge. Ignition of the leaking fuel by an electric arc in the landing gear bay or through contact with the hot parts of the engine with forward propagation of the flame causing a very large fire under the aircraft’s wing and severe loss of thrust on engine 2 then engine 1. In addition, the impossibility of retracting the landing gear probably contributed to the retention and stabilisation of the flame throughout the flight. RECOMMENDATIONS

The investigation did not bring to light the need for any other urgent recommendations However, on several points, some improvements specifically linked to Concorde seem desirable in the light of information from the investigation. These improvements, which are the subject of the following recommendations, were brought to the attention of the French airworthiness authorities and were taken into account in the context of the aircraft’s return to service. 1. For any transport aircraft, it is essential that feedback, through analysis of in-service incidents, be as effective as possible. Taking into ccount the small number of aircraft in service and their limited operations, in-service experience on Concorde is particularly limited. It is, however, both an ageing and a complex aircraft. It has been noted that the rate of malfunctions in certain systems or equipment was higher than current rates on other aircraft. Consequently, the BEA recommends that: The airworthiness authorities, the manufacturers and the operators of Concorde reinforce the means available for the analysis of the functioning of aircraft systems and in-service events and for the rapid definition of corrective actions. . The Concorde Flight Manual stipulates that a red alarm must lead to an immediate reaction by the crew… In the same manual, dealing with an engine fire is consistent with this general instruction. However, the Air France Operations Manual requires that no action be taken before reaching four hundred feet. Consequently, the BEA recommends that: Air France ensure that the emergency procedures in the section on Concorde utilisation in its Operations Manual be coherent with the Flight Manual. 3.

Recording the engine parameters which allow engine speed to be determined only every four seconds slowed down and complicated some work essential for the technical investigation. This characteristic also tends to mask certain facts during examination of incidents for which it would not be possible to devote as much time and effort as for the 25 July 2000 accident. In contrast to Air France’s Concorde aircraft on the day of the accident, British Airways aircraft are equipped with recorders that allow the parameters from all four engines to be recorded every second. Consequently, the BEA recommends that:

Air France equip its Concorde aircraft with recorders capable of sampling at least once a second the parameters that allow engine speed to be determined on all of the engines. 4. The technical investigation brought to light various malfunctions relating to the operation of the aircraft, for example the use of non-updated flight preparation data, the absence of archiving of certain documents or incomplete baggage management. Equally, omitting the left bogie spacer was a consequence of non-respect of established procedures and of the failure to use the appropriate tool. Consequently, the BEA recommends that:

The DGAC undertake an audit of Concorde operational and maintenance conditions within Air France. Beyond specific improvements to Concorde, the investigation showed the need for progress in safety in various areas. This general progress is the subject of the following recommendations. 5. Tests and research undertaken in the context of the investigation confirmed the fragility of tyres against impacts with foreign bodies and the inadequacy of the tests in the context of certification. Recent examples on other aircraft than Concorde have shown that tyre bursts can be the cause of serious damage. Consequently, the BEA recommends that:

The DGAC, in liaison with the appropriate regulatory bodies, study the reinforcement of the regulatory requirements and demonstrations of conformity with regard to aviation tyres. 6. The investigation showed that a shock or a puncture could cause damage to a tank according to a process of transmission of energy from a projectile. Such indirect processes, though known about, are complex phenomena which had never been identified on civil aircraft. Equally, the ignition of the kerosene leak, the possible forward propagation of the flame, its retention and stabilisation occurred through complex phenomena, which are still not fully understood.

Consequently, the BEA recommends: The DGAC, in liaison with the appropriate regulatory bodies, modify the regulatory certification requirements so as to take into account the risks of tank damage and the risk of ignition of fuel leaks. 7. The loss of a metallic strip by the Continental Airlines DC10 has been identified as resulting from maintenance operations that were not in accordance with the rules of the art. Consequently, the BEA recommends that: The FAA carry out an audit of Continental Airlines maintenance both in the United States and at its foreign sub-contractors. MODIFICATIONS TO CONCORDE

After deciding that the main cause of the accident was the ignition of the kerosene flowing from the rupture in a fuel tank; the best possible source of protection was found to be lining the insides of certain tanks with Kevlar-rubber panels. These panels would stop any foreign object from penetrating the tank even if such object could rupture the tank casing, therefore preventing fuel from pouring out. The liners were designed to reduce the flow rate due to any tank rupture to around 0. 5 litres/sec. The rupture that caused the Paris accident was allowing fuel to escape at around 100 litres/sec

At EADS’ request, the Goodyear tyres were replaced by Michelin NZG tyres, especially developped for Concorde and other new aircrafts such as the Airbus A380. This new aircraft tire technology, christened NZG for “Near Zero Growth”, uses a high-modulus reinforcement material. This offers higher damage resistance and substantial weight gains, two key qualities in the field of aviation. These tyres were tested on an Air France Concorde, again at Istres the military test base in the Rhone delta region of France, during a series of ground and flight tests that took place in May 2001.

MICHELIN NZG TYRE SPECS External Diameter 110cm Width 40cm Weight 80kg Tread 4 Groves Pressure 16 Bar Max Speed 280MPH No. per Aircraft 8 Other modifications included: •Changing the normal operating procedures so that the 115V supply to the carbon brake cooling fan is deactivated during takeoff run, these fans can then be turned back on once the undercarriage has been retracted to fully cool the brakes •Armour plating the wiring in the undercarriage bay. •New seating and fittings to reduce the mass by 400 kg to make up for added weight of safety improvements