Loss of Pitch Control During Takeoff Air Midwest Flight 5481

8

Loss of Pitch Control During Takeoff Air Midwest Flight 5481

Raytheon (Beechcraft) 1900D, N233YVCharlotte, North CarolinaJanuary 8, 2003

〔图〕插图

Aircraft Accident Report

NTSB/AAR-04/01

PB2004-910401

Notation 7545B

National Transportation Safety Board Washington, D.C.

Aircraft Accident Report

Loss of Pitch Control During Takeoff Air Midwest Flight 5481 Raytheon (Beechcraft) 1900D, N233YV Charlotte, North Carolina January 8, 2003

NTSB/AAR-04/01

PB2004-910401

Notation 7545B

Adopted February 26, 2004

〔图〕w3VSARD\n National Transportation Safety Board

490 L'Enfant Plaza, S.W.

Washington, D.C. 20594

References

Contents

iv

Aircraft Accident Report

1. 12.4.1 Pitch Trim Control System $\ldots$ 39

1. 13 Medical and Pathological Information $\ldots$ 39

1. 14 Fire .................................................................................................................................... 39

1. 15 Survival Aspects 40

1. 15.1 Emergency Response $\ldots$ 40

1. 16 Tests and Research ............................................................................................................................................................................................................................................................................................................................................................................................................................................................................................

1. 16.1 Ground Tests $\ldots$ 42

1. 16.1.1 Baseline Configuration $\ldots$ 42

1. 16.1.2 Turnbuckle Length Adjustments $\ldots$ 42

1. 16.1.3 Cable Tension Adjustment Procedural Errors $\ldots$ 43

1. 16.1.4 Cable Stretch $\ldots$ 43

1. 16.2 Airplane Performance Study $\ldots$ 43

1. 16.2.1 Flight 5481 Weight and Balance Study $\ldots$ 44

1. 16.2.1.1 Air Midwest Weight and Balance Program $\ldots$ 44

1. 16.2.1.2 Airplane Component Buildup Method $\ldots$ 44

1. 16.2.1.3 Takeoff Ground Roll Weight Analysis 46

1. 16.2.2 Load Manifest Study $\ldots$ 47

1. 16.2.3 Elevator Control Authority Study ..........................................................................................................................................................................................................................................................................................................................................................................................................................................................................

1. 16.2.3.1 Relationship Between Pitch Control Position and Actual Elevator Position ................................................................................................................................................................................................................................................................................................................................................................................................................................

1. 16.2.3.2 Available Elevator Travel $\ldots$ 50

1. 16.2.4 Beech 1900D Simulation Studies .................................................................................................................................... 50

1. 16.2.4.1 Elevator Required to Match Accident Flight Data 51

1. 16.2.4.2 Elevator Required to Establish Controlled Flight $\ldots$ 52

1. 16.2.4.3 Kinematics Extraction $\ldots$ 52

1. 16.2.4.4 Simulator Trims. $\ldots$ 53

1. 17 Organizational and Management Information $\ldots$ 54

1. 17.1 Air Midwest $\ldots$ 54

1. 17.1.1 Air Midwest Maintenance Training Program $\ldots$ 55

1. 17.1.2 Huntington Maintenance Station Operations. $\ldots$ 56

1. 17.1.3 Air Midwest Audits of the Huntington Maintenance Station 58

1. 17.1.4 Postaccident Actions $\ldots$ 59

1. 17.2 Raytheon Aerospace $\ldots$ 60

1. 17.2.1 Contract Between Raytheon Aerospace and Air Midwest 62

1. 17.3 Structural Modification and Repair Technicians 62

1. 17.3.1 Contract Between Structural Modification and Repair Technicians and Raytheon Aerospace 63

1. 17.4 Federal Aviation Administration Oversight. $\ldots$ 63

1. 17.4.1 Continuing Analysis and Surveillance System $\ldots$ 65

1. 17.4.2 Correspondence From the Federal Aviation Administration to Air Midwest. $\ldots$ 66

