HISTORY OF FLIGHT
On October 28, 1999, at approximately 2132 central daylight time, a Boeing B757-222, N575UA operated by United Airlines (UAL) as Flight 225, experienced an airframe vibration and diverted to O'Hare International Airport, Chicago, Illinois, for a precautionary landing. The 14 CFR Part 121 revenue flight had departed Dulles International Airport, Herndon, Virginia, at 1943 eastern daylight time with San Francisco International Airport, San Francisco, California, as the destination airport. While in cruise flight, a series of airframe vibrations occurred. The airspeed was reduced to 250 knots and the decision to divert to O'Hare was made. The landing was uneventful. None of the 2 pilots, 4 flight attendants, or 62 passengers were injured. Visual meteorological conditions prevailed and the flight was on an IFR flight plan.
After the incident, the Captain made the following write-up in the airplane's maintenance logbook:
"We encountered substantial vertical control flutter FL310-350 .78-.80 Mach with left, center, right autopilot and no autopilot. Flutter stopped when IAS was decreased. Descended to FL310, 250 kts., .75 Mach and had no more flutter. Two similar write-ups Oct. 17 and 21. Called SAMC [Systems Aircraft Maintenance Control]. They said they did not know what was wrong and that we might have some loose parts in the tail. Dispatch indicated Lt. to Mod. turbulence ahead over Nebraska and Colorado. Their advice was to divert to ORD. I agree and we diverted to ORD."
The First Officer reported the incident in the following manner:
"We left IAD on time at approximately 19:45. The departure and climb-out were normal. However, shortly after leveling out at FL 350 we felt a moderately strong vibration and were advised by the flight attendants that the entire cabin had felt it as well.
The most accurate way to describe it would be to say that it felt as if something in the tail section was moving, causing the front section of the plane to oscillate very rapidly up and down.
We disconnected the center autopilot and descended to FL 310, slowing to approximately mach .78. The vibration stopped temporarily, returning a few minutes later. We tried the left autopilot, and the vibration occurred with it as well, and did not stop until we descended and slowed later. [The Captain] requested that I handfly while he got on the radio with SAM, SFO Engineering, and Chicago Dispatch. We called a UAL pilot from the back to come up to the cockpit and help us out. At this point we were passing abeam ORD on our flight planned route. After [the Captain] talked with maintenance in San Francisco and was advised we might have 'a loose part in the tail,' and considering possible moderate turbulence past Denver, we elected to divert to Chicago. ATC gave us an immediate right turn and vectored us to ORD, giving us priority handling without declaring an emergency. We requested emergency equipment standby for the landing, which turned out uneventfully. We released the equipment upon rollout, and after debriefing maintenance at the gate, we exchanged planes and continued on into San Francisco."
During an interview, the Captain related the events concerning the in-flight vibration incident. He reported that the departure and climb out from Dulles were normal. The airplane leveled off at 35,000 feet with the center autopilot (AP) engaged. About 20 to 25 minutes after departure and south of Pittsburgh, the airplane experienced flutter (Captain's word). He reported that he clicked the center AP off and slowed the aircraft. He descended to 31,000 feet and stabilized at 31,000 feet.
He reported that he increased the airspeed with the left AP engaged. A short period later the flutter came back. He did not know what speed the airplane was flying. He reported that the flutter came back for 5 to 10 seconds. During that time, the Captain reported, he turned the left AP off and engaged the right AP. The flutter continued. He turned the right AP off and flew the airplane by hand. The flutter continued. He reported that when he reduced airspeed the flutter stopped. He reported that he thought the airspeed was controlling the vibration. He reported he had reduced the airspeed to about .78 mach or about 250 kts.
He reported that the vibration was felt throughout the airplane, but that it was not a violent vibration. He reported the vibration was a vertical vibration caused by the elevators.
He reported the flight attendants and the dead heading pilot felt the vibration. He characterized the vibration as what you would feel when taxiing over the bridge at O'Hare Airport. He felt the aircraft was completely sound.
He reported that he checked the flight manual, but could not find a procedure that applied to the situation.
He reported that he checked the aircraft logbook for previous write-ups. He found that the write-up from 8/21/99 seemed to be similar.
He contacted SAMC for their assistance. He described the problem and SAMC said they would check with engineering and get back to them. The Captain reported that it took SAMC about 20 to 30 minutes to call back.
