The NTSB blames the crew for failing to set their flaps prior to takeoff, leading to the accident and the deaths of all but one person on the airplane as well as two more on the ground. I think that is definitely true.

— James Albright

image

Updated:

2019-10-16

image

Northwest Airlines N312RC,
Airnikon Collection

The report also speculates that the flight crew might have pulled a circuit breaker to silence a nuisance warning that would have also disabled the takeoff configuration warning. There isn't any evidence of this, but it is the most likely reason for the absent warning.

Finally, the report speculates that they could have survived the flaps-up stall had they properly executed stall recovery procedures.

We aren't here to add to the list of things the pilots did wrong, only to learn from what could have caused the crash so we can avoid a similar fate. So I'll assume all three items are true and apply what I call the don't/don't/do rules:

  1. Don't get busy

    — If you find yourself in a rush it could be that you are behind the airplane and it is time to regroup. This is easier said than done in flight, but it is easier done than said on the ground: set the parking brake. This crew missed a ground frequency change and missed a turn.
  2. Don't get smart

    — The checklist exists for a reason and the minute you are first tempted to skip it things will only get worse. The cockpit voice recorder shows the checklist was not called for or completed.
  3. Do things for a reason

    — Pulling a circuit breaker to silence a warning horn might be easier than correcting the reason the warning horn is sounding, but it sets you up for failure.

1 — Accident report

2 — Narrative

3 — Analysis

4 — Cause

5 — Postscript

image

1

Accident report

  • Date: 16 August 1987
  • Time: 2045
  • Type: McDonnell Douglas DC-9-82
  • Operator: Northwest Airlines
  • Registration: N312RC
  • Fatalities: 6 of 6 crew, 148 of 149 passengers, 2 on the ground
  • Aircraft Fate: Destroyed
  • Phase: Takeoff
  • Airport: (Departure) Detroit-Metropolitan Wayne County Airport, MI (KDTW)
  • Airport: (Destination) Phoenix-Sky Harbor International Airport, AZ (KPHX)

2

Narrative

  • On August 16, 1987, a Northwest Airlines (Northwest) flightcrew picked up a McDonnell Douglas DC-9-82 airplane, N312RC, at Minneapolis, Minnesota, and operating as flight 750, flew the airplane to Saginaw, Michigan, with an en route stop at Detroit Metropolitan Wayne County Airport (Detroit-Metro), Romulus, Michigan, arriving at Saginaw about 1840 eastern daylight time. At Saginaw N312RC became flight 255 and was flown by the same flightcrew which had brought the airplane in. Flight 255, was a regularly scheduled passenger flight between Saginaw and Santa Ana, California, with en route stops at Detroit and Phoenix, Arizona. The flight was to be conducted in accordance with the provisions of 14 Code of Federal Regulations (CFR) Parts 91 and 121. About 1853, flight 255 departed Saginaw and about 1942 arrived at its gate at Detroit- Metro. Except for taxiing past and having to make a 180” turn to return to its assigned arrival gate, the flight to Detroit was uneventful.

Source: NTSB AAR 88/05, ¶ 1.1

The report later notes that besides missing the turn to their arrival gate, they forgot to turn off their weather radar at the gate. (The report speculates the crew may have not completed, or even started, the appropriate checklist.)

  • About 10 to 15 minutes before the flight was due to depart the gate, a company transportation agent brought the flight release package to the airplane. He was met by the first officer who told him that the captain was not on board. The first officer inspected the package which contained the dispatch documents, signed the release, and returned the signed copy to the agent. As the agent left the airplane, he met the captain who had been conducting a walk around inspection of the airplane and showed him the signed copy of the flight release. The captain studied the release, told the agent that it was all right, and thanked him.

Source: NTSB AAR 88/05, ¶ 1.1

The report notes later that the first officer wasn't authorized to sign the release and that this was part of a pattern showing the first officer set the tempo on this crew. The captain was 57-years old, had 31 years experience with several airlines which had been bought or merged their ways into Northwest, and was highly experienced. The first officer was 35-years old and much less experienced. I'm not so sure the first officer set the tone for the crew, it seems more like the captain was a bit laissez-fare (absent from command).