1. 17.5 Raytheon Aircraft Company $\ldots$ 68

1. 18 Additional Information 68

1. 18.1 Weight and Balance Procedures. $\ldots$ 68

1. 18.1.1 Federal Aviation Administration $\ldots$ 68

1. 18.1.1.1 Postaccident Average Weight Survey $\ldots$ 70

1. 18.1.2 Air Midwest $\ldots$ 72

1. 18.1.2.1 Postaccident Weight and Balance Revisions. $\ldots$ 74

1. 18.1.3 US Airways Express $\ldots$ 75

References

Contents

vi

Aircraft Accident Report

1. 5.2.2 Oversight of Continuing Analysis and Surveillance System Programs $\ldots$ 118

1. 5.2.3 Oversight of Maintenance Training Programs. $\ldots$ 119

1. 5.2.3.1 Programs to Reduce Human Error in Aircraft Maintenance. $\ldots$ 120

1. 5.2.4 Oversight of Weight and Balance Programs $\ldots$ 120

1. 5.2.4.1 Approval of Air Midwest's Weight and Balance Program $\ldots$ 121

1. 5.2.4.2 Use of Average Weights $\ldots$ 123

1. 5.2.4.3 Technological Advances $\ldots$ 123

1. 5.3 Raytheon Aircraft Company $\ldots$ 124

1. 6 Cockpit Voice Recorders Installed in Beech 1900

Series Airplanes 125

3. Conclusions $\ldots$ 128

1. 1 Findings $\ldots$ 128

1. 2 Probable Cause $\ldots$ 131

4. Recommendations. 132

1. 1 New Recommendations $\ldots$ 132

1. 2 Previously Issued Recommendation Reiterated in This Report $\ldots$ 135

1. 3 Previously Issued Recommendation Resulting From This Accident Investigation 135

1. 4 Previously Issued Recommendation Classified in This Report $\ldots$ 135

5. Appendixes

A: Investigation and Hearing. $\ldots$ 137

B: Cockpit Voice Recorder Transcript $\ldots$ 138

C: Air Midwest's Elevator Control System Rigging Procedure at the Time of the Accident $\ldots$ 192

D: Air Midwest's Revised Elevator Control System Rigging Procedure 200

Factual Information

5

Aircraft Accident Report

Beechcraft 1900D Maintenance Program Manual

DATE: 08/25/00 PAGE: A.149

SIXTH DETAILED INSPECTION PROCEDURES CHECKLI

$$ \frac{62}{57} $$

Figure 1. Detail six work card at the time of the accident.

Factual Information

7

Aircraft Accident Report

〔图〕FORM M0UeAixwaye/0xpxea/ hix MidnestPAGEDATE 01/06/03AIRCRAFT MAINTENANCERECORD LATS/O#05-66----COPyAIRCRAFT S: UE233N23

Figure 2. Aircraft maintenance record of nonroutine items for January 6, 2003, at the Huntington, West Virginia, maintenance station.

Factual Information

8

Aircraft Accident Report

In addition, the mechanic stated that, while adjusting the cable tension, he bypassed several steps of the complete elevator control system rigging procedure (section 27-30-02).$^{17}$ The Beech 1900D AMM did not contain a stand-alone procedure for checking elevator cable tension, as called for on the D6 work card, or for adjusting elevator cable tension without rigging the entire elevator control system.$^{18}$ The quality assurance inspector stated, during a postaccident interview, that he and the mechanic discussed which steps to bypass and that he allowed the mechanic to adjust only the cable tension.

The accident airplane returned to service on the morning of January 7, 2003. The airplane flew a total of nine flight legs before the accident flight.

1.2 Injuries to Persons

Table 1. Injury chart.

Note: The minor injury involved a US Airways mechanic who was treated for smoke inhalation.

1.3 Damage to Airplane

The airplane was destroyed by impact forces and a postcrash fire.

1.4 Other Damage

The airplane struck a US Airways maintenance hangar located on CLT property.

Factual Information

16

Aircraft Accident Report

〔图〕(Side View)Primary Control StopsElevatorControl HornAft BellcrankTurnbuckles\n

Figure 3. Beech 1900D pitch control system.

〔图〕ElevatorsTurnbucklesPrimaryControl ColumnControl StopsControl CablesSecondary Control StopsFlight Data Recorder Sensor\n

Figure 4. Side view of Beech 1900D tail section.

Factual Information

18

Aircraft Accident Report

〔图〕插图

Figure 5. Turnbuckle assemblies.

