We just didn't know enough about wing tip vortices back then and were probably lucky this type of accident didn't happen more often.

— James Albright

image

Updated:

2012-10-10

image

Wing Tip Vortex, from
NASA Langley Research Center

But now we do, there is no excuse for not knowing about proper Wing Tip Vortices separation.

1 — Accident report

2 — Narrative

3 — Analysis

4 — Cause

5 — Postscript

image

1

Accident report

  • Date: 30 MAY 1972
  • Time: 07:24 CDT
  • Type: Douglas DC-9-14
  • Operator: Delta Air Lines
  • Registration: N3305L
  • Fatalities: 4 of 4 crew, no passengers on board
  • Aircraft Fate: Destroyed
  • Phase: Approach
  • Airports: (Departure) Fort Worth-Greater Southwest International Airport, TX (GSW) (GSW/KGSW), United States of America; (Destination) Fort Worth-Greater Southwest International Airport, TX (GSW) (GSW/KGSW), United States of America

2

Narrative

image

Flight Track, AAR-73-3, App. D

  • Delta Air Lines, Inc., Flight 9570 (DL9570) departed from Love Field, Dallas, Texas, on a training flight at 0648 on May 30, 1972, and proceeded to the Greater Southwest International Airport (GSW) to perform approaches and landings.
  • A short time later, the pilot requested approval for a landing on Runway 13, behind [a] DC-10 which was inbound on the ILS. The clearance to use Runway 13 was issued with an advisory, “caution turbulence.”
  • On approach to the runway threshold, the DC-9 started to oscillate around its longitudinal axis and then took and extreme roll to the right. The airplane had achieved approximately 90° of roll when the right wingtip contacted the runway surface approximately 1,240 feet beyond the threshold. The airplane continued to roll to a nearly inverted attitude before the main body struck the runway. The fuselage and empennage separated upon impact and slid approximately 2,400 feet, coming to rest off the right side of the runway.
  • The flight was routine until approximately 11 seconds before impact. At that time, the right seat occupant, the Delta check pilot, commented, “A little turbulence here.” The flight data recorder disclosed a corresponding vertical acceleration trace excursion to + 1.7 g. The altitude at this time was approximately 670 feet m.s.1., which is 100 feet above the airport elevation.
  • Data extracted from the cockpit voice recorder disclosed that an attempt was made to “go-around” and that the stall warning system was actuated just prior to impact with the ground.
  • The airplane was destroyed by the impact forces and the general fire which developed subsequent to impact. The four occupants sustained fatal injuries.
  • It was determined from information obtained from the airplane flight data recorders that the DC-9 had traversed the same track as the DC-10 during the final phase of the approach. The time separation between the two airplanes was 53 to 54 seconds.
  • Meteorological information. 0723. Local, 10,000 feet scattered, 25,000 feet thin broken, visibility 30 miles, temperature 67” F., dew point 59’F., wind 340’ 7 knots, altimeter setting 30.03 inches.