  • About 2029, the final weight tabulation (weight tab) was delivered to the flightcrew. About 2032, flight 255 departed the gate with 149 passengers and 6 crewmembers on board. During the push back, the flightcrew accomplished the BEFORE (engine) START portion of the airplane checklist, and, at 2033:04, they began starting the engines.
  • At 2034:40, after the engines had been started, the ground crew disconnected the tow bar from the airplane, and, at 2034:50, the west ground controller cleared the flight to “taxi via the ramp, hold short of (taxiway) delta and expect runway three center [3C] (for takeoff). . . ." The controller also informed the flightcrew that Automatic Terminal Information Service (ATIS) Hotel (“H”) was now current and asked them if they had the information. The flightcrew repeated the taxi instructions and stated that they had the ATIS information. At 2035:43, the ground controller cleared flight 255 to continue taxiing, to exit the ramp at taxiway Charlie (C), to taxi to runway 3C, and to change radio frequencies and then contact the ground controller on 119.45 Mhz. At 2035:48, the first officer repeated the taxi clearance, but he did not repeat the new radio frequency nor did he tune the radio to the new frequency. Thereafter, the first officer told the captain, “Charlie for three center, right.”
  • ATIS “H” had been transcribed at 2028:35 and was being broadcast at the time of the accident. Examination of the cockpit voice recorder (CVR) recording showed that the flightcrew had not received information “H” before they began to taxi. However, at 2035:18, information “H” began on the first officer’s radio channel, and, at 2035:55, he told the captain that he was leaving the airplane’s No. 1 radio “to get the new ATIS.”
  • About 2025, the tower supervisor began coordination to change Detroit-Metro from a runway 21 configuration to a runway 3 configuration. The change was completed at 2028. ATIS “H” was the first ATIS transcription to contain and broadcast this information. It also described the ceiling and visibility and stated in part that the temperature was 88° F, that the wind was 300° at 17 knots, and that "". . . ILS approaches are in use to runways three left (3L) and three right (3R) departing runways three . . . low level windshear advisories are in effect. . ."
  • The takeoff performance data in the flightcrew’s dispatch package was based on using either runways 21 L or 21 R; however, the flight had been instructed by the ground controller to taxi to runway 3C, the shortest of the three available runways. The final takeoff weight for the airplane was 144,047 pounds. At 2037:08, the captain asked the first officer if they could use runway 3C for takeoff. Because of the runway change, the first officer had to refer to the company’s Runway Takeoff Weight Chart Manual to verify that their takeoff weight was below the allowable limits for runway 3C. The takeoff weight chart showed that with the flaps set at 11°, the maximum allowable takeoff weights for runway 3C at 85°F and 90°F were 147,500 pounds and 145,100 pounds, respectively. After consulting the manual, the first officer told the captain runway 3C could be used for takeoff and the captain concurred with the first officer’s evaluation.

Source: NTSB AAR 88/05, ¶ 1.1

The crew got the correct runway and correctly verified they had the performance necessary to takeoff. But this did distract the first officer for a while.

  • During the taxi out, the captain missed the turnoff at taxiway C. When the first officer contacted ground control, the ground controller redirected them to taxi to runway 3C and again requested that they change radio frequencies to 119.45 Mhz. The first officer repeated the new frequency, changed over, and contacted the east ground controller. The east ground controller gave the flight a new taxi route to runway 3C, told them that ATIS “H” was still current, that windshear alerts were in effect, and that the altimeter setting was 29.85 in-Hg. The flightcrew acknowledged receipt of the information.
  • At 2042:11, the local controller cleared flight 255 to taxi into position on runway 3C and to hold. He told the flight there would be a 3-minute delay in order to get the required “in-trail separation behind traffic just departing.” At 2044:04, flight 255 was cleared for takeoff.
  • The CVR recording showed that engine power began increasing at 2044:21 that the flightcrew could not engage the autothrottle system at first, but, at 2044:38, they did engage the system, and that the first officer called 100 knots at 2044:45.6. At 2044:57.7, the first officer called “Rotate,” and, at 2045:05.1, the stall warning stick shaker activated and continued operating until the CVR recording ended. At 2045:09.1, 2045:11.4, 2045:14.3, and, 2045.17.1, the aural tone and voice warnings of the supplemental stall recognition system (SSRS) also activated. Between 2044:01 and 2045:05.6, the CVR recording did not contain any sound of the takeoff warning system indicating that the airplane was not configured properly for takeoff.
  • Witnesses generally agreed that flight 255’s takeoff roll was longer than that normally made by similar airplanes. They stated that the flight began its rotation about 1,200 to 1,500 feet from the departure end of the runway, agreed that it rotated to a higher pitch angle than other DC-9s, and agreed that the tail of the airplane’came close to striking the runway.
  • Only a few witnesses recalled any details about the position of the airplane’s leading edge wing slats, trailing edge wing flaps, or landing gear. Most of these witnesses said that the landing gear was retracted after liftoff. Two Northwest first officers recalled that the flaps and slats were extended. One first officer was in the airplane directly behind flight 255 in the takeoff sequence. According to her, “the flaps were extended, which is normal, but I could not. . . state the actual degree of flap extension.” She did not describe the position of the slats. The second first officer’s airplane was parked on taxiway “A” between the ramp and taxiway “J.” The airplane was facing runway 3C and about 150 feet from it. He testified that he observed the flaps and slats as flight 255 rolled past his airplane and, “The slats and flaps were extended.” However, he was unable to estimate their degree of extension.