The control cable assemblies are routed through pulleys from the aft bellcrank, through the rear pressure bulkhead, beneath the rear cargo compartment and cabin floor, and to the forward bellcrank beneath the cockpit pedestal and the floor under the first officer's seat. The forward bellcrank contains two secondary stop bolts (one up and one down), which have AMM-specified deflection settings that allow slightly more travel than the primary stop bolts at the elevator control horns. Structure above and below the forward bellcrank contains holes for a rig pin.$^{29}$ A forward push-pull tube, located beneath the cockpit pedestal, connects the forward bellcrank with the bottom of a T-shaped control column assembly.

Two control wheels—one for the captain and one for the first officer—are located in front of each cockpit seat. The control wheels (also known as control yokes) are connected to the top of the T-shaped control column assembly by a shaft and a flexible coupling. Each shaft passes through a control column support assembly, with a roller slider assembly mounted inside, providing support to each shaft where it enters the instrument panel. Each roller slider assembly consists of three rollers, one of which has an eccentric screw that is adjusted to set roller clearance. The captain’s control column support and shaft assembly contains a hole for a gust lock pin, which is used only when the airplane is parked and shut down. The gust lock pin immobilizes pitch and roll controls to prevent wind damage to the control surfaces and control systems.

Factual Information

25

Aircraft Accident Report

NOTE

INSURE THAT THREADS ARE VISIBLE THROUGH THE INSPECTION HOLE AT THE END OF THE PUSHROD, AFTER ADJUSTMENTS ARE MADE.

i. Adjust the elevator bellcrank pushrod ends as required to obtain 0 readings on the travel boards for both elevators. Tighten the locknuts.

j. Connect pushrods to the aft elevator bellcrank.

k. Use a cable tensiometer to measure control cable tension.

1. Tighten elevator control cables uniformly to the center of the maximum and minimum tensions.

m. Slowly adjust the elevator cable(s) as required until the rig pin in the forward bellcrank will fit with minimum insertion force.

n. Remove the rig pin from the forward bellcrank.

o. Read the cable tension of both cables. Record the values as T up and T down. The allowable limit for combined cable tension is $66 \pm 8.0$ lbs. Compute the combined tension load (Tc) by adding the up and down values together and dividing by 2, or:

$$ \begin{array}{c} Tc = (T \text{ up} + T \text{ down}) / 2 \\ \text{Example:} \\ T \text{ up} = 57 \text{ lbs.}, T \text{ down} = 75 \text{ lbs.} \\ Tc = (T \text{ up} + T \text{ down}) / 2 \\ Tc = \frac{57 + 75}{2} = \frac{132}{2} = 66 \text{ lbs.} = \text{acceptable} \end{array} $$

p. Compare the combined tension load to cable tension graph.

q. If the combined tension load is above the maximum range, start the procedures over and set cable tension at the minimum, or shorten the pushrod of the forward elevator bellcrank and repeat Steps 1 through p.

r. Install lock clips on turnbuckles.

s. Remove the rig pin from the aft elevator bellcrank.

t. Set the forward elevator bellcrank stop bolts for a clearance of $0.37 \pm 0.06$ inch.

u. Verify the bob weight stop bolt clearance is $0.5 \pm 0.06$ inch. Adjust the stop bolt, if necessary. The push-pull tube assembly between the control column and the forward elevator bellcrank may need to be adjusted to achieve the required clearance (refer to Step e).

v. The forward elevator bellcrank stop bolt is to make contact before the bob weight stop bolt makes contact with the bob weight. In order to verify that the forward bellcrank bolt makes contact before the bob weight stop bolt makes

Factual Information

27

Aircraft Accident Report

According to the Air Midwest Beech 1900D load manifest form for flight 5481, the operating empty weight was 10,673 pounds,$^{39}$ the passenger weight was 3,325 pounds, the weight in the coat closet was 10 pounds, the AFT1 cargo compartment weight was 775 pounds, the AFT2 cargo compartment weight was 45 pounds, the zero fuel weight was 14,818 pounds, the fuel weight at takeoff was 2,200 pounds, and the gross takeoff weight was 17,018 pounds. The flight crew made a 10-pound addition error when summing the weights that comprise the zero fuel weight. As a result, the calculated zero fuel weight was actually 14,828 pounds,$^{40}$ and the calculated gross takeoff weight was actually 17,028 pounds. The Beech 1900D maximum gross takeoff weight is 17,120 pounds.

