the learning never stops!

Delta Air Lines 191

Accident Case Study




Figure: Typical microburst wind field, from NTSB Report, Figure 4.

Accident Report

  • Date: 02 AUG 1985
  • Time: 18:05
  • Type: Lockheed L-1011 TriStar
  • Operator: Delta Air Lines
  • Registration: N726DA
  • Fatalities:8 of 11 crew, 126 of 152 passengers, 1 on the ground
  • Aircraft Fate: Destroyed
  • Phase: Approach
  • Airport: (Departure) Fort Lauderdale International Airport, FL (FLL/KFLL), United States of America
  • Airport: (Destination) Dallas/Fort Worth International Airport, TX (DFW/KDFW), United States of America


[NTSB AAR 86-05, ¶1.1]

  • A t 1743:45, Fort Worth ARTCC cleared flight 191 to descend to 10,000 feet, gave it a 29.92 in Hg altimeter setting, and suggested that the flight turn to a heading of 250’ “to join the Blue Ridge ‘zero one zero radial inbound and we have a good area there to go through.” The captain replied that he was looking at a “pretty good size” weather cell, “at a heading of two five five . . . and I’d rather not go through it, I’d rather go around it one way or the other.” Fort Worth ARTCC then gave the flight another heading and stated “when I can I’ll turn you into Blue Ridge, it’ll be about the zero one zero radial.” At 1746:50, the center cleared flight 191 direct to Blue Ridge and to descend to 9,000 feet, and flight 191 acknowledged receipt of the clearance.
  • At 1748:22, the captain told the first officer, “You’re in good shape. I’m glad we didn’t have to go through that mess. I thought sure he was going to send us through it.” At 1751:19, the flight engineer said, “Looks like it’s raining over Fort Worth.” At 1751:42, Forth Worth ARTCC instructed flight 191 to contact DFW Airport Approach Control (Regional Approach Control), and at 1752:08, the flight contacted approach control stating that it was descending through 11,000 feet and had received ATIS Information Romeo. At 1756:28, Regional Approach Control’s Feeder East controller transmitted an all aircraft message which was received by flight 191. The message stated in part, “Attention, all aircraft listening . . . there’s a little rainshower just north of the airport and they’re starting to make ILS [instrument landing system] approaches . . . tune up one oh nine one for one seven left.”
  • At 1759:47, the first officer stated, "we're gonna get our airplane washed,” and at 1759:54, the captain switched to Regional Approach Control’s Arrival Radar-l (AR-1) frequency and told the controller that they were at 5,000 feet. At 1800:36, the approach controller asked American Air Lines flight 351 if it was able to see the airport. (Flight 351 was two airplanes ahead of flight 191 in the landing sequence for runway 17L.) Flight 351 replied, “As soon as we break out of this rainshower we will.” The controller then told flight 351 that it was 4 miles from the outer marker, and to join the localizer at 2,300 feet; the controller then cleared the flight for the ILS approach to runway 17L. Ail of the transmissions between the controller and flight 351 were recorded on flight 191's CVR.
  • At 1800:51, the approach controller asked flight 191 to reduce its airspeed to 170 knots indicated (KIAS), and to turn left to 270°; flight 191 then acknowledged receipt of the clearance. Flight 191 had been sequenced behind a Learjet Model 25 (Lear 25) for landing on runway 17L.
  • At 1802:35, the approach controller told flight 191 that it was 6 miles from the outer marker, requested that it turn to 180° to join the localizer at or above 2,300 feet, and stated, “cleared for ILS one seven left approach.” The flight acknowledged receipt of the transmission. At 1803:03, the approach controller requested flight 191 “to reduce your speed, to one six zero please,” and the captain replied, “Be glad to.” Thereafter, at 1803:30, he broadcast, “And we’re getting some variable winds out there due to a shower . . . out there north end of DFW.” This transmission was received by flight 191, and at 1803:34, the CVR’s cockpit area microphone (CAM) showed that an unidentified flightcrew member remarked, “Stuff is moving in.”
  • At 1803:46, the approach controller requested flight 191 to slow to 150 KIAS, and to contact the DFW Airport tower. At 1803:58, the captain, after switching to the tower’s radio frequency, stated, “Tower, Delta one ninety one heavy, out here in the rain, feels good.” The tower cleared the flight to land and informed it, “wind zero nine zero at five, gusts to one five.” At 1804:07, the first officer called for the before-landing check. The flightcrew confirmed that the landing gear was down and that the flaps were extended to 33°, the landing flap setting.
  • At 1804:18, the first officer said, “Lightning coming out of that one." The captain asked, “What,” and the first officer repeated “Lightning coming out of that one.” The captain asked, “Where,” and at 1804:23, the first officer replied, “Right ahead of us.”
  • Flight 191 continued descending along the final approach course. At 1805:OS the captain called out "1,OOO feet.” At 1805:19, the captain cautioned the first officer to watch his indicated airspeed and a sound identified as rain began. At 1805:21, the captain warned the first officer, “You’re gonna lose it all of a sudden, there it is.” At 1805:26 the captain stated, “Push it up, push it way up.” At 1805:29, the sound of engines at high rpm was heard on the CVR, and the captain said “That’s it.”
  • At 1805:44, the Ground Proximity Warning System’s (GPWS) g/ “Whoop whoop pull up” alert sounded and the captain commanded “TOGA". At 1805:48 and 1805:49, two more GPWS alerts were recorded. At 1805:52 a sound similar to that produced by a landing airplane and the sound of the takeoff warning horn were recorded. At 1805:56, the local controller in the tower told flight 191 to “go around,” and the CVR recording ended at 1805:58.