Source: NTSB AAR 88/05, ¶ 1.1

You would think there is no better eye witness than one current and qualified in the same airplane working for the same airline. But expert witnesses are often wrong because they tend to see what they want to see. If you tell them to check the position of the flaps and slats on an airplane they know well, they will do a good job of that. But if you ask them about what they saw about a passing airplane without any hint that something is wrong, they may see what they expect to see.

  • After flight 255 became airborne it began rolling to the left and right. Witnesses estimated that the bank angles during the rolls varied from 15° to 90°. Some witnesses stated that the airplane wings leveled briefly and then banked to the left just before the left wing hit a light pole in a rental car lot. Most witnesses did not see fire on the airplane until it was over the rental car lot. The first officer of the Northwest airplane parked on taxiway “A” testified that flight 255 was intact until the left wing struck the light pole in the auto rental car lot. After the wing struck the pole, he saw what appeared to be “a four- to five-foot chunk of the wing section . . ." fall from the airplane. He did not see any fire on the airplane until after it struck the light pole and then he saw “an orange flame. . ." emanating from the left wing tip section.
  • After impacting the light pole, flight 255 continued to roll to the left, continued across the car lot, struck a light pole in a second rental car lot, and struck the side wall of the roof of the auto rental facility in the second rental car lot. Witnesses stated that the airplane was in a 90° left wing-down attitude when it struck the roof and that it continued rolling and was still rolling to the left when it impacted the ground on a road outside the airport boundary. The airplane continued to slide along the road, struck a railroad embankment, and disintegrated as it slid along the ground. Fires erupted in airplane components scattered along the wreckage path. Three occupied vehicles on the road and numerous vacant vehicles in the auto rental parking lot along the airplane’s path were destroyed by impact forces and or fire.
  • On board flight 255, 148 passengers and 6 crewmembers were killed; 1 passenger, a 4-year-old child was injured seriously. On the ground, two persons were killed, 1 person was injured seriously, and 4 persons suffered minor injuries.

Source: NTSB AAR 88/05, ¶ 1.1

  • Virtually all of the interviewed first officers and other captains who had flown with the captain described him as a competent and capable pilot. They stated that the captain always used the airplane checklist. One first officer stated that the captain had a reputation “as a strict, by-the-book pilot who would not tolerate any deviation from standard procedures.”
  • With the exception of one training report during his early probationary period with the airline, all of the captains with whom the first officer had flown graded his performance as average or above average. Comments contained in some of his grade sheets described him as follows: “competent pilot,” “easy to work with,” “good in all respects,” and “very personable, thorough job. . ." One captain with whom the first officer recently had flown stated that he appeared to be a good pilot. Although he did not remember if the first officer had initiated checklists, he stated that the first officer did not appear to be a “yes man” and that he remembered the first officer handling a very busy period “very well and calling a potential problem [to his] attention.” Other captains who recently had flown with the first officer described his ability and performance in favorable terms.