SAMC and Dispatch contacted the flight. SAMC informed the Captain that there might be "...loose parts in the tail," and that some bushings or bearing might have more play than normal. During the three-way discussion between the Captain, SAMC, and Dispatch, it was decided that the flight should divert into Chicago.
During the interview, the Captain was asked if he had any theories concerning the vibration. He responded that he thought it was related to the airspeed. When the aircraft was slowed, the vibration stopped.
During the interview, the Captain was asked if he would do the same thing under similar circumstances. He responded that he would. He reported that there were no specific procedures for this type of occurrence, and that he would rely on his experience and airmanship.
During the interview, the Captain was asked why he tried all three AP's. He responded that he suspected the vibration was related to the AP. He was making a quick check of each AP since they are stand-alone systems.
During the interview, the Captain was asked if he had considered landing at Dulles since the flutter occurred within the first 20 to 25 minutes of a flight that lasted approximately 1.5 to 1.8 hours. He responded that he did not think it was necessary.
AIRCRAFT INFORMATION
The airplane was a twin engine Boeing B-757-222, serial number 26689, manufactured in 1993. The airplane seated 196 and its maximum gross weight was 241,000 pounds. The engines were Pratt and Whitney PW2037 turbofan engines that delivered 37,000 pounds of thrust each. The last inspection was on November 16, 1998, and was part of the continuous maintenance inspection program. The airplane had accumulated 3,160 hours since the last inspection and had a total time of 23,631 hours.
TESTS AND RESEARCH
UAL Maintenance and Test Flights
A maintenance inspection was performed on N575UA at the UAL maintenance hangar at O'Hare Airport. The free play of the elevator was found to be out of limits. It measured 0.5 inch and the allowable limit was 0.340 inch. Parts that were replaced included bushings, the left center power control actuator (PCA), two reaction links, and four idler links.
The first test flight was conducted on October 31, 1999. The airplane failed the test flight since it still exhibited a vertical vibration throughout the airplane. The logbook test pilot narrative write-up read as follows:
"ACFT fluttered noted at FL350 and .81M with CTR AP engaged, disengaged CTR AP, selected L & R AP's fluttering continued, turned AP's off and fluttering continued. Descended to FL310 and reduced speed to 250 kts and fluttering subsided. Fluttering occurred in pitch motion, all trim indications normal per crew. [name deleted]. See previous history."
During a conference call, the test pilot reported that the airplane still had a vertical bounce. He characterized the bounce in the cockpit like bouncing a baby on the knee at 2 to 3 cycles per second. He reported he observed the food carts in the aft galley were bouncing slightly off the floor. He reported the vibration was mach related, but that it would go away with slight control wheel pressure. He reported that the bouncing lasted about two minutes. He reported that it would stop briefly and then return. He reported that he had no difficulty controlling the airplane. He reported the free-play in the elevator did not get translated to the yoke because of the PCA's.
N575UA landed at Indianapolis International Airport (IND) at the conclusion of the test flight. Further maintenance was performed on the airplane at the UAL maintenance facilities. The free-play and rigging were re-checked. The only significant problem they identified was the cable tensions from the pilot and copilot yokes back to the elevator actuators. The cable tension required was 135 pounds. The cable tension was measured at 110 pounds.
On November 1, 1999, the airplane was flight tested for a second time. The airplane failed the test flight. The logbook pilot narrative write-up read as follows:
"Aircraft still has rapid pitch oscillation. Oscillation started while climbing through 32,000', mach 0.813. Oscillation occurs with any autopilot or no autopilot engaged. Oscillation stopped immediately when applying very slight<<"
N575UA returned to IND for further maintenance actions performed by UAL and Boeing technicians.
Boeing provided assistance to perform an instrumented test flight on N575UA. The following is a list of maintenance actions that were completed prior to the third test flight:
1. All hinge bearings on the right and left elevators were removed and replaced.
2. The PCA from the right outboard position was removed due to questionable rod end bearing corrosion/wear.
3. The PCA from the right middle position was removed due to questionable rod end bearing corrosion/wear.
4. The PCA from the left outboard position was removed due to questionable rod end bearing corrosion/wear.
5. The borrowed PCA from America West Airlines that had been used on the left center elevator during the two test flights was replaced.