The load manifest form also indicated that the calculated CG index for the accident flight was 81 (37.8 percent MAC). The Air Midwest CG takeoff limits range from indexes of about 23 to 85 (16.7 to 39.2 percent MAC) when a Beech 1900D airplane is at a gross takeoff weight of 17,028 pounds. The Beech 1900D aft CG limit is 40 percent MAC.

Table 2 shows information reported on Beech 1900D load manifest forms for flights flown by the accident airplane after the January 6, 2003, D6 maintenance check. The table shows that the accident flight was calculated to be the most aft loaded of all of the postmaintenance flights. Table 5 in section 1.16.2.2 provides information from the load manifest forms of selected flights before the D6 maintenance check.

Table 2. Weight and balance information for the 10 flights after maintenance.

Factual Information

38

Aircraft Accident Report

〔图〕AND Tumbuckle7.30 inchesANU Turnbuckle5.54 inches -Tumnbuckle BarrelTooling HoleThreaded Cable Terminals\n

Figure 6. Turnbuckles as found in the wreckage.

The AND turnbuckle was extended 1.76 inches more than the ANU turnbuckle. After the accident, Air Midwest surveyed its entire fleet of 42 Beech 1900D airplanes, which represented 25 percent of the 164 Beech 1900D airplanes active in the North American fleet. Air Midwest data submitted to the Safety Board indicated that, on average, the AND turnbuckle was extended 0.04 inch less than the ANU turnbuckle.

The forward bellcrank had fractured into two pieces, and the fractured surface did not exhibit the soot that covered the bellcrank side surfaces and surrounding structure. The forward bellcrank forward (AND) stop bolt was bent and measured 0.93 inch in length. The control cables and the aft end of the forward push-pull tube were attached to the forward bellcrank.

The forward push-pull tube and rod end were bent. The push-pull tube was measured in segments at the Safety Board's Materials Laboratory, and the total length of the rod was 15.33 inches. The rod end had six threads visible.$^{56}$

Factual Information

45

Aircraft Accident Report

weight (in the coat closet and the AFT1 and AFT2 cargo compartments); and the fuel weight. The information sources for the airplane component weights used in the Board's weight and balance study are shown in table 3.

Table 3. Information sources for airplane component weights.

Note: The Air Midwest average weight program in effect at the time of the accident is discussed in section 1.18.1.2.

The results of the airplane component buildup method for flight 5481's weight and CG position were based on the following: the airplane operating empty weight reported on maintenance records; the flight crew and passenger weights documented by the medical examiner; the dry weight of personal effects and the dry weight of checked, carry-on, and crew baggage,$^{68}$ 45 pounds in the AFT2 cargo compartment (the actual weight of a tire stored in the compartment); 10 pounds in the coat closet; and 2,310 pounds of fuel. On the basis of this information, the Safety Board estimated that flight 5481 had a minimum gross takeoff weight of 17,443 pounds and a CG position of 43.3 percent MAC, which was outside of the Beech 1900D weight and CG envelope.

Factual Information

46

Aircraft Accident Report

1.16.2.1.3 Takeoff Ground Roll Weight Analysis

Because the accident airplane was partially consumed by a postcrash fire, some of the wreckage evidence was not recoverable. As a result, the Safety Board used FDR data from the accident flight and a model of the airplane's takeoff ground roll dynamics to calculate the airplane's actual weight.

The accident airplane's weight during takeoff was calculated by evaluating the forces and moments on the airplane during the takeoff ground roll. Engine, groundspeed, and downrange distance data provided by Raytheon Aircraft Company for three Beech 1900D flight test takeoffs were used to validate this approach. As shown in table 4, the Safety Board's force and moment model predicted the actual flight test airplane weights within about ±200 pounds for groundspeeds between the 70- and 100-knot range.

Table 4. Validation of takeoff ground roll model.

The accident flight's weight was subsequently derived using FDR data for the left and right engine rpm, right engine torque, altitude, airspeed, and corrected longitudinal acceleration.$^{69}$ Thrust was calculated using the FDR airspeed and the radar-derived groundspeed data. Because of the lack of reliable FDR data for left engine torque, FDR right engine torque data were used for both of the engines. The FDR left and right engine rpm values were averaged to define the propeller rpm.