Source: NTSB AAR 88/05, ¶ 1.5

  • Examination of the airplane flight and maintenance logbooks did not reveal any discrepancies or malfunctions that would have contributed to the accident. In addition, the examination disclosed that, at the time of the accident, there were no discrepancies or malfunctions in the logbooks involving minimum equipment list (MEL) items.
  • According to the Northwest DC-9-82 Airplane Pilots Handbook (APH), the maximum certificated takeoff weight of the airplane is 149,500 pounds. The airplane is limited to a maximum tailwind of 10 knots for takeoff and landing and a maximum demonstrated crosswind of 30 knots for takeoff and landing. The actual airplane weight for the takeoff at Detroit Metro was 144,047 pounds, its computed center of gravity (cg.) for the ensuing takeoff was 9.8 percent of the mean aerodynamic chord (MAC) of the wings and was within the forward and aft c.g. limits of 3.1 percent and 24.4 percent MAC, respectively.
  • The CVR showed that the latest runway temperature information known to the flightcrew was the 88°F reading contained in ATIS “H.” The CVR also showed that the flightcrew planned to use 11° flaps for the takeoff. Based on the 88°F ambient temperature, flaps at 11°, and the slats at the takeoff or mid-sealed position, the company’s takeoff weight chart showed that the maximum allowable takeoff weight for runway 3C was 146,060 pounds and that reduced engine thrust could not be used for takeoff. The required engine pressure ratio (EPR) for the ensuing takeoff would have been 1.95. The takeoff weight charts provided weight corrections based on headwind or tailwind components. On runway 3C, the maximum allowable weights either could be increased by 230 pounds for each knot of headwind or had to be decreased by 960 pounds for each knot of tailwind.
  • The wing trailing edge flap system consists of an inboard and outboard flap segment on each wing. Six fixed position detents are located along the left side of the flap handle, track, or race: UP/RET, 0°, 11°, 15°, 28°, and 40°. When the flap handle is in the UP/RET detent, the flaps and leading edge slats are retracted. When the flap handle is in the 0° detent, the flaps are still retracted, but the slats are extended to the mid-sealed position. Positioning the flap handle to the 15° or higher degree detents will move the slats to the extended position.

Source: NTSB AAR 88/05, ¶ 1.6


3

Analysis

  • The airplane’s maintenance records disclosed that it had been maintained and operated in accordance with applicable Federal aviation regulations and company operations specifications, rules, and procedures. Except for the possible failure of the takeoff warning system to provide an aural warning for an improper takeoff configuration, there was no evidence of any preexisting malfunctions or failures of any airplane structures or systems which would have been a causal factor to the accident.

Source: NTSB AAR 88/05, ¶ 2.1

  • Examination of the CVR and DFDR data readouts showed immediately that the airplane had not encountered a decreasing headwind type of windshear. The DFDR data showed that, at liftoff, the airplane’s airspeed was about 169 KIAS and that instead of decelerating over the last 14 seconds of the flight, the airplane accelerated to about 184 KIAS and climbed about 48 feet. This performance was not consistent with the expected performance of an airplane that is caught in a decreasing head windshear. The fact that the airplane did not encounter a windshear was further corroborated by the lack of divergence between the airplane’s ground speed and indicated airspeed during the time it was airborne.
  • The correlated CVR and DFDR readouts showed that during the 14-second flight, the airplane’s stick shaker remained activated continuously, and its SSRS activated four times. With the flaps at 11° and the slats in the mid-sealed position, the airplane’s stall speed was about 121 KIAS; if the flaps were retracted and the slats remained in the mid-sealed position, the stall speed would increase to 128 KIAS. Despite the fact that the 169 to 184 KIAS recorded during the flight exceeded the worst of the two stall speeds by 36 to 56 KIAS, the stall warnings persisted. The investigation indicated that the only wing configuration that would continue to activate the stall warnings between 169 and 184 KIAS was a wing that was in cruise configuration, i.e., slats and flaps retracted.

Source: NTSB AAR 88/05, ¶ 2.2

The Digital Flight Data Recorder acted as expected, in that the data stream was interrupted whenever the parking brake was set.

  • The DFDR readout showed that the recorded values for the flaps and slats were identical at the beginning and at the end of each of these two data stream interruptions. The recorded values showed that the flaps and slats were in the retracted position and that there was no disagreement between the slat position and the flap handle position. In addition, the DFDR readout showed that, from push back to impact, during the entire period that power was on the DFDR, the flaps were always retracted, the slats were always retracted, and there was no disagreement between the positions of the flap handle and slats.
  • The only position of the flap handle that will place and keep the slats in the retracted position is the UP/RET detent. Moving the flap handle to any other select position on the flap handle track will move the slats out of the retract position to either the mid-sealed or the extended position as the case may be.