Source: AAR-73-3, §1.1


3

Analysis

  • The Delta training flight from Love Field, Dallas, to the Greater Southwest International Airport was routine in every respect.
  • The airplane was performing normally, its systems and powerplants functioning without any difficulty. The airplane’s weight and c.g. were well within prescribed limits. There was no in-flight fire, nor was there any incapacitation of the flightcrew. The airplane was properly equipped for the flight. The aircraft was operated in accordance with a VFR flight plan filed with the company dispatch office. The pilot requested and received an IFR clearance for the flight from Dallas to Greater Southwest Airport. The issuance and acceptance of that clearance placed the flight under the protection provided by the ATC system for aircraft operations under IFR. Upon arrival in the GSW area, the flight was cleared for an ILS approach to Runway 13 and sequenced behind the American Airlines DC-10. The separation between the DC-9 and the DC-l0 was established by radar vectors and exceeded 6 nautical miles. The ILS approach and the subsequent full-stop landing were uneventful. There were no indications that the flight experienced turbulence of any nature or any other difficulty.
  • The IFR clearance issued by ATC at Dallas was terminated by the landing at GSW. To reinstate an IFR clearance would have required a request from the pilot. This was not done. After the next takeoff, the DC-9 was advised by the tower to “Maintain VFR” and to contact approach control.
  • On the second ILS approach, the DC-9 was again sequenced behind the DC-10, with approximately 6 NM separation. Again there was no indication of an encounter with abnormal turbulence. This approach was terminated with a “missed approach,” after which radio contact was reestablished with approach control. The DC-9 requested and received radar vectors to position the airplane for a VOR Runway 35 approach to include a circling approach to land on Runway 17. The approach was initiated, and the crew was advised to contact the tower and to report passing the 5 NM fix. The subsequent tower contact was followed by a clearance for the circling approach to Runway 17.
  • Although the DC-9 was not specifically advised by GSW or approach control that radar service was terminated, it was, in fact, terminated at the time the radio was switched to the tower local control frequency. There is no reason to believe that the flightcrew of DC-9 thought otherwise. Recorded conversation between the check airman and the captain trainee verified the intention of the crew to “watch the traffic.” They subsequently saw both the B-727 and the DC-10 in the traffic pattern for an ILS approach to Runway 13.
  • As the approach continued, the flightcrew of the DC-9 visually assessed the traffic situation and became concerned that the DC-10 landing on Runway 13 would conflict with the planned circling approach to Runway 17. Consequently, they requested a revised clearance for continuation of the circling approach to terminate with a full-stop landing on Runway 13 behind the DC-10. The local controller responded with “Okay that’ll be fine use one three for full stop! Caution turbulence.”
  • Subsequent comments by the crew have been interpreted as referring to a visual assessment of spacing with reference to the DC-10, which was then on final approach. The intra-cockpit conversation, “All right, there’s twenty seconds,” indicates that the DC-9’s downwind leg was extended 20 seconds beyond a position abeam the end of Runway 13. At the time of the 20-second comment, the pilot initiated a left turn to the final approach. At that time, the computed flight track shows that the DC-9 was almost directly abeam the DC-10.
  • The procedure of turning onto the base leg when abeam the preceding aircraft is a common practice for establishing separation from preceding aircraft. At the prescribed approach speed and bank angle, the turn to final approach approximates a standard rate (3°/sec.) turn. After completing the turn 1 minute later, the following aircraft would be in approximately the position occupied by the preceding aircraft when the turn was begun. Experienced pilots, and particularly the FAA air carrier inspector, should have been aware of the 2-minute criterion for separation from “heavy” jets in IFR conditions. This category included the DC-10, although the term “heavy” was not used on the radio.
  • The Board believes that if either the pilots of the DC-9 or the FAA inspector had recognized the hazard of their situation at that time, they would have extended the downwind leg to increase the separation interval. The fact that such an action or recommendation was not taken is attributed to one or more of the following factors:
    1. The pilots and the FAA inspector might have been engrossed in conducting the circling approach in accordance with the procedures specified in the Delta DC-9 operating manual. The DC-9 flightcrew adhered to these procedures explicitly and, in doing so, might not have recognized the hazard of their proximity to the DC-10 with regard to wake turbulence.
    2. The flightcrew and the FAA inspector were aware of the proximity of the “heavy” jet and, although cognizant of the nature of turbulence associated with trailing tip vortices, did not correctly assess the hazard to a DC-9. The evidence indicates an apparent widespread belief that the vortex hazard was a problem for small general aviation aircraft and that, although uncomfortable, it is not dangerous to an airplane of the size and weight of the DC-9. The fact that this is only the second instance in which vortex wake turbulence has been considered a causal factor in the crash of a moderately large airplane lends further support to this misconception.
    3. Finally, the flightcrew’s complacent attitude toward the tower controller’s “caution turbulence” advisory might have resulted from the “cry wolf’ syndrome. This syndrome might well have existed in this case because the crew of the DC-9 had behind the DC-10 without apparent difficulty. Frequent caution advisories without resultant encounter with a vortex may lead pilots to disregard such notices.
  • The turn to the final approach was conducted at the circling approach altitude in accordance with the prescribed procedures. The rollout into the final approach was accomplished at 072256.0 at a position slightly to the right and below the path traversed by the DC-10. The time separation was 55 seconds and the DC-9 was 2.25 NM behind the DG10, which was lifting off following the completion of the touch-and-go landing.
  • The DC-9 was above the influence of the wake turbulence until it was inside the middle marker. The DC-9 flightpath approached the left wingtip vortex of the DC-10 from left to right, descending into the disturbed air. The onset of turbulence was apparent on the flight data recorder vertical acceleration trace at approximately 0723:23, at which time the pilot’s recognition of turbulence was evident by the comment, "a little turbulent here."
  • The vortex encounter simulation shows that the onset of turbulence would have been signaled by a moderate left roll which would probably not have been of serious concern to the flightcrew. The reflex reaction of the pilot would have been to make a right lateral control input in an attempt to maintain a wings-level attitude. This action, once initiated, would have caused the airplane to penetrate deeper into the vortex. The induced left rolling moment continued to increase and at 0723:28.2 the pilots became concerned. An order by the check airman to “go around” was followed immediately by a call for “takeoff power.”
  • At an altitude of approximately 50 feet above the ground, the pilot’s most immediate concern would have been devoted to maintaining level flight. Whether takeoff power was actually applied could not be determined. It is possible that the attention of both pilots was diverted by the roll problem to the point that there might have been some hesitation to remove one hand from the control yoke to effect power lever movement. It is also possible that the power levers were moved forward but that the normal 1ig in engine acceleration delayed the thrust response to maximum r.p.m. Engine acceleration could have been compromised further by transient compressor stalls which could have been caused by the vortex-produced airflow disruption at the engine inlet. There were no indications of engine over-temperature conditions; however, a transient stall might not have produced discernible over-temperature evidence.
  • Whatever the reason, there was no positive indication in the evidence provided by the flight data recorder that the airplane responded to an application of takeoff thrust.
  • At 0723:30, the DC-9 was at a critically low altitude, deep within the influence of the vortex flow field. The stall warning (“stick shaker”) was activated as a result of the high angle of attack, which was induced by the vortex vertical flow component. The pilots were, in all probability, still countering a left rolling tendency by application of right aileron control when the airplane moved into the vortex core. The resultant load reversal would have induced a sharp roll to the right. The pilots would then have responded immediately with a control reversal; however, at this time, the magnitude of the induced rolling moment, and the normal lag in pilot’s reaction, control system, and aerodynamic response were such that recovery was impossible. The DC-9 right wingtip struck the runway surface in an uncontrolled attitude.
  • The Board believes that the actions of the flightcrew were normal for the circumstances. Because of the moderate nature of the initial roll, and possibly because of the uneventful experiences with past encounters with less severe vortices, neither the flightcrew nor the FAA inspector associated the initial turbulence with impending loss of control. As a result, the decision to execute a missed approach was not made in time to avoid this accident.
  • The crosswind component caused the vortex to remain in the runway centerline area by preventing the lateral motion normally produced by ground interaction. The slight tailwind component further aggravated the situation by moving the turbulent air mass back into the runway threshold area.
  • The Board, therefore, concludes that the pilot of the DC-9 accepted the clearance for a visual approach, and that in accordance with effective directives, it was his responsibility to establish separation or institute other vortex avoidance measures. The controller’s responsibility was to advise the pilot of the position of the heavy jet and to issue a caution for wake turbulence.
  • In addition, the Board concludes that the responsibility for vortex avoidance should not be placed solely with the pilot because of the difficulty he has in complying with techniques which require him to visualize an invisible hazard.

