Aircraft Accidents and Lessons Unlearned LIII: Northwest Airlines Flight 255

On August 16, 1987, at about 19:45 (7:45 PM) Eastern Standard Time, Northwest Airlines flight 255 (NWA255) crashed shortly after taking off from Runway 03 Center at Detroit Metropolitan Wayne County Airport in Romulus, Michigan (MI). The Douglas DC-9-82 aircraft, registration number N312RC, failed to climb out before striking light poles, a building’s roof, then the ground. The aircraft broke up as it slid, never veering from its takeoff heading.

The same pilots were flying this third leg of four bound for Phoenix, Arizona, that originated in Minneapolis, Minnesota. The National Transportation Safety Board (NTSB) assigned NWA255 accident number DCA87MA046; Accident report AAR-88/05, published on May 10, 1988, stated, “… that the probable cause of the accident was the flight crew’s failure to use the taxi checklist to ensure that the flaps and slats were extended for takeoff.”

The probable cause was correct. Examination of the cockpit voice recorder (CVR) showed that during taxi out on departure, the flight crew did not run the preflight checklist and, based on the flight control positions found after the accident, did not extend flaps and slats for takeoff. Unlike the stabilizer trim being set without verbal confirmation (the sound of the stabilizer trim ‘in-motion’ horn recorded on the CVR), there was no aural indication that flight controls were extended. From here the report should have examined why the pilots did not run the preflight checklist, to get to the Root Cause of the accident, whether it was a failure on the airline’s procedures; a lack of productive check rides or even something human factors related. However, accident investigators failed to pursue Root Cause.

The report’s reference to the takeoff configuration warning system (TCWS) and its apparent electrical failure diverted attention away from the sole probable cause: pilot failure to use the checklist. As AAR-88/05 stated in the Probable Cause, “Contributing to the accident was the absence of electrical power to the airplane takeoff warning system which thus did not warn the flight crew that the airplane was not configured properly for takeoff. The reason for the absence of electrical power could not be determined.” This was a coincidence that the pilots’ break with procedure aligned with a system that did not function correctly, but should the TCWS have been a contributing factor or a Finding?

Using the accident number, DCA87MA046, the NTSB docket page was queried for Field Notes and Team Lead Factual Reports, especially the Systems Lead’s TCWS notes. The Docket Search Result, however, was “Zero Dockets” as was the docket search result for Delta Airlines flight 1411’s accident report notes, which will be discussed later.

To be clear, the TCWS is a redundant system; its purpose is to alert the pilots of the misconfiguration. TCWS is not designed to remind pilots nor is it a hazard warning in the course of normal flight, e.g., terrain or pending midair collision warnings. Although, the TCWS warns when procedures are not followed, it should never be relied upon to sound. TCWS is a last resort, designed to never be used. It was not an accident cause nor was it a contributing factor. The NWA255 pilots’ failure to run the preflight checklist was the only Probable Cause. The checklist failure, however, was not a Root Cause.

This distinction is important because responsibility – in some cases, as in NWA255, sole responsibility – needs to be defined. If we are to learn the true lessons from aircraft accidents, we must ignore unrelated distractions and narrow the initiates down to root causes, otherwise the lessons are clouded. In this case, the cause was procedural; the NTSB focused attention away from the pilots and called for technology to fix the problem, clouding the problem even further. This diminished pilot skills, their responsibility given to technology. Pilots became more obsolete.

What was the TCWS? The TCWS in the NWA255 DC-9-32 was a mechanical system that employed a series of switches and sensors that reacted to the position of, e.g., the engines’ throttle cables, landing gear sensors, cable or hydraulically driven flight control drive units. TCWS could also work during the landing cycle, assuring flight controls and landing gear were in position for approach and landing. Compared to today’s digital TCWS, the warning system was very rudimentary, but effective. If the flaps, slats, landing gear and/or spoilers were out of takeoff configuration, an alarm horn would sound. It was unlikely that the TCWS was a deferrable item; if non-functional, the plane was not airworthy until repaired.

However, if the TCWS were recognized as a cause, NTSB investigators made a significant omission by not interviewing the Minneapolis maintenance crew and another error by ignoring the maintenance history. It was a critical mistake that Maintenance information and research was absent. Had the throttle cables recently been replaced; the flaps rigged; the landing gear time-changed? Had the mechanic in Minneapolis moved the throttles forward to see if the horn would sound? Was there a preflight inspection conducted by Maintenance and, if not, why not? This could have led to proactive recommendations.