Source: NTSB AAR 88/05, ¶ 2.2

  • The flap handle was in the UP/RET position when it was found in the wreckage.
  • The most reliable physical evidence showed that the flaps and slats were retracted and in agreement with the full forward position of the flap handle at the start of the impact sequence.
  • Since only the flightcrew could extend the airplane’s flaps and slats after it was pushed back from the gate, the Safety Board also concludes that the flightcrew did not extend the flaps and slats and did not configure the airplane properly for takeoff. However, the CVR transcript showed that the takeoff warning system, which was designed to warn the flightcrew that the airplane was not configured properly for takeoff, failed to provide the proper warning to the crew.

Source: NTSB AAR 88/05, ¶ 2.4

  • The testimony of a Northwest first officer who rode in the jump seat from Detroit to Saginaw indicated that the takeoff warning system had functioned after the airplane landed at Saginaw.
  • The sound spectrum analysis testing conducted in the Safety Board’s audio laboratory permitted the Board to identify the takeoff warning’s failure mode. Of primary importance to this analysis was the fact that the two SSRS alarms are connected to different power supplies in the CAWS unit: SSRS-2, the first officer’s alarm, was connected to CAWS power supply-3; and SSRS-1, the captain’s alarm, was connected to CAWS power supply-2. The takeoff warning system also was connected to power supply-2.
  • When both SSRSs operate, an echo effect will be heard. The sound spectrum analysis of the actual warning generated by the accident airplane’s CAWS unit showed that there was no echo effects, that only one SSRS had provided the alarm, and that, based on the frequency components of the word, SSRS-2 provided the alarm recorded by the CVR.
  • The evidence showed that the stall alarm was generated from power supply-3 of the CAWS unit’s, and that, based on the facts that the takeoff warning system and SSRS-1 did not operate, power supply-2 of the unit was inoperative.
  • Since the bus was powered and the wiring from the P-40 circuit breaker to the CAWS unit was intact, but power supply-2 of the CAWS unit was not functioning, the process of elimination leads to the only remaining component in the input circuit where a power interruption most logically could occur--the P-40 circuit breaker.
  • Because the P-40 circuit breaker was badly damaged during the accident, it was impossible for the Safety Board to determine positively its preimpact condition. There were three possible conditions that would have caused power to be interrupted at the P-40 circuit breaker: the circuit breaker was intentionally opened by either the flightcrew or maintenance personnel, the circuit breaker tripped because of a transient overload and the flightcrew did not detect the open circuit breaker, or the circuit breaker did not allow current to flow to the CAWS power supply and did not annunciate the condition by tripping.
  • [ . . . ] it was possible that the power supply could have been disabled by the flightcrew for a nuisance warning other than the takeoff warning. The Safety Board cannot rule out this possibility.
  • The second possibility considered was that the circuit breaker opened electrically due to an undetermined transient overload condition, and that the crew did not detect the tripped circuit breaker. In this case, there would be no warning that such a condition existed and the location of the circuit breaker is such that a tripped breaker might not be visually detected, especially in low ambient light conditions. Although flightcrew members normally check the circuit breaker panels on entering the cockpit, the sixth item on the BEFORE START checklist requires a circuit breaker inspection and both crewmembers are required to accomplish this step and are required to respond to the challenge.
  • The third possibility examined was that the P-40 circuit breaker, for undetermined reasons, did not allow current to flow even though the latch appeared mechanically closed to the flightcrew. [ . . . ] The evidence makes it impossible for the Safety Board to rule out that the current flow through the P-40 circuit breaker was inhibited by the presence of a dielectric film on the bus bar contact.

