February 2001

Alaska Airlines Flight 261 Update:
The Causal Mystery Is About to Be Solved 

by Robert F. Hedrick

This article represents the views and opinions of the author – Ed.

On January 31, 2000, Alaska Airlines Flight 261 crashed into the ocean off Southern California, killing all 88 persons aboard. Author Robert Hedrick practices and teaches aviation law in Seattle, and represents several of the victims' families. Last December, the National Transportation Safety Board held a hearing in Washington, D.C., to establish the facts surrounding the crash. The facts and probable cause for the crash, as established by the NTSB, may play a critical role in the litigation over this event pending before a multidistrict litigation judge in U.S. District Court in San Francisco. Hedrick attended the hearing and reports here on the detailed and highly technical evidence presented at the hearing, which explains why the frantic effort of the flight crew to control their aircraft was ultimately unsuccessful.

The National Transportation Safety Board (NTSB) is the federal agency responsible for investigating aviation accidents. Although the agency has no rulemaking authority, it wields significant influence because of its mandate to determine the facts and probable cause of aviation disasters. Immediately after Flight 261 crashed, the NTSB sent its "go team" to the scene, and assigned board member John Hammerschmidt to lead the investigation.

NTSB investigative units for aircraft operations/human performance; air traffic control; witnesses; structures; systems/powerplants; and maintenance records were assigned to the case. There were also groups formed to investigate the flight data recorder, cockpit voice recorder, aircraft performance, and lubricating grease. In addition, the Federal Aviation Administration (FAA), Alaska Airlines, The Boeing Company, Pratt & Whitney Engines, the Air Line Pilots Association, the Aircraft Mechanics Fraternal Association, the Association of Flight Attendants, and the National Air Traffic Controllers Association were invited to participate in the investigation, and to provide outside technical assistance.

In major air disasters a public hearing is held to create a full record of the facts, circumstances and conditions of the accident. Such hearings are not adversarial, but are fact-finding proceedings. In this case, hundreds of exhibits were submitted and many witnesses were examined by NTSB technical staff, board members, and other parties to the investigation. Later this year, after all post-hearing test results are in, the NTSB will issue its final accident report containing its factual determinations and probable cause findings. The report does not allocate fault to the culpable parties, but sets out the relevant factors that likely caused the accident.

The crew for Flight 261 arrived in Puerto Vallarta, Mexico, on January 30, 2000, the day before the accident. The crew who had flown the MD-83 south to Puerto Vallarta (the next day) did not report any significant problems. The aircraft departed Puerto Vallarta at 1:37 p.m. PST, with one scheduled stop in San Francisco, and a final destination of Seattle.

The flight data recorder indicated that the autopilot was engaged during climb-out, but was turned off 13 minutes later. At 3:46 p.m., the autopilot was re-engaged for three minutes, disengaged for 20 seconds, and then re-engaged.

At 3:49 p.m., the crew contacted Alaska Airlines dispatch and maintenance control in Seattle to discuss a problem with the horizontal stabilizer, which they thought was jammed, and to consider diverting to Los Angeles. After an extended discussion, the crew decided to land at Los Angeles.

At 4:09 p.m., as the aircraft was at 30,000 feet over the Pacific Ocean (west of L.A. and east of Santa Catalina Island), the autopilot was turned off. At that moment, two faint thumps are heard on the cockpit voice recorder (CVR), and the pitch trim position moved 2.4 degrees, nose down, over the next two seconds. This was the first recorded stabilizer movement since climb-out from Puerto Vallarta.

At 4:10 p.m., the crew reported: "Center, Alaska 261, we are, uh, in a dive here … yeah, we're out of 26,000 feet, we're in a vertical dive … not a dive yet, but twenty-three seven request, uh … yeah we've got it back under control there." At this moment, a second voice from the background disagrees, "No, we don't." The aircraft descended for two minutes, reaching a maximum speed of 353 knots. Speed brakes were deployed, and the elevator was used to pull the aircraft out of the dive. Once level, leading edge slats and trailing edge flaps were deployed on the wings, but were retracted 20 seconds later.