The airplane was instrumented with eight accelerometers to determine the source of the vibration. A third test flight was flown on November 10, 1999, and all vibrations detected at all speeds and flight levels were within normal ranges. The logbook pilot narrative write-up read as follows:
"Aircraft test flown with Boeing instrumentation after elevator bearing replacement. Aircraft flew with no vibs-oscillations noted at speeds up to MM0 and altitudes up to 41000 feet. Aircraft was flown using each A-P individually and manually- Hydraulic systems were turned off, input to elev, rudder and aileron were made - all normal. APU started at 35000 - no problems noted. Aircraft can be returned to revenue service."
UAL Parts Removal and Inspection
Numerous parts were removed from N575UA for inspection and analysis during the course of the investigation. In particular, the elevator bearings, bushings, reaction links, and PCA's were inspected at the UAL San Francisco Maintenance Base, Flight Control shop, with FAA oversight.
A cursory examination of the bearings revealed that a majority of the bearings had sustained varying degrees of removal damage. The following is a list of the elevator hinge bearings with their vendor part number and position on the stabilizer from left to right.
Item # Part Name Part number Position
1) Bearing DAS6-23A-48 1
2) Bearing DAS6-23A-48 2
3) Bearing DAS6-23A-48 3
4) Bearing DAS6-23A-48 4
5) Bearing DAS6-23A-48 7
6) Bearing DAS6-23A-48 8
7) Bearing DAS6-23A-48 9
8) Bearing DAS6-23A-48 10
9) Bearing DAS6-23A-48 13
10) Bearing DAS6-23A-48 14
11) Bearing DAS6-23A-48 15
12) Bearing DAS6-23A-48 16
13) Bearing DAS8-27B-48 5
14) Bearing DASS-27B-48 6
15) Bearing DAS8-27B-48 11
16) Bearing DAS8-27B-48 12
The following is a list of the upper reaction link rod ends and PCA's that were inspected:
1) Left hand, Middle, Upper Reaction Link Rod End, P/N DRX34B, S251N214-11
2) Left hand, Inner, Upper Reaction Link Rod End, P/N DRX34B, S251N214-11
3) Left hand, Outboard, PCA (s/n 9293315) Rod End, P/N DRX32B
4) Right hand, Outboard, PCA (s/n 5304) Rod End, P/N DRX32B
Notes:
1) Part number DAS6-23A-48 falls under Boeing Specification BACB10CH60C
2) Part number DAS8-27B-48 falls under Boeing Specification BACB10CH85C
3) The serial number for the right hand PCA is not the actual unit serial number. The data plate was missing when the unit arrived in the shop. The number 5304 was derived from a pressure differential sensor serial number. Both units are identified with the following UAL part number and dash number: Left hand, Outboard, PCA (s/n 9293315) UAL p/n MR27314 - 269 Right hand, Outboard, PCA (s/n 5304) UAL p/n MR27314 - 268
The UAL examination of the 16 hinge bearings, two reaction links and two PCA rod end bearings yielded the following results:
1. All hinge bearings P/N DAS6-23A-48 (positions 1-4, 7-10 and 13-16) were received with the retaining sleeve still installed over the outside diameter and were observed to have moderate to extensive damage to the sleeves and bearings as a result of the removal process. Several bearings would not rotate as a result. All of these bearings exhibited a radial play of approximately 0.001 to 0.0015 inch with a load applied.
2. The four hinge bearings P/N DAS8-27B-48 (positions 5, 6, 11 and 12) were observed to have moderate to extensive damage to the bore inside diameter as well as the dust shields as a result of the removal process. These bearings do not utilize sleeves for retention. All bearings would rotate. Bearing numbers 5, 6 and 12 exhibited a radial play of approximately 0.001 to 0.0015 inch with a load applied. Bearing number 11 exhibited a radial play of approximately 0.035 to 0.040 inch with a load applied and an axial play of approximately 0.095 to 0.100 inch. External examination revealed the lubricant was present, but appeared reddish in color. Removal of the dust shield from one side, which was partially dislodged from the removal process, revealed lubricant that was reddish in color with a dried cake-like appearance. The bearing was mildly cleaned and microscopic examination showed severe corrosion damage to the rollers and races.