Which is why the NTSB’s fixation with an alleged TCWS electrical system power loss was odd. The docket was empty of any enlightening data related to what the NTSB Systems engineer proved or if he/she was looking in the right place. During the aircraft breakup, sensors were jolted out of place; cables became excessively stretched or broken. It also, unless circuit breakers were physically open, raised the question: How did the Systems engineer determine that power was not available to the TCWS? More importantly, why was the supposed electrical power loss considered a contributing factor?

This is why going off on tangents was wrong. It was dangerous to divert resources and attention away from the investigation path. Additionally, the Probable Cause was for serious information. Investigations that branch off into unrelated departures, leaves the correct causes to be diminished, to get lost in the minutiae of other theories. Speculation should be raised in the investigation’s Analysis phase; if it cannot be proven it should not make the accident report and should be edited out with other theories. If the Systems investigator gave credence to phantom electrical problems, why not question a TCWS sensor design; a switch location; a throttle rigging procedure? All of these components could have just as easily affected the TCWS warning horn. Where would speculation end?

The investigator spent four pages of the report talking about an electrical problem that could neither be found nor proven even existed. In those four pages the investigator did not move the investigation forward, nothing productive was learned. Was the investigator-in-charge unable to bring the conversation back to facts?

It was unclear from the accident report, whether TCWS function was confirmed by the flight crew since the CVR transcript began after the pilots conducted their pre-flight. Any aural warning checks performed by the flight crew during their pre-flight were omitted from the CVR transcript. Even if the CVR transcript recorded the pilots’ pre-flight check, the TCWS could have malfunctioned during pushback or taxi-out. During a review of the transcript, there were five unidentified identical noises titled: “((sound of click))” that occurred between 20:43:11 and 20:44:39 as the aircraft was powering up for takeoff and running down Runway 3-Center. Was this the TCWS aural warning trying to function? The investigator never identified this clicking sound.

It was unfortunate that time was wasted on tangents. Proactive measures could have been worked out with the Federal Aviation Administration (FAA) to identify the Root Causes of the pilots’ failure to follow procedures. On August 31, 1988, 381 days after the NWA255 accident (113 days after AAR-88/05 was adopted), Delta Airlines flight 1141 (DAL1411) crashed during takeoff at Dallas-Fort Worth International Airport. The DAL1411 accident report, AAR-89/04, stated as the Probable Cause, “(1) the Captain and First Officer’s inadequate cockpit discipline which resulted in the flight crew’s attempt to take off without the wing flaps and slats properly configured.”

Following AAR-88/05’s adoption, the FAA issued Air Carrier Operations Bulletin #8-88-4 in June 1988; the Delta Certificate Management Office received Bulletin 8-88-4 on August 30, 1988, and Delta received the bulletin on September 5, 1988, five days after the DAL1411 accident. Bulletin 8-88-4 directed an airline’s FAA Principal Operations Inspector to review, “… overall takeoff warning system performance … and ensure that the checklists appropriately support required crew actions …” Checklist non-compliance was the fundamental cause of NWA255 and, unfortunately, DAL1411. With NWA255 having occurred over a year prior, its investigators might have prevented DAL1411, whose pilots did not run the preflight checklist, just like NWA255’s pilots.

AAR-88/05’s recommendations were toothless, as if the NTSB called them in. Expedite the issuance of guidance materials? What does that mean to the industry? Have Principal Operations Inspectors emphasize the importance ‘disciplined application of standard operating procedures’; ensure training includes cockpit resource management? These were not recommendations; they were the restating of the obvious. If Northwest Airlines messed up their required training, then say so; state it clearly.

Finally, two points about why analysis of 34-year-old accidents is important, especially when digital aircraft monopolize our skies. First, the NTSB has not changed its approach to accident investigation procedures; they still employ decades-old practices, e.g., using engineers with zero industry experience, who do not understand airline culture. That is why all Maintenance issues were missed. Second, the NTSB did not get NWA255 right. What improvements, then, have been implemented into the aviation industry today? If the fixes were wrong in 1988, are we any safer in the 21st century?