The flight stabilized at 24,000 feet. The crew reduced airspeed and began to troubleshoot the problem. At 4:12 p.m., they told Alaska's Los Angeles maintenance base that they thought they had a runaway trim. According to an Alaska mechanic, the crew stated: "We are in a worse situation than we were … I'm afraid to try it again to see if we can get it to go in the other direction."

At the same time the crew told the passengers: "We have a flight-control problem up front here. We're working it. That's Los Angeles off to the right there. That's where we're intending to go. I'd anticipate us parking there in about 20 to 30 minutes."

At 4:15 p.m., the crew reported to Los Angeles Air Traffic Control: "We have a jammed stabilizer and we're maintaining altitude with difficulty, uh, but uh, we can maintain altitude we think. … I need to, uh, get down about 10, change my configuration, make sure I can control the jet, and I'd like to do that out here over the bay. …" The controller read the L.A. altimeter setting, and the crew acknowledged: "Thank you." This was the last radio call from Flight 261.

By 4:18 p.m., the flight had descended to 18,000 feet. At 4:19:35 p.m., holding an airspeed of 270 knots (about 300 mph), the flaps started to deploy. At that time, an extremely loud noise is heard on the CVR. The aircraft then rapidly pitched nose down with a pitch rate of nearly 25 degrees per second. The G-forces reached negative three Gs within three seconds. After six seconds, the aircraft reached 80 degrees nose down and then rolled over, flying upside down.

The pilots struggled for control. They applied forces to the rudders, ailerons and elevator. They deployed the speed brakes, but nothing worked. The aircraft dove nearly straight down for three miles at a high rate of speed, impacting the ocean at 4:20:57 p.m.

The crash occurred in the Channel Islands National Marine Sanctuary, off Point Loma, California, 2.7 miles from Anacapa Island. The debris and wreckage sank in 700 feet of water, spreading out on the ocean bottom across an area the size of a football field. Approximately 90 percent of the wreckage was recovered.

On February 8, 2000, most of the horizontal jackscrew assembly was recovered. Even on sight of the untrained eye, the jackscrew appeared to have failed, as threads torn from the acme nut wrapped around the screw as if they had been torn clean from the nut.

The horizontal stabilizer is the fixed horizontal surface on the tail of an airplane. On the MD-83 it is on top of the tail. The stabilizer is 40 feet long and moves up and down in flight, allowing the pilot to set the trim of the aircraft in either nose up, nose down or level attitude. The horizontal stabilizer is normally a long-term control, which is set and left alone for extended periods of flight. The elevator is the moveable (hinged) control surface located at the trailing edge of the horizontal stabilizer. It is operated by the pilot's yoke control, and allows immediate changes in flight trim (up/down) attitude. With a proper trim setting the aircraft can fly at any normal attitude without undue forces on the pilot controls.

On MD-83 aircraft, the horizontal stabilizer is operated by one threaded shaft called a "jackscrew." It is two feet long and attached to the forward edge of the stabilizer. Two electric motors (primary and secondary) rotate the shaft, which is located directly below the motors. The jackscrew rotates through a stationary acme nut that is attached to the vertical stabilizer. Jackscrew rotation causes the entire assembly to move up and down. Since the nut is fixed to the vertical stabilizer, rotation of the screw moves the forward edge of the horizontal stabilizer up and down, which helps control aircraft pitch in flight. Horizontal stabilizer trim is set by the pilot or operated through the autopilot.

The type of jackscrew assembly involved in the accident was originally designed by Douglas Aircraft in 1965 for the DC-9. It weighs 100 pounds, is two inches in diameter, and costs $60,000. The accident jackscrew was supplied to McDonnell Douglas on June 28, 1990 by the Peacock Company of Norwalk, California. Boeing took over McDonnell Douglas in 1997.