3. The left hand, middle, upper reaction link rod end (P/N DRX34B, S251N214-11) was examined and exhibited a radial play of approximately 0.008 inch with a load applied. External observations revealed lubrication that is reddish in color and some, non-location specific, gritty feel when rotated by hand.
4. The left hand, inner, upper reaction link rod end (P/N DRX34B, S251N214-11) was examined and exhibited a radial play of approximately 0.008 inch with a load applied. External observations revealed lubrication that is reddish in color and some, non-location specific, gritty feel when rotated by hand.
5. The left hand outboard PCA (UAL -269) Rod End and the right hand outboard PCA rod end (UAL -268) bearings both exhibited a radial play of approximately 0.001 to 0.0015 inch with a load applied.
6. The right hand outboard PCA rod end (UAL -268) bearing had a gritty feel during rotation by hand. The lubrication, that was visible externally, was reddish in color. Removal of the dust covers and mild cleaning revealed moderate to severe corrosion common to all roller surfaces and the ball. The corroded side was opposite the zerk fitting. The side common to the zerk fitting has minimal corrosion on the rollers and moderate corrosion on the ball.
UAL reported that it had complied with all the requirements of the Boeing 757 Maintenance Planning Document (MPD) and utilized Boeing's maintenance manual procedures to maintain the elevator control system.
Boeing Inspection of Parts
The 16 hinge bearings and the two reaction link rod ends were sent to Boeing for examination and testing.
On 19 May 2000, Boeing reported on the inspection results of the parts provided to them by UAL. The following is a summary of Boeing's findings for the elevator PCU reaction link bearings:
Internal Fit Clearance: 0.0003 Inch.
1. UAL bearing marked "X".
Axial movement: 0.032 Inch
Radial movement: 0.008 Inch
2. UAL bearing marked "T".
Axial movement: 0.025 Inch
Radial movement: 0.008 Inch
The grease from bearing "T" revealed properties consistent with Aeroshell 7 grease. Rust was also evident in the grease sample.
The Boeing report stated, "Although the bearings showed wear in the axial and radial direction, we do not believe that the bearings alone could have caused the vibration experienced by the listed airplane."
The hinge bearings were inspected. The Boeing findings on the elevator hinge bearings #2, #11, and #12 revealed the following results:
1. All three hinge bearings were found to contain the appropriate grease.
2. Evidence of iron/steel, copper, and cadmium corrosion and wear debris were detected in the grease of all three bearings.
3. The radial and axial clearances of all three bearings exceeded the nominal requirements.
Internal Clearance Limits:
Axial Clearance: 0.0016 - 0.0036 inch
Radial Clearance: 0.0006 - 0.0014 inch
The radial clearances for bearings #2, #11, and #12 were 0.017 inch, 0.038 inch, and 0.0045 inch, respectively.
The axial clearances for bearings #2, #11, and #12 were 0.0045 inch, 0.0880 inch, and 0.0005 inch, respectively.
4. Moderate to severe corrosion was found on the rollers, inner surfaces of the outer rings, and outer surfaces of the inner rings of all three bearings.
5. The outer rings and rollers of all three bearings met minimum hardness HRc 60 requirements.
The Boeing report stated, "... it is unlikely that the wear exhibited by subject bearings would result in a large magnitude of vibration. However, if additional elevator hinge bearings have excessive wear and the PCA bearings are also worn; the onset of vibration may be likely. Initially, the vibration would be a forced response type of vibration which would be near the natural vibration modes of the stabilizer/elevator system (about 20 hertz). As wear increases, the vibration would be strong enough to excite the rest of the airplane and would be felt through the airframe. When the first body vertical mode of vibration (3.5 to 4 hertz) starts, the vibration would be most noticeable in the flight deck and the aft galley."
The Boeing report stated that the key to maintaining the bearings and reducing wear in the bearings is proper lubrication. The report recommended the following to improve the effectiveness of the lubrication process:
1. Be diligent in doing the lubrication. Ensure that lubrication is done per 757 AMM 12-21-00 and 757 AMM 12-21-04.
2. Ensure that the grease is going into the bearing, and not coming out between the bearing outer surface and into inner surface of the hinge fitting lug.