Source: NTSB AAR 88/05, ¶ 2.5

  • The CVR recording showed that the flightcrew neither called for nor accomplished the TAXI checklist. The first item on the TAXI checklist required both pilots, in response to the checklist’s challenge, to check and verify orally that the flaps and slats were positioned correctly. This item was not performed, and the flightcrew did not discover that the airplane was configured improperly for takeoff. The omission of the TAXI checklist was further corroborated by the flightcrew’s inability to engage the autothrottles at the start of the takeoff because they did not, as required by the TAXI checklist place the TCI in the “T.O.” mode. However, they were able to rectify this omission by the time the airplane accelerated to 100 KIAS. The Safety Board concludes that the failure of the flightcrew to accomplish the TAXI checklist in accordance with required procedures was the probable cause of this accident.
  • The Safety Board could not determine conclusively why the first officer did not lower the flaps. Northwest procedures authorized first officers to extend the flaps after the airplane begins to taxi and has cleared the parking ramp and its associated obstacles. The CVR recording showed that at the time the first officer was authorized to extend the flaps, several intervening events might have diverted his attention. Almost immediately after receipt of the taxi clearance and about the time the airplane began moving, the first officer had to select the ATIS radio frequency and listen to and copy the contents of the ATIS message. After receiving the message, he then had to get the takeoff performance chart and verify if they could use runway 3C for takeoff. Thus, the possibility existed that he might have intentionally delayed lowering the flaps, perhaps anticipating a different flap setting due to the runway change. The testimony of and interviews with Northwest flight personnel indicated that the flap extension procedure had become a very strong habit pattern among the DC-9 first officers. As such, the first officer may never have experienced an occasion when he had either inadvertently failed to extend flaps or had failed to extend them when the airplane began taxiing. The habit pattern of extending the flaps may have caused a lessening of his awareness of the omission, because by the time the first officer completed copying the ATIS message and analyzing the takeoff weight data, the airplane had taxied well beyond the point where he would have routinely extended the flaps. Based on this well developed habit pattern of extending the flaps, the first officer might have believed that this task, which was always completed shortly after the captain began to taxi or by the time the airplane departed the terminal ramps, had been completed as it always was.
  • The flap extension procedure did not require the captain to be either notified or to approve repositioning the flaps and slats. Therefore, unless he happened to either observe the first officer move the flap handle, or observe the movement of the flap indicator or the illumination of the slat advisory lights, he would not know that the procedure had been accomplished.
  • The Safety Board believes that it is the captain’s responsibility to structure the manner in which his crew will accomplish its duties. While he must be open to information input from his crew, he must set the tone for how this information will be proffered. Except for the BEFORE START CHECKLIST, he did not call for any of the other checklists nor did he point out to the first officer that checklists were not being accomplished in accordance with company procedures. After push back, the captain initiated three conversations which were not germane to duty requirements and which diverted the crew’s attention from task-related activities.
  • The evidence indicated that the first officer was either given, or assumed he had been given, the duties of leading the crew’s task-related activities up to and including the signing of the flight release, a responsibility assigned to the captain by regulation. While it is possible the captain intended to discuss this problem with the first officer, he made no move to point out to the agent, for the agent’s future knowledge, that only the captain is authorized to sign the release. The first officer’s assumption of the role of leader placed him in a position of structuring the crew’s approach to activities while at the same time trying to satisfy the captain that he was carrying out his subordinate role in a satisfactory manner. In the area of checklist initiation, the first officer’s assumption of initiation responsibilities greatly increased his work and planning load and relegated the captain’s function to that of observer. The evidence also indicated that deference by a captain to a first officer also can inhibit crew effectiveness because the captain cannot presume that the first officer will always assume all of the captain’s responsibilities. The captain appears to have become dependent upon checklist initiation by the first officer instead of on his own active initiation responsibilities. Therefore, when the first officer became distracted, the captain’s passive involvement with checklist initiation did not provide a backup to the first officer’s memory.

Source: NTSB AAR 88/05, ¶ 2.6

The report notes this crew made several procedural error from the time of their arrival until takeoff, each constituting a deviation. The report concludes:

  • In essence, when these deviations are assessed together with the flightcrew’s checklist performance, the Safety Board believes that their performance was below the standards of an air carrier flightcrew.

Source: NTSB AAR 88/05, ¶ 2.6


4

Cause

The National Transportation Safety Board determines that the probable cause of the accident was the flightcrew’s failure to use the taxi checklist to ensure that the flaps and slats were extended for takeoff. Contributing to the accident was the absence of electrical power to the airplane takeoff warning system which thus did not warn the flightcrew that the airplane was not configured properly for takeoff. The reason for the absence of electrical power could not be determined.

Source: NTSB AAR 88/05, ¶ 3.2

References

(Source material)

NTSB Aircraft Accident Report, AAR-88/05, Northwest Airlines, Inc., McDonnell Douglas DC-9-82, N312RC, Detroit Metropolitan Wayne County Airport, Romulus, Michigan, August 16, 1987