On the basis of these FDR data, radar data, the validated force and moment model, and component weight and location information, the Safety Board determined that the accident airplane had an actual weight of 17,700 ±200 pounds and an actual CG position of 45.5 ±2 percent MAC. Figure 7 shows the results of the Board's calculations overlaid on the Beech 1900D weight and CG envelope.

Factual Information

47

Aircraft Accident Report

〔图〕插图

Figure 7. Weight and center of gravity information for flight 5481.

1.16.2.2 Load Manifest Study

The Safety Board reviewed load manifest data for 86 flights of the accident airplane that occurred between December 22, 2002, and January 8, 2003, and identified 9 flights that were documented to be similarly loaded to the accident flight. As shown in table 5, flight 5512 on December 27, 2002, was the most closely loaded to the accident flight. FDR data from flight 5512 were used (along with other data) to validate the Board's Beech 1900D simulator and kinematics extraction tool (see section 1.16.2.4).

Factual Information

48

Aircraft Accident Report

Table 5. Flights documented to be similarly loaded to the accident flight.

Note: The accident flight appears in bold face print.

ªThe CG indexes that correspond to these CG positions are 81, 70, 69, 74, 72, 75, 81, 67, 67, and 72, respectively.

$^{b}$ This figure includes the 10-pound addition error made by the flight crew on the load manifest form.

$^{c}$ One of the 15 passengers was a child between 2 and 12 years of age. In calculating the weight and balance of the airplane, the flight crew used the weight figure for children between 2 and 12 years of age for this passenger (80 pounds).

1.16.2.3 Elevator Control Authority Study

The Safety Board conducted an elevator control authority study based on data from the accident airplane's FDR to determine (1) the relationship between the FDR pitch control position and the actual elevator position and the change in that relationship before and after the D6 maintenance check and (2) the elevator travel that was available after maintenance. The results of the elevator control authority study are discussed in sections 1.16.2.3.1 and 1.16.2.3.2, respectively.

1.16.2.3.1 Relationship Between Pitch Control Position and Actual Elevator Position

The Safety Board analyzed the pitch control position and pitch trim control position data that were recorded during the climb and cruise portions of 84 flights available on the accident airplane's FDR.$^{70}$ The flight phase (climb or cruise) and the

Factual Information

49

Aircraft Accident Report

flight date (before or after the D6 maintenance check) were recorded for each FDR pitch control position and pitch trim control position data pair.

Raytheon Aircraft Company provided flight test certification data for elevator and elevator trim tab climb and cruise positions for various Beech 1900D weights and CG positions. These data were used to verify the absolute position of the accident airplane's elevator for a given FDR pitch control position and pitch trim control position pair.

The position of the trim tab is the predominant predictor of elevator position during climb and cruise flight. Figure 8 shows that a strong correlation exists between elevator (FDR pitch control position) and trim tab (FDR pitch trim control) data. The figure shows two distinct data groups: the larger group represents flights before the D6 maintenance check, and the smaller group represents flights after the D6 maintenance check.

〔图〕插图

Figure 8. Flight data recorder pitch control and pitch trim control positions.

The larger group of data shown in figure 8 (that is, the flights before D6 maintenance) compared well with the Beech 1900D flight test certification data, indicating that the premaintenance FDR pitch control position data represented the actual elevator position to within about 1°. However, the smaller group of data (that is, the flights

°

°

Factual Information

51

Aircraft Accident Report

〔图〕插图

Figure 9. Elevator travel available before and after the detail six maintenance check.

The Safety Board's Beech 1900D simulator was subsequently used to derive the elevator required to (1) match the accident airplane's flight motion recorded on the FDR using the actual airplane weight and CG position and (2) establish controlled flight from an initial upset condition defined by the accident flight profile using the actual airplane weight and CG position. The results of the simulation studies are presented in sections 1.16.2.4.1 and 1.16.2.4.2, and the results of the kinematics extraction are presented in sections 1.16.2.4.3 and 1.16.2.4.4.

