Aircraft Accidents and Lessons Unlearned XXXVI: Midwest Express Flight 105

On September 6, 1985, Midwest Express flight 105, a twin-engine Douglas DC-9-14, registration number N100ME, crashed while taking off out of General Billy Mitchell Field in Milwaukee, Wisconsin (accident number DCA85AA036). As per National Transportation Safety Board (NTSB) report AAR-87/01, the flight was the victim of an accelerated stall after reaching 450 feet of altitude. The aircraft suffered what the report called, an uncontained failure (UF) of the number (#) 2 (right) engine, before rolling almost ninety degrees. The loss of lift led to impact.

The Probable Cause, per the report, “was the flight crew’s improper use of flight controls in response to the catastrophic failure of the right engine during a critical phase of flight, which led to an accelerated stall and loss of control of the airplane. Contributing to the loss of control was a lack of crew coordination in response to the emergency.” What were missing in the report were supporting facts.

After I posted Lessons Unlearned XXXIV: Avianca Flight 52, a former NTSB colleague of mine said that, “I was unfair to the Board; that I was hard on the NTSB investigators.” I never intended to be ‘unfair’ or ‘hard’ on any of the investigative agency reports I have reviewed, foreign or domestic. Instead, my focus has been – and will always be – on aviation safety and the improvement of safety by drawing attention to glaring mistakes that were missed; glaring mistakes that may have contributed to later accidents.

1st mistake: Per AAR-87/01, (no NTSB Archive information available) the #2 engine’s UF occurred at the 9th and 10th stage high pressure compressor; the high-pressure compressor spacer failed at takeoff power. The investigator reported that the sleeve had been reworked because its air seal was damaged. Regrettably, the investigator did not verify the process followed to repair the air seal was accomplished per the manufacturer’s instructions or whether the manufacturer would even approve the air seal repair.

2nd mistake: The NTSB concluded that the accident did not result from engine failure, but from the pilots’ actions, i.e. Pilot Error. The NTSB concluded that limited crew communications led to the accident. They implied the pilots intentionally flew into an accelerated stall. Both probable guesses were unlikely.

Both pilots were proficient in the DC-9; their training was current; a majority of their experience was in the DC-9. The cockpit voice recorder (CVR) transcript did not suggest otherwise. The flight began routinely; conversation was professional and friendly. ADDED NOTE: the flight attendant (FA) deserved credit for admirable calm under stress. As the plane stalled, the FA gave controlled orders to the passengers to assume a crash position. The FA was a professional to the end.

The DC-9 aircraft was designed to be flown on one engine, even continued, safe flight in the event of a single engine failure on takeoff. There was an experienced crew, clear weather and a well-designed aircraft with hundreds of thousands of hours of proven reliability in operation. How and why would the pilots intentionally place the airliner in an unrecoverable stall? From the reported facts, they didn’t. Several inconsistencies in the report demonstrated inexperienced investigator problems.

3rd mistake: The CVR and the FDR both showed a data ‘gap’ that occurred simultaneously with a loud ‘clunk’ (identified as the #2 engine’s UF). The NTSB report said the incredible coincidence was from “a jump in the foil [recording tape] position of the recorder”. However, per the CVR transcript, at takeoff, the #2 engine generator powered the recorders. When the #2 engine failed, the #1 engine took over the electrical load – as designed – resulting in a momentary gap. There was no ‘jump in the foil’ tape.

4th mistake: The #1 engine’s power reduction was ignored; this was an inexcusable oversight. Normally the #1 engine throttle would have been advanced – not retarded – to fly out of the emergency. Per the CVR, the pilots never said they advanced the #1 the throttle or not. Was the #1 engine fuel line cut by sharp metal flung out by the #2 engine UF? Investigators never looked at this. Instead, in the absence of FDR engine data, the NTSB made ambiguous CVR interpretations to draw conclusions. In addition, the CVR transcript never suggested the crew lacked coordination or that they deliberately flew the aircraft into a stall. The accident report’s Probable Cause was based on inexperienced speculation, not facts.