3. Make sure the grease gun pressures are per 757 AMM 12-21-00.
The Boeing report stated that, "Based on the data that was reported on the datum airplane we feel that we do not have enough data to pinpoint the exact cause of the airframe vibration. …However, if the elevator hinge bearings have excessive wear and the PCA bearings are also worn, the onset of vibration may be possible."
ADDITIONAL INFORMATION
On June 9, 1988, Boeing issued Service Bulletin No. 757-27A0086 with a revision date of July 27, 1989. The subject of the Service Bulletin was: "Flight Controls-Elevator Power Control Actuators Reaction Links-Rod End Bearing Inspection/Replacement." The Service Bulletin stated the following:
"This service bulletin provides instructions to inspect and replace worn power control actuator (PCA) rod end and reaction link rod end bearings, so that rotational freeplay of the elevator is limited and unacceptable vibration during flight is prevented. The new bearings incorporate improved lubricant seals.
Six operators have reported excessive wear of elevator PCA rod end bearings on in-service airplanes. Four of these airplanes experienced intermittent airframe vibrations, and five experienced pitch oscillations with the autopilot engaged. These experiences occurred as a result of PCA rod end and reaction link rod end bearing wear. Subsequent examination of an operators fleet has identified a significant number of worn PCA reaction link rod end bearings. These bearings are prone to excessive wear because of corrosion resulting from insufficient lubrication and seals which permit moisture to penetrate the bearings. Worn elevator bearings can result in excessive elevator freeplay, and unacceptable airframe vibration during flight."
A Delta Airlines (DAL) B-757, N602DL, had experienced a series of aircraft vibrations in August and September 1999. Boeing reported that they had provided Delta technical assistance on what to look for and replace. Some of the elevator power control units were replaced along with the stabilizer trim motor/gearbox and jackscrew, as well as parts from the stabilizer pivot. A subsequent flight test showed the problem fixed, and the airplane was put back into service.
A stabilizer trim mechanism teardown was conducted by DAL and Boeing. Boeing reported the following concerning the inspection and its relevance to the UAL B757 incident:
"... dimensional checks of the individual loadpath components took place. Although slightly beyond design specification for new parts, none of the parts measured (either individually or as a built up unit) showed beyond service limits for wear, etc. The same checks were performed for the hinge bearings and the horizontal stabilizer pivot bearings. Here again, slight wearing of the parts, but nothing dramatic or what could be identified as a significant contributor to the noted vibration/pitch oscillation.
This is much like what has been found (so far) on the UAL airplane. The exception is that the UAL airplane stabilizer trim mechanism was checked on the airplane and found to be within free-play limits; it was not replaced. But again, following the bearing replacement, etc. the UAL airplane showed no problems and was put back into service."
On January 28, 2001, Boeing issued a change to the B-757 maintenance manual. The procedures for the elevator free-play checks outlined in the maintenance manual from September 20, 1997 to January 28, 2001, were incorrect. The incorrect procedures stated the following steps in the Elevator Free-play Check:
"Step 8. Use the force scale to apply an 80 pound force +/- 10 pounds to the bottom of the elevator for 15 seconds or until the elevator movement stops. Apply this force along the elevator rib opposite the powered PCA and approximately 2 inches forward of the elevator trailing edge.
Note: The right hydraulic system powers the outboard actuators. The left hydraulic system powers the center actuator. The center hydraulic system powers the inboard actuators.
Step 9. Slowly remove the 80 pound force.
Step 10. Record the dial indicator or scale reading."
The January 28, 2001, revised maintenance manual stated the following steps in the Elevator Free-play Check:
"Step 8. Use the force scale to apply an 80 pound force 10 pounds to the bottom of the elevator for 15 seconds or until the elevator movement stops. Apply this force along the elevator rib opposite the powered PCA and approximately 2 inches forward of the elevator trailing edge.
Note: The right hydraulic system powers the outboard actuators. The left hydraulic system powers the center actuator. The center hydraulic system powers the inboard actuators.
Step 9. Record the dial indicator or scale reading.
Step 10. Remove the 80 pound force."
In both maintenance manual revisions, if the free play measured was more than 0.340 inch, then the PCA rod ends and the reaction link bearings needed to be examined for too much wear.