1.16.2.4.1 Elevator Required to Match Accident Flight Data

The Safety Board's simulator requires that the aircraft flight condition (altitude, airspeed, weight, and CG), the configuration (flaps and gear), the flight control inputs or control surface inputs, and engine data be specified.$^{71}$ The aerodynamic, propulsion, and mechanical models were used to compute the forces and moments acting on the airplane. The integration of the governing equations of motion yielded the resulting aircraft motion as a function of time.

Factual Information

57

Aircraft Accident Report

〔图〕插图 Note: RAMS, Regional Airline Maintenance Service.

Figure 10. Organizational structure of Air Midwest's Huntington, West Virginia, maintenance station operations.

During weekdays at HTS, the overnight shift started about 2200, an airplane arrived about 2230, and the shift ended about 0630. During weekends at HTS, the overnight shift started about 1800; an airplane arrived about 1830; and the shift ended between about 0200 and 0400, depending on the workload. Between five and seven maintenance personnel normally worked the overnight shift. The maintenance personnel indicated that they would remain on duty until the work was completed and that they were not pressured to return an airplane to service.

The RALLC site manager stated that the workload at HTS was “fairly stable.” He indicated that HTS was scheduled for one airplane per night and that a detail check would normally be scheduled only on 1 or 2 nights per week. The workload was managed according to Air Midwest maintenance control’s 3-day forecast.

Seven maintenance personnel (a quality assurance inspector, a foreman, and five mechanics) were on duty when the accident airplane received its D6 maintenance check on January 6, 2003. The quality assurance inspector and the foreman had been employed at HTS since it opened in July 2002. The five mechanics had been employed at HTS between 2 and 7 weeks. The quality assurance inspector and the foreman had completed

Factual Information

71

Aircraft Accident Report

Table 6. Federal Aviation Administration Notice 8400.40 survey results by operator.

Note: Operator “A” is Air Midwest. Operator “J” did not properly conduct the survey; thus, that operator’s data are not reflected in the survey results. The symbol “in this table indicates that the operator used actual weights.

Table 7. Federal Aviation Administration Notice 8400.40 survey results by category.

If the average sample weights were greater than the average weights in the operators' weight and balance program, the operators were to use the new average weights. If an operator thought that the increased weights were inappropriate, then the operator could either conduct a new validation survey using the criteria prescribed in AC 120-27C to determine new average weights or use actual weights. At the public hearing on this accident, an FAA air safety investigator from the Air Carrier Operations Branch stated that all 15 operators that used average weight programs adjusted the weights in one or more categories by 5 to 25 percent.

Factual Information

75

Aircraft Accident Report

on May 11, 2003, and effectively reduced the Beech 1900D airplane capacity to a maximum of 17 passengers. The average weight for adult passengers increased from 175 pounds (November through April) and 170 pounds (May through October) to 200 pounds all year.$^{103}$ These passenger weights still include 10 pounds for items that are normally carried on board by passengers, such as handbags, briefcases, and laptop computers. Articles that are stored in the coat closet are counted as 10 pounds per bag or item. The average weight for checked baggage increased from 25 to 30 pounds, and the average weight for carry-on baggage stowed in the AFT1 cargo compartment decreased from 25 to 20 pounds. These weight and baggage revisions were incorporated into the Flight Operations Procedures Manual on June 12, 2003.

Table 8. Air Midwest revised weights compared with previous weights.

Air Midwest also revised its Beech 1900D load manifest form. The revised form showed the new passenger and checked baggage average weights and added the carry-on baggage weights. The revised form also included a table to calculate the total weight and CG index number for the flight crew, which was previously calculated as part of the airplane's basic operating weight, and an updated passenger index table. The revised load manifest form and instructions were incorporated into the Flight Operations Procedures Manual on April 25, 2003.

1.18.1.3 US Airways Express

Because Air Midwest flight 5481 was operating under a code-share agreement with US Airways Express, the flight was subject to US Airways Express' procedures for passenger, baggage, and cargo loading. Gate and ramp agents use the US Airways Express Load Report, form OF-11E, to account for all passengers, baggage, and cargo loaded on a US Airways Express flight. The OF-11E must be given to the flight crew on all flights except for ferry flights. The flight crew uses this information to complete the load manifest form.