The jackscrew has a maximum up-and-down limit, and a two-part failsafe system. First, the electric motors (primary and secondary) that turn the screw have a built-in electrical shutoff. Second, there are mechanical stops at the upper and lower limits, which are set to stop the screw from rotating when the acme nut meets the mechanical end stops (the secondary backup). The electrical shutoff stops just short of the mechanical stops. If the electrical shutoff fails, the mechanical stops are supposed to prevent further travel. The operating mechanism (including the motors) is located at the top of the assembly, and the screw is located below it.

The maximum movement limits for the electrical shut-off are 2.2 degrees up stop (which moves the horizontal stabilizer up, causing the aircraft nose to go down) and 3.1 degrees down stop. The mechanical stops are set at 2.6 degrees up and 3.6 degrees down. Importantly, even if the horizontal stabilizer is stuck at any of these maximum limits, aircraft pitch can still be controlled by the elevator. However, if these maximums are exceeded, at a certain point (depending on various aerodynamic factors) the elevator will no longer be able to counter the forces of the horizontal stabilizer, and the aircraft will not be able to maintain level flight.

In light of the sensitivity in trim changes, the electric motors on the jackscrew assembly have a trim rate of one-third degree per second when operated by the primary electric motor, and one-tenth degree per second when operated by the alternate motor. Primary pitch during flight is controlled by the elevators, not the horizontal stabilizer.

The screw part of the assembly is a titanium core (quill) torque tube with a double-threaded outer layer of hardened steel. The two independent threads that wind the screw are supposed to be redundant design, with one winding backing up the other.

A washer and a nut, screwed directly into the torque tube, are at the bottom and top of the screw. Attached to the screw on the inside of the nuts are the mechanical stops that prevent the screw from rotating beyond the maximum limits. The stops were not designed to handle direct vertical loads in the unlikely event that the acme nut becomes stripped.

The load is carried primarily on the torque tube and is transferred through the acme nut assembly, which is stationary and attached to the vertical stabilizer. The nut is eight inches long and pivots to accommodate the slight rotation of the jackscrew assembly as it moves up and down. The acme nut is made of aluminum bronze, which is a softer metal than the hardened steel on the jackscrew. As such, the acme nut will wear before the jackscrew, allowing the less expensive acme nut to be replaced once wear limits are reached. There is a single grease zerk fitting on the acme nut for injecting lubricating grease into threads contained in the nut. A four-by-six-inch access panel is located in the vertical stabilizer, which is used for inspection and maintenance of the jackscrew.

When the jackscrew was found, threads stripped from the acme nut were wrapped around it. At the NTSB hearing, the jackscrew design was challenged for a number of reasons. There does not appear to be a redundancy (backup) in the event the jackscrew threads are stripped. The dual thread design is not redundant when threads are stripped (by lack of lubrication) because all the threads will strip. Larger aircraft, such as the DC-8, DC-10 and MD-11, have two jackscrew assemblies operating the horizontal stabilizer. If there is a complete failure of one jackscrew, the other will take the load. It appears that the mechanical end stop/nut and bolt located at the bottom of the jackscrew severed off. In other words, the torque tube fractured. In order for the aircraft to pitch down at a rate of 25 degrees per second (going into the last dive), the horizontal stabilizer would have likely rotated up 22 degrees (or more), which is at least 10 times the maximum design limit. Though disputed by Boeing, it was suggested that the mechanical end stop broke off in flight and not when the aircraft hit the water. This might explain the loud noise right before the dive, and the horizontal stabilizer deflection well beyond its normal limits.

Like most moving mechanical parts, the jackscrew assembly has to be maintained or it will eventually fail. The acme nut and jackscrew must have adequate lubrication in order to operate properly. In addition, jackscrew wear must be periodically checked to ensure it is within safe limits.

There are two main lubricating issues: lubricating intervals, and the type of lubricating grease used.

Originally the DC-9 jackscrew was to be lubricated every 600 to 900 flight hours. Over time, lubrication intervals were increased, but varied from airline to airline because of "demonstrated reliability." For example, at the time of the accident, Alaska had the longest interval between jackscrew lubrications. Alaska and US Airways checked the jackscrew and lubricated every 2,500 flight hours. At that same time the interval was 500 hours at Airborne Express, 600 hours at Scandinavian Airlines (SAS), 900 hours at American Airlines, and 1,000 hours at Hawaiian Airlines.