On October 11, 2001, the Federal Aviation Administration issued Airworthiness Directive (AD) 2001-20-11, with an effective date of November 15, 2001. The AD superseded AD 89-03-05, which had an effective date of March 6, 1989. The new AD stated, "The actions specified by this AD are intended to prevent unacceptable airframe vibration during flight, which could lead to excessive wear of bearings of the elevator PCA load loop and hinge line and result in reduced controllability of the airplane. This action is intended to address the identified unsafe condition."
AD 2001-20-11 is applicable to all Boeing 757 series airplanes and its purpose is, "To prevent unacceptable airframe vibration during flight, which could lead to excessive wear of elevator bearings and result in reduced controllability of the airplane...."
The new requirements of AD 2001-20-11 required elevator free-play checks of all Boeing 757 series airplanes per Boeing Service Bulletin 757-27A0086, Revision 2, dated July 27, 1989. The AD stated, "Before further flight after the free-play checks, lubricate the bearings in the elevator PCA load loop and hinge line. Use of either the BMS3-24 or BMS3-33 grease will be acceptable as long as the grease types are not intermixed on any individual bearing." AD 89-03-05 did not require lubrication of the hinge line bearings.
The new AD required repetitive elevator free-play checks and lubrication within intervals of 4,000 hours, or 3,000 hours for applicable models, of the most recent inspection per AD 89-03-05, or 18 months after the effective date of this AD (November 15, 2001), whichever occurred first. AD 89-03-05's terminating action was replacement of old bearings with new, improved bearings. AD 2001-20-11 removed the terminating action and changed the repetitive elevator free-play checks from every 4,000 hours, or 3,000 hours for applicable models, to 18 month inspection cycles. It also added a lubrication task, lubricating the hinge line bearings and actuator load loop, to be done at the time of each free-play check.
The Boeing Company published a magazine article titled, "In-flight Airplane Vibration and Flight Crew Response" in the October 2001 Issue, No. 16, pages 3-9, of the AERO Magazine, a publication of Boeing Commercial Airplanes.
The article stated the following concerning flight crew response to a vibration event:
"There is concern that a drastic measure taken by the flight crew to rectify vibration could actually increase the severity of the problem. For this reason, the best flight crew reaction to an abnormal vibration is to smoothly extract the airplane from the operating region where the vibration occurs.
If performance considerations do not override the severity of the vibration, the flight crew should reduce airspeed and engine speed. Essentially, the crew should return to level flight at reduced airspeeds and avoid unnecessary stress on the airplane."
The article stated the following concerning in-flight observations that can assist during maintenance troubleshooting:
"Identifying and correcting the cause of in-flight airplane vibration often is accomplished through trial and error, which can consume many maintenance hours. The causes of airplane vibration are numerous; however, flight crew observations and detailed reporting can provide very important clues to the potential source of the vibration.
Post flight, flight crews generally report two types of vibration. The first is a high-frequency tactile vibration (typically more than 25 Hz) that is felt in either the hands or feet. This vibration is sometimes associated with sound and usually relates to a small-mass component acting on the airframe, such as a loose door, access panel, or fairing. This type of vibration can be constant during all phases of flight, but it may vary with airspeed.
The other type of vibration is of a lower frequency (typically less than 20 Hz) that can be felt by the entire body. This type of vibration usually relates to a large-mass component acting on the airframe, such as the rudder, horizontal stabilizer, or elevator."
The article further stated the following concerning the in-flight observations that assist in determining the vibration source:
"In-flight observations can provide essential clues to the source of a vibration. Information about the airplane speed, flight conditions, engine power settings, and effects of changes made to airplane systems and flight controls on vibration can assist in identifying the source.
For example, a low-frequency vibration in the vertical direction that is felt in both the forward and aft cabin may be the result of excessive free play in the elevator or stabilizer surfaces. Pitch-control flight surfaces with excessive free play can cause the body to vibrate vertically, with the motion felt most strongly in the forward and aft locations of the airplane. Slight mis-trimming of the airplane using the stabilizer and elevator may dampen out this type of vibration because the free play of the surface is removed with aerodynamic loading."
The article went on to state that when moderate to severe vibrations are felt in the aft fuselage on B-757 airplanes in cruise flight, with the vibration decreasing with airspeed, then the elevator free play should be checked. The maintenance action requires the elevator power control actuator and link bearings be replaced if the free-play check fails.
Parties to the investigation included the Federal Aviation Administration, United Airlines, and Boeing.