[NTSB AAR 86-05, ¶1.7]

  • The NWS terminal forecast for the DFW Airport pertinent to the accident indicated a slight chance of a thunderstorm with a moderate rain shower. The NWS area forecast pertinent to the accident called for isolated thunderstorms with moderate rain showers for northern and eastern portions of Texas. The terminal forecast was issued by the NWS Forecast Office in Fort Worth, Texas, and the area forecast was issued by the National Aviation Weather Advisory Unit in Kansas City, Missouri.
  • There were no SIGMETS, convective SIGMETs, Severe Weather Warnings, Local Aviation Warnings, Severe Weather Watches, or Center Weather Advisories (CWA) in effect for the time and area of the accident.
  • Flight 191 was third to land behind flight 351 (a Boeing 727) and a Learjet 25; American flight 539 (a McDonnell Douglas MD-80) was to land behind flight 191.
  • The captain of flight 351 testified that he had been directed to execute a missed approach because the airplane ianding ahead of him had not been able to clear the runway in time. During his approach to the DFW Airport, the captain said he saw only scattered clouds and one “thunderstorm northeast of the field.” He said that his Bendix monochromatic weather radar was set in contour mode and the cell did not contour. He could see the cell from the cockpit and “it looked harmless . . . like showers." The captain testified that after passing the outer marker inbound he did not encounter any rain or turbulence, and he did not see any lightning. After the missed approach, flight 351 was vectored to the downwind leg for runway 17L and sequenced into the traffic flow for another approach. After turning on base leg at 2,500 feet, the flight encountered a windshear and lost about 20 KIAS traversing the area of the shear.
  • Flight 351 was cleared for its second approach to runway 17L at 1800:38 and landed about 1804. The captain testified that he did not go through any weather cells and that, while on final, the nearest one was about 2 miles east of his aircraft. The captain said that, after departing the outer marker inbound he encountered heavy rain which lasted until he descended through 1,000 feet. He did not encounter any turbulence or windshear, and he did not see any lightning during the approach.
  • The Learjet preceding flight 191 in the landing sequence had a Primus model 400 color weather radar. The pilot stated that he used the radar until he was about 25 nmi from DFW Airport and that “nothing looked bad.” He was able to see the cells visually. At the public hearing, he testified that he saw this “little buildup*’ as he approached the airport, and that “it looked harmless.” Although his weather radar was still on, he did not recall looking at his radar as he turned on the final approach course.
  • About 1803, as the Learjet approached the outer marker, the pilot retarded power to decelerate the airplane from 170 to 153 KIAS, the maximum flap extension speed. At 153 KIAS, with power still retarded, he extended the landing gear and flaps and placed the airplane into its landing configuration. While the flaps and landing gear were extending, the airspeed dropped from 153 to 125 KIAS. Since the airplane’s power had been reduced %onsiderablyf’ to slow it from 170 KIAS, and since he had not added power while the flaps and landing gear were in transit, the pilot testified that he did not perceive the deceleration from 153 to 125 KIAS to be the result of a windshear encounter.
  • With regard to reporting the weather to the tower, the captain testified that he had nothing to report, “the only thing that we encountered was the heavy rain.”
  • Flight 539 was the next airplane behind flight 191 in the landing sequence. Flight 539 was equipped with a Bendix model RDR-4A color radar which, in the opinion of the captain, was “generally a very effective radar.”
  • The captain said that they maintained visual contact with flight 191 until it entered the rain shower beneath the buildup. He estimated that flight 191 was about 800 feet AGL when it entered the rain, and he also saw lightning in the area where he lost sight of flight 191. His first officer stated that a cell “with abundant lightning” was directly off the approach end of runway 17L and he saw flight 191 “penetrate the cell.”
  • Although the captain of flight 539 testified that, based on what he had observed visually and on his radar, he was considering rejecting the approach, he continued inbound on the approach until, at 1806:21, the local controller requested flight 539 to “go around.”