After the accident, the FAA ordered inspection of all jackscrews on the DC-9/MD-80 series aircraft. Alaska had the most reported problems — six, of which five failed the acme nut wear test. Airborne Express and American did not report any problems. However, Hawaiian Airlines had three aircraft (out of 15) that reported problems. The FAA issued an Airworthiness Directive mandating that the jackscrews be checked for lubrication every 650 hours.

Mobil 28 was the lubricating grease used throughout the U.S. on the jackscrew (and flight controls, doors and landing gear) on the DC-9 and MD-80 series aircraft. In 1995, based on Boeing specifications, Aeroshell 33 was developed. The advantage of Aeroshell 33 was that it performed better and had stronger corrosion protection for steel surfaces, which would increase part life.

Boeing had "no technical objection" to Alaska changing from Mobil 28 to Aeroshell 33 on MD-80 series aircraft. But Alaska was the only major U.S. airline that used Aeroshell 33 to lubricate jackscrews in its fleet. The transfer from Mobil 28 to Aeroshell 33 was accepted by the FAA as a "routine change," and considered insignificant. In December 1997, Alaska began replacing Mobil 28 with Aeroshell 33.

The procedure for changing lubricant was to remove all Mobil 28, add Aeroshell 33, run the system to spread the Aeroshell 33, remove all lubricant again, and then lubricate a second time with Aeroshell, and run the system. Mobil 28 and Aeroshell 33 are considered incompatible, and produce an inferior product when mixed together. In May 1999, Boeing suggested that aircraft operators avoid mixing the two greases and remove all old grease from their systems.

Testing at the U.S. Naval Laboratory determined that the grease on the Flight 261 jackscrew was "contaminated" because it contained both Mobil 28 and Aeroshell 33. It could not be determined what percentage of each grease was in the test sample. The sample did contain aluminum bronze particles that were from the stripped threads of the acme nut. Further tests are being performed to determine whether or not Aeroshell 33 causes aluminum bronze (the acme nut material) to corrode.

It was also determined that the zerk fitting on the acme nut was clogged, and had to be cleared with a flat punch. (A zerk fitting allows lubricant to be injected into a closed system under pressure without leaking out. The zerk fitting has a ball shape on the outside that is gripped by the nozzle of a grease gun pushed into it. The grease flows to the threads through a one-way valve. When grease is seen flowing out of both ends of the acme nut, the nut is considered to be lubricated. When the grease gun is removed, the valve in the Zerk fitting closes and prevents grease from leaking out and debris from getting in.) The acme nut contained eight inches of threads which would not have been lubricated if the fitting was clogged at the time of the last lubrication before the accident. In addition, Alaska may not have followed its own internal steps for approving the change from Mobil 28 to Aeroshell 33.

When originally certified by the FAA, the jackscrew assembly could be used as long as it passed the end-play test. The test measures the clearance between the threads of the screw and the threads of the nut, and reflects the amount of wear. If the clearance was more than .040 of an inch, the assembly had to be replaced. According to Boeing, when the clearance is at the .040 limit, the unit should handle 10 times the anticipated structural load. Even when the clearance is doubled to 0.080, it should hold five times the load. In cruise flight at 300 knots and proper trim, the load is 4,000 pounds.

Originally, wear checks were required every 3,600 hours; this was eventually increased. McDonnell Douglas required the jackscrews to be inspected for wear every 30 months or 7,200 flight hours, whichever came first. In 1996, Alaska convinced the FAA to increase those inspections by removing the flight-hour requirement, leaving only the 30-month inspection, no matter how many hours were flown.