Source: AAR-73-3, §2.1


4

Cause

The National Transportation Safety Board determines that the probable cause of the accident was an encounter with a trailing vortex generated by a preceding “heavy” jet which resulted in an involuntary loss of control of the airplane during the final approach. Although cautioned to expect turbulence the crew did not have sufficient information to evaluate accurately the hazard or the possible location of the vortex. Existing FAA procedures for controlling VFR flight did not provide the same protection from a vortex encounter as was provided to flights being given radar vectors in either IFR or VFR conditions.

Source: AAR-73-3, §2.2.b


5

Postscript

The Causes listed in the NTSB report tell us what happened, but not why it happened. We can often get a better sense of that by reading the recommendations:

As a result of the investigation of this accident, the Safety Board on June 30, 1972, issued two recommendations (A-72-97 and 98), directed to the Administrator of the Federal Aviation Administration. Copies of the recommendation letter and the Administrator’s response thereto are included in Appendix G. On July 28, 1972, the FAA issued special instructions to all controllers which called for new and increased separation for all aircraft operating behind the DC-10 or L-1011. Specifically, the new standards required 5 miles spacing for all aircraft, with the exception of the 747 or C5A operating behind the DC-10 and L-1011. Previously, a wide-bodied jet following another “heavy” jet required only 3 miles spacing.

Source: AAR-73-3, §3

Therefore, the National Transportation Safety Board recommends that the Federal Aviation Administration: Reevaluate wake turbulence separation criteria for aircraft operating behind heavy jet aircraft.

Source: AAR-73-3, App. G.

If I were to write the causal section, I would have said that the FAA failed to adequately study the effects of heavy aircraft wake turbulence on all other aircraft types prior to certifying heavy aircraft, and to implement any necessary mitigations. It took this accident to do just that. More about this: Wake Turbulence.

It appears we have learned that lesson but there is another one that remains. After this 1972 accident and the revised wake turbulence separation rules, it was tempting to think that we solved the problem. Years later we discovered the Boeing 757 created a wake unusual for its weight. Years after that we discovered the need to give greater caution to the Airbus A380 and that perhaps there was more to learn. The lesson, I think, is that we are not as smart as we think we are. Just because a group of self-appointed experts devise criteria that proclaim everything is safe, doesn't make it so.

References

(Source material)

Aviation Safety Network

NASA Langley Research Center, URL no longer active.

NTSB Aircraft Accident Report, AAR-73-3, Delta Air Lines, Inc., McDonnell Douglas DC-9-14, N3305L, Greater Southwest International Airport, Fort Worth, Texas, May 30, 1972.