5th mistake: What proof was there that any liberated metal penetrated the fuselage? On page 12 of AAR-87/01, section 1.11.5, “the FDR indicated an excessive increase in the climb-rate …”. The NTSB explained this away as an FDR and side-slip induced errors. However, any breach in the fuselage’s pressure vessel would have equalized the cabin pressure to outside pressure, resulting in … an increase in climb-rate. An investigator with a maintenance background would have known this.

6th mistake: The emergency event lasted 15.5 seconds, during which the NTSB incorrectly speculated about a ‘lack of crew coordination’. At 4.5 seconds after the loud ‘clunk’, the Captain (the pilot flying) asked, “What do we got here, Bill?” before saying 3.1 seconds later, “Here …”. Bill, the first officer (FO), transmitted to air traffic control (ATC) that there was an emergency (at 7.6 seconds). This was confusion, not disorganization. Clearly the pilots were puzzled by multiple conflicting instrument readings. Why would both engines lose power? In the remaining seven seconds, did they see hydraulic pressure drop? Did fuel pressure go to zero? Maybe split elevators? Did the aircraft slowly begin rolling? They did not even have enough time to reach for the emergency procedures. Most likely the pilots suffered from astonishment in an untrained for event. They did not suffer from cockpit mismanagement.

7th mistake: What would hydraulic pressure loss or flight control jamming have to do with the #2 engine UF? The Midwest Express 105 accident occurred less than four years before United flight 232 and eleven years before Delta flight 1288; both of these accidents resulted from uncontained engine failures. They demonstrated the catastrophic damage that engine parts, e.g. compressor blades or internal components, could do to the fuselage and systems in line with the separated component’s trajectory. From AAR-87/01, investigation into the #2 engine’s cowls, mounts, the vertical or horizontal stabilizers, was rudimentary; the factual information documented in the report about such damage was limited.

The engines each provide hydraulic pressure from a hydraulic pump mounted on each engine’s accessory gear box. It was more likely that numerous compressor blades were suddenly slung outwards during takeoff power, not in one direction but in many directions, ricocheting off structure; they would have sliced through the engine case, tore fuel lines, cut into hydraulic lines or the pump. Hydraulic fluid would have vented to atmosphere and drove pressure to zero. Total hydraulic pressure loss would have made the flight controls less manageable, with the aircraft ‘low and slow’. This possibility was never looked at.

Another possibility would be liberated compressor blades or components that damaged systems inside the aircraft’s tail or stabilizers, as was later seen with United 232 and Delta 1288. Elevator or rudder control cables would have been bent, cut; the bellcranks or pulleys damaged; flight control hinges jammed by expelled engine pieces; liberated compressor blades shot upwards that struck the elevator as the horizontal stabilizer moved into the blade’s trajectory. These possibilities were never looked at.

These potential events would explain why the cockpit was unusually quiet during the event. The captain and FO were mentally searching their training; the cautious call to ATC, “… uh, we’ve got an emergency here”; the stick shaker added chaos to the crisis. The atmosphere within the cockpit was one of growing confusion followed by what was more likely an uncommanded roll and stall. The pilots were dealing with conflicting information and ambiguous gauge readings within an inadequate allotment of time at a low altitude. It was a cruel hard fact: the aircraft could not be saved, no matter what they did.

It is an archaic process that NTSB Board Members, with no transportation experience, should still wield the power to cripple safety with ignorance. On one hand, the wrong party was blamed; on the other, the true root causes were ignored. Once published, an accident report is permanent; there are no appeals, no do-overs; no one to suggest that the investigation was done incorrectly. In the case of Midwest Express 105, the pilots have, for eternity, been saddled with the brand of Pilot Error for the crime of being in an aircraft that could no longer fly.

On February 3, 1987, the NTSB issued accident report AAR-87/01. The report was ambiguous; it was pure speculation; the report solely blamed two qualified pilots for an accident that was more a mechanical series of tragic events than operational mistakes. Root cause analysis would have properly interpreted the evidence. Safety was not improved. And isn’t improving safety what these accident reports are all about?

Stephen CarboneComment