The airplane that crashed was scheduled for inspection two months after the accident. At the time of the accident, the aircraft had logged 8,884 hours since the last inspection in September 1997. During the inspection at Alaska's Oakland repair facility, lead mechanic John Liotine performed the end-play check and determined that the play was .040 inch, the maximum limit. Technically, replacement was required if play is more than .040, but Liotine wanted the jackscrew replaced because it was showing a lot of wear related to the age of the aircraft. A few days later Liotine was overruled by a supervisor, who tested the jackscrew and determined that it was within its normal limits. Liotine was never told that his decision had been reversed. Sixteen months before the crash, Liotine told federal investigators that maintenance records were being falsified at Alaska. It is not known if Alaska had a replacement jackscrew assembly available at the time (as the aircraft was due to return to service), raising speculation about the word "panic" written in a log entry about efforts to obtain more parts for the aircraft. Liotine believed that replacing the jackscrew would have delayed release of the aircraft.

Had the 7,200-hour interval been in effect at Alaska, the jackscrew on Flight 261 would have undergone an end-play check 1,684 hours before the accident. After the accident, the FAA's airworthiness directive decreased the wear-check interval to every 2,000 hours.

Concern was also raised about the accuracy of the measuring instrument, which was apparently made by Alaska, and not subject to mandatory calibration. A related issue exists regarding the possibility that grease on the jackscrew could cause inaccurate test results. Alaska now requires the jackscrew assembly to be wiped clean before wear tests are performed.

Prior to the accident, an internal computer alerted Alaska of potential wear problems with the jackscrew. In 1999, Alaska commenced computer tracking of the jackscrew end-play test results. In June of that year, a jackscrew failed the test during heavy maintenance. Five months later, another jackscrew failed the test, and a third was checked and later replaced. Three days after the accident, company software generated statistical analysis of maintenance that triggered an alert to the jackscrew problem. At the time of the accident, Alaska was not tracking wear rates of individual jackscrews.

The Federal Aviation Agency (FAA) is the primary rulemaking and oversight authority for airlines and the aircraft industry. The FAA approved the jackscrew assembly design and was responsible for overseeing maintenance practices at Alaska Airlines.

With regard to aircraft design, the cardinal safety rule (which is also in the Federal Aviation Regulations (FARs)) is that no single failure of a critical aircraft part should cause a crash. Heavier aircraft have two jackscrew assemblies that operate the horizontal stabilizer, and back up each other. On the DC-9 and MD-80 series aircraft, there is only one jackscrew assembly. The alleged redundant design is (1) the dual thread screw winding, and (2) the mechanical end stops. The FAA approved this system.

Concerns were raised at the hearing about the FAA's approval of the design, which did not appear to have adequate redundancy in the event that the threads stripped. Nevertheless, the FAA believed that the design met the relevant design regulations at the time.

With regard to aircraft maintenance, the FAA has a Maintenance Review Board (MRB), which outlines the minimum scheduled inspection program for specific aircraft. The MRB recommends certain maintenance practices, however those recommendations are not mandatory.

Pursuant to the FARs, each airline establishes its own maintenance program (consistent with the manufacturer's recommendations), and that program is reviewed and accepted by the designated FAA principal maintenance inspector (PMI). PMIs establish close working relationships with air carriers, and are allowed considerable discretion in reviewing and approving maintenance practices and procedures. That is why the practices vary from carrier to carrier. If a carrier demonstrates reliability, it can request that the PMI modify the maintenance schedule, usually to lengthen the intervals for inspection, lubrication and overhaul, which saves the carrier money.

The PMI approved the changes at Alaska involving the increased lubrication interval, the increased end-play check interval, and the change of lubrication from Mobil 28 to Aeroshell 33. Some of these decisions will probably be criticized in the NTSB's report. It was also suggested that the FAA was lax in its oversight of Alaska maintenance, and that the FAA failed to discover or take action on maintenance items before the accident.

There is no doubt that failure of the jackscrew caused Flight 261 to crash. One might consider the failure a product defect, but the story of human error begins more than 35 years before the accident. A complex series of mistakes, including errors in design, maintenance, testing and government oversight, all likely contributed to the eventual failure of the horizontal stabilizer system, resulting in the loss of 88 lives. Despite the intense investigation of the NTSB, we will never know to what extent each element contributed to the disaster.

Back to table of contents >>