[NTSB AAR 86-05, ¶1.10.1]

  • The Low Level Wind Shear Alert System (LLWAS) at the DFW Airport was operational at the time of the accident. The system, which has no recording capability, consists of six 20-foot-high vector-vane type of sensors located strategically throughout the airport property. The northeast, southeast, southwest, west, and northwest sensors are located on the airport perimeter; the centerfield sensor is located about 4,463 feet south of the thresholds of runways 17L and 17R and midway between the two runways. The northeast and northwest sensors were nearest to the storm and are located about 3,000 feet north of the thresholds of runways 17 left and right and 18 left and right, respectively.
  • The LLWAS has several limitations: winds above the sensors are not detected; winds beyond the peripheral sensors are not detected; updrafts and downdrafts are not detected; and if a shear boundary happens to pass a particular peripheral sensor and the centerfield sensor simultaneously, an alarm will not occur. In addition, the dimensions of some meteorological phenomena-- microbursts or macrobursts--may be smaller than the spacing between the sensors and thus may not be detected. However, since the downward flow in macrobursts and microbursts turns horizontally as it approaches the ground, an outward flowing shear boundary is established which eventually affects one of the sensors and places the system in alert. The controllers in the DFW Airport tower cab stated that the LLWAS went into alert either about the time the storm reached the north end of the airport or about 10 to 12 minutes after the accident, and when they checked the display, all sensors were in alarm.

[NTSB AAR 86-05, ¶2.4]

