Investigations Visuals Guestbook In Memoriam Further Info Search


Introduction / Summary

Crew Competent and Alert

No Ice Contamination Pre-Impact System Failures In-Flight Fire/Explosion Conclusion


CASB Minority Report

Pre-Impact System Failures

Power Lost Before Crash

There is no doubt that all engines were turning just before the crash, nor that the right outboard engine (the number four engine) was turning slower than the others. But air passing through a jet engine keeps it turning even after an in-flight shutdown, so rotational speed does not necessarily equate to power production.

The majority could not determine if the low rotational speed of the number four engine was "the result of an in-flight power loss" or was "the result of tree fragment ingestion prior to ground impact". In our opinion, the evidence is conclusive that engine number four was operating at low power before it contacted trees.

The "tree model" shows the fuselage pitched up about nine degrees and yawed to the right at about 10 degrees when the tail struck the first tree. With this pitch angle, the engines could have ingested "tree fragments" for only part of the time that the aircraft plunged through the trees for less than one second.

The "tree model" also shows that when the aircraft first hit trees, it was banked to the right at about seven degrees. Since the wing tips are tilted up at about seven degrees with respect to the roots (6.5 degree dihedral), the two right-hand engines (numbers three and four) entered the tree canopy at virtually the same instant. It follows that "tree fragment ingestion prior to ground impact" cannot account for the vast differences in damage to these engines.

The spare conifers and slender deciduous trees typical of the accident site could not have damaged the number four engine extensively during this short period. We know this from the inlet guide vanes which present the first obstacle to foreign material entering an engine. Seventeen (of 23 total) inlet guide vanes of the number four engine were available for inspection. The leading edges of the inlet guide vanes "generally were in good condition" with no apparent damage from trees or other foreign material.

The bottom of the number four engine case was crushed both front and rear. The fractured blades on the first two (low pressure) compressor stages indicate that the (low pressure) shaft was in fact turning as the front of the engine was crushed; that is, the engine was either windmilling or producing some (unknown amount of) power at this instant. The turbine stages, attached to the same shaft as these compressor stages, were damaged as the back of the case was crushed an instant later. These turbine stages "exhibited relatively little rotational damage." In fact, there is no detectable rotational damage on the final turbine stage. When the engine case was crushed against it, rotation had already ceased.

Low rotational damage to turbine stages at the rear of the engine is consistent with high rotational damage to compressor stages near the front if, and only if, initial impact on the engine was near the front.

Thus, the low pressure spool of the number four engine was stopped during the small interval of time between the instants when the blades were torn from the compressor stages (by crushing at the front) and when the final turbine stage was damaged. Power produced during this time interval would augment the inertial torque tending to twist the hollow shaft. Yet, this shaft remained essentially untwisted on the number four engine.

The low pressure turbine shafts on the other three engines were all twisted in excess of 30 degrees even though both front and rear stages showed heavy rotational damage. This is clear evidence that the number four engine was rotating substantially slower (and by inference producing substantially less power) than the other engines. This finding is substantiated by the open bleed valve on the number four engine, corresponding to idle power or less. Wood fibres found in this valve also suggest that it was open (hence the engine was not producing power) prior to any ingestion of "tree fragments."

The majority reports that attempts to compare the pre-impact power output of the number four engine to that of the others led to contradictory results. The manufacturer of the engines found little difference between the rotational speeds (and hence between the presumed power) at impact. CASB investigators concluded that this difference was greeter than 40 per cent.

We note that large percentage differences in rotational speed would be consistent with small absolute differences if all engines were at low speed. The observation that the bleed valves on engines remains uncertain.

Both witnesses who observed the "orange/yellow glow" from directly under the flight path believed that the engines were not running

Witness 1: "The airplane passed right over my truck. When it passed right over us, the engines were not running. I did not hear any whine from the engines. I had gone by there hundreds of times when planes were taking off and you could hear the engines. But I could not hear the engines yesterday. There was no whine but there was some type of rumble... I'm certain that when the aircraft passed over its the engines were not working."
Witness 2: "I heard the noise. I looked. I could see the plane coming over. It didn't sound like engine noise... I live fairly close to the Sydney Airport and I've heard planes taking off before. This one didn't sound right.... There was no roar from him at all".

This "ear witness" testimony is all the more striking since the engines would sound louder than normal as the aircraft flew lower than normal over the trucks.

To us, spooling down of all engines provides a more plausible explanation of the tremendous deceleration than does a massive increase in drag due to 0.03 or O.04 inches of ice on the wing.

Thrust Reversers May Have Been Deployed

The sliders on the lower tracks of all four thrust reverser assemblies suggested that the reversers had not been fully forward (that is, not latched in the stowed position) at the time of impact. The position of the number four thrust reverser doors further suggested that they had been deployed prior to impact. The majority concluded that the displacement of all the reverser assemblies (translation rings) and the damage to the number four unit were due to rearward "dragging action during impact." Thus, the majority ruled out in-flight deployment of a thrust reverser as a factor in this accident.