  • At 1759:37, flight 191 was about 7 nmi northeast of the cell and was requested to turn right to 340’. Between 1751 and 1800, the cell had intensified from a VIP level 1 to VIP level 4, and flight 191’s nose was pointed at the cell until 1759:37. Except for a period between 1755:53 and 1757:19 during which a portion of the checklist was being , completed, the flightcrew was relatively free of in-cockpit duties. During this period the flightcrew would have been free to use the weather radar to scan the cell and to manipulate the antenna tilt to acquire the best possible radar picture. Since the storm celI had reached a VIP level 4 by 1800, the cell would have reached contouring levels of intensity for their radar sometime during this period. However, the CVR contains no conversation referring either to what was or was not displayed, difficulties involved with manipulating the radar antenna tilt, or the inadequacies of the radar in this area of flight.
  • The Safety Board is concerned that the present training within the industry for windshear encounters on the final approach seems to advocate the philosophy that the retrieval of the approach profile is the desired end result and not escape from the environment. For example, the landing windshear procedures in the Delta L-1011 POM advised the pilot “to be prepared to apply thrust immediately to maintain a minimum of Vref when encountering the shear and to be prepared for a prompt reduction of thrust once normal target speed and glide path is reestablished.” The Safety Hoard believes that training should emphasize that in an environment wherein extreme pitch attitude changes and large applications of engine thrust are required to maintain altitude and minimum airspeeds, flightcrews should be taught that the only objective of the procedure is to escape and thereafter place the maximum distance between the ground and the airplane as soon as possible.
  • The first officer was apparently able to apply the above techniques to keep the airplane on the ILS glideslope as it passed through and beyond the initial portion of the microburst. When the airplane descended into the vortex, the combination of an airspeed loss of 20 KIAS and a strong updraft most likely caused a momentary (l-second) activation of the stickshaker. The Safety Board believes that the first officer acted reflexively when the stickshaker activated to exert a 20- to 25 pound forward push on the control column. This control column force and the longitudinal stability of the airplane resulted in the airplane nosing over to a -8.5’ pitch attitude, a rapid departure from the ILS glideslope, and a descent rate which approached 5,000 fpm for an instant.
  • The NASA analysis of alternate flight paths showed that ground impact might have been avoided had the pushover force not been applied. However, the Safety Board recognizes that the airplane was in an extremely turbulent environment, and because of the rapid reversals of the vertical winds, the airplane was subjected to rapid changes in angle of attack, longitudinal pitch forces, and fluctuations of indicated airspeeds. Consequently, under these circumstances, the ability of the first officer to apply an optimum or recommended pitch control technique would have been subjected to a severe test.
  • The flightcrew had applied maximum thrust shortly before the airplane departed rapidly from the glideslope, and the captain called for TOGA within 3 seconds of glideslope departure. When TOGA was engaged, the command bars presented a "fly-up" command, and the airplane pitched upward in response to the first officer’s application of a substantial nose-up control correction. During this period, the vertical wind changed from a 40-fps downdraft to a 10-fps updraft. The reversal in wind component combined with a substantial nose-up pitch rate increased the angle of attack rapidly. At 1805:48, 3 seconds after TOGA was engaged, a +2.0 g vertical acceleration was recorded and the stickshaker probably again activated for about 1 second. At 1805:50, the airplane began to pitch down. During this time, the magnitude of the “fly-up” command presented by the command bars had decreased; however, they were still presenting a “fly-up” command when the airplane began to pitch down. The data contained in the performance analysis and the flight director study do not permit the Safety Board to conclude that the first officer was “flying the command bars” during the short time that the TOGA Mode was engaged. The data suggest that, in response to the stickshaker, the first officer ignored the command bars and applied nose-down control to silence the stickshaker. The data also show that when the stickshaker activated, the airplane’s pitch attitude was 6° nose-up, the airspeed was about 150 KIAS, and the airplane was accelerating. Consequently, had the first officer been able to match the airplane’s pitch attitude with the command bar position, the airplane might have cleared the ground. The shallow tire marks in the soft ground about 1 mile before the runway 17L threshold indicates a rather mild touchdown and additional evidence that the airplane’s descent had almost been arrested. However, because of the uncertainties in the dynamic wind analysis, and in further recognition of the turbulent environment affecting the flightcrew, the Safety Board cannot conclude that other pilots would have been able to avoid ground contact. The Safety Board believes, however, that avoidance of ground contact could only have been assured positively if the missed approach had been executed when the captain perceived the first indications of a microburst windshear, when the airplane was between 700 and 800 feet AGL.
  • Regardless of the first officer’s response to the command bars, the flight director’s TOGA mode did not provide optimum pitch command guidance for penetrating windshears. In this instance, 1.25 Vs was about 131 KIAS and stickshaker activation speed was about 111 to 113 KIAS. The TOGA logic was designed to maintain 1.25 Vs and, therefore, would present pitch command guidance that would sacrifice altitude to maintain 131 KIAS, even though that airspeed was well above stickshaker activation airspeeds. The sacrifice of altitude to maintain airspeed is contrary to present windshear penetration doctrines and, in this instance, it sacrificed the climb performance which was available down to and at stickshaker speed. The Safety Board notes that other air carriers have cautioned against the use of the TOGA mode during takeoff and go-arounds during windshear encounters; however, the Delta L-1011 POM provided no guidance regarding the limitations of the flight director system TOGA mode under such circumstances.

Probable Cause

[NTSB AAR 86-05, ¶2.4] The National Transportation Safety Board determines that the probable causes of the accident were the flightcrew’s decision to initiate and continue the approach into a cumulonimbus cloud which they observed to contain visible 1ightning; the lack of specific guidelines, procedures, and training for avoiding and escaping from low-altitude windshear; and the lack of definitive, real-time windshear hazard information. This resulted in the aircraft’s encounter at low altitude with a microburst-induced, severe windshear from a rapidly developing thunderstorm located on the final approach course.

See Also


Radar Techniques


NTSB Aircraft Accident Report, AAR-86/05, Delta Air lines, Inc., Lockheed L-1011-385-1, N726DA, Dallas/Fort Worth International Airport, Texas, August 2, 1985

Revision: 20120401