A different appreciation of the evidence may be gained by considering how the rotational damage on all engines establishes the direction of the initial impact force.

We have noted that the engines could not have been in contact with the trees for more than about a second during which the aircraft was pitched up and yawing to the right. Consideration of the possible magnitude and direction of resulting Forces shows that tree contact prior to the main ground impact can not account for "dragging action" on the thrust reverser. The direction of the initial ground impact force can be readily established from the rotational damage on the engines. The direction of twist on the low pressure shafts of the numbers one, two, and three engines indicates that initial impact was at the front of these engines. The low pressure shaft of the number four engine remained essentially untwisted. But, the progressively decreasing rotational damage shows that the number four engine also struck ground first at the front. Thus, the initial axial deceleration would have exerted high forward G-Forces on all components of all engines - including all the thrust reversers.

The reversers (translating rings) are normally latched to prevent rearward movement. But, once unlatched, they move relatively easily on their sliders. Had the reversers been stowed in their normal (forward) position when the front of the engine struck the ground, decelerative forces would have tried to drive them even further forward, forcing the latch links even more firmly into the locked position. Under these circumstances, the sliders and witness marks would have been found at extreme forward positions, not "near" the forward positions as observed.

If, however, the reversers had been deployed (that is, positioned at the rear of their tracks) at the moment of initial impact, decelerative forces would have driven them forward. The forward motion of the translating rings would have tended to close the deflector doors. Under such circumstances, the deflector doors could be deployed, stowed, or anywhere in between at the time of subsequent secondary impacts. Witness marks from secondary impacts could correspond to the stowed or nearly stowed positions, or anywhere in between.

Since the cylinders of the hydraulic actuators are double acting, they would split from rapid forward extension as readily as from the rearward extension postulated in the majority report. Detailed examination of scuff marks on the interior of the cylinder walls might have been able to establish which way the pistons were moving at impact.

We also note that the S-shaped bends in the number four thrust reverser lower track (evident in Fig. 1.10.) suggest buckling due to compressive forces. The apparent failure in tension of the attachment links of the deflector door mechanisms also suggests failure during forward movement. These observations support the hypothesis that the reverser was driven forward by decelerative forces.

Similar re-interpretation could be made of the majority findings with respect to the other three reversers. Figures DO2 and DO3 show the deflector doors of the number one reverser, for example. The orientation and lack of continuity of scratches and buckles across the door/housing interfaces suggest that the doors were deployed at initial impact. At least two of the thrust reverser control valves (which are located in the engine pylons) were apparently recovered. The position of the sliders in these valves may have shed additional light on the pre-impact states of the thrust reversers. Unfortunately, these parts appear to have been discarded without examination.

We believe that all the evidence cited by the majority can be re-interpreted in the light of the large axial decelerative forces at initial ground impact. Such re-interpretation supports the hypothesis that the number four and likely the other three thrust reversers were deployed prior to the crash.

Multiple Malfunctions

The majority concluded that the Arrow Air DC-8's flaps were extended to the expected 18-degree takeoff position even though the wreckage yielded inconclusive and contradictory evidence.

The piston in one of the six recovered flap actuators left a clear imprint corresponding to 25-degree extension. Another had two imprints corresponding to 17- and to 32-degree extension. The remaining four actuators with less clear indications were initially assessed as corresponding to 23, 27, 40, and 43 degrees. Eight of 10 flap track pairs were recovered. Most tracks showed multiple imprints corresponding to a range of settings from 3 to 50 degrees. The flap position indicator read 38 degrees.

All three flap lockout cylinders were recovered, although in severely damaged condition. Two suggested that the naps were fully extended, while the third suggested a settling near mid-range. These findings could be explained by two simultaneous hydraulic line failures. The majority found this explanation improbable and attributed the contradictory indications to post-impact damage.

To us, this contradictory evidence does not support a determination of a pre-impact flap position of 18 degrees. We are less ready than the majority to rule out improbable multiple failures in such a complex accident.

Multiple failures are also suggested by the landing gear which remained extended. The captain, an experienced instructor/pilot, would have reacted to declining airspeed after take-off by calling for full power and raising the gear. Disintegration of the cockpit area precluded determination of the position of the landing gear lever. But, if the crew did attempt to raise the gear, the extended landing gear could signal another apparently independent failure.

We believe it unlikely that the contradictory evidence about flaps, spoilers, EPR gauges, N1 tachometers, and other systems can let explained separately through unrelated hypotheses. To us, the extent of the contradictory evidence suggests simultaneous multiple system failures due to a common cause.

Top of page Next page Previous page

[ Home | Investigations | Visuals | Guestbook | In Memoriam | Further Info | Search ]
©Copyright 1997-2008 JWS