The Air Midwest Flight Operations Procedures Manual and the US Airways Express Ground Operations Manual, Section 5, “Baggage & Cargo Loading,” dated September 1, 2002, cited the specific items that needed to be completed on the OF-11E.

Factual Information

80

Aircraft Accident Report

Figure 11 shows flight 5481's weight and CG information compared with that for the flight 103 accident airplane. Two data points in the figure show the CG range for the flight 103 airplane.

〔图〕插图

Figure 11. Flight 5481 weight and center of gravity information compared with flight 103 weight and center of gravity information.

1.18.4.2 ValuJet Airlines Flight 592, Everglades, Near Miami, Florida

On May 11, 1996, ValuJet Airlines flight 592, a Douglas DC-9-32, N904VJ, crashed into the Everglades, near Miami, Florida, about 10 minutes after takeoff from Miami International Airport. The 2 pilots, 3 flight attendants, and all 105 passengers were killed. Flight 592 was operating under 14 CFR 121 with a scheduled destination of the William B. Hartsfield International Airport, Atlanta, Georgia.

A fire erupted in the airplane's class D cargo compartment. The fire was initiated by the actuation of one or more oxygen generators being improperly carried as cargo. The oxygen generators were prepared and packaged for carriage aboard flight 592 by SabreTech, a 14 CFR Part 145 repair station in Miami that performed heavy maintenance for ValuJet Airlines.

Factual Information

90

Aircraft Accident Report

1.18.6 Other Passenger and Baggage Weight Information

1.18.6.1 Civil Aviation Authority Report

The Civil Aviation Authority (CAA), the FAA's counterpart in the United Kingdom, conducted a survey between November 1981 and January 1982 to determine the distribution of passengers and their carry-on baggage aboard scheduled air carrier flights. The survey was conducted 18 times at Heathrow Airport because of its high turnover during the day and its wide variety of international and domestic flights. Departing international and domestic passengers and their carry-on baggage were weighed on a scale with a digital readout; testing at the beginning and the end of the survey found that the scale was accurate to within 0.5 percent.

International passengers were asked about where they were traveling, whether they were traveling for business or leisure, and whether they had checked any baggage. The international passengers and their carry-on baggage were then weighed. Domestic passengers were asked whether they had checked any baggage, and then the passengers and their carry-on baggage were weighed. The CAA's survey results appear in table 9.$^{122}$

Table 9. Results of the Civil Aviation Authority's weight and carry-on baggage survey.

$^{a}$ For this report, the average weights were converted from kilograms to pounds and then were rounded to the nearest whole number.

$^{b}$ Of the 1,928 male passengers surveyed, 1,632 were international passengers, and 296 were domestic passengers.

$^{c}$ Of the 769 female passengers surveyed, 692 were international passengers, and 77 were domestic passengers.

$^{d}$ The child and infant passengers surveyed were not categorized according to whether they were international or domestic passengers. A child was considered to be between 2 and 11 years of age, and an infant was considered to be under 2 years of age.

e The carry-on baggage was not categorized according to whether the bags belonged to international or domestic passengers.

According to the CAA's report, the average weight for male passengers was between 13 and 16 pounds more than the CAA's male passenger weight at the time of the survey (about 165 pounds). The average weight for female passengers was between 2 and

Factual Information

92

Aircraft Accident Report

sliding-scale method of calculating standard weights particular to the aircraft capacity should be adopted.

The Flight Standards Division proposed passenger standard weights to be used when actual passenger weights are not used. The proposed standard weights for adult males and females, infants, children, and adolescent males and females decreased as the maximum seating capacity of the airplane increased. The standard weights for an airplane with a maximum seating capacity of 15 to 19 passengers and crewmembers are shown in table 10. (The Beech 1900D has a maximum seating capacity of 19 seats.)

Table 10. Australian Department of Transport and Communications proposed standard passenger weights for airplanes with a maximum seating capacity of 15 to 19 seats.

ª For this report, the average weights were converted from kilograms to pounds and then were rounded to the nearest whole number.

1.18.6.3 Civil Aviation Authority of New Zealand Survey

In 2003, the CAA of New Zealand commissioned a survey to identify the current average weight of domestic passengers traveling aboard large airplanes with more than 30 passenger seats, medium-sized airplanes with 10 to 30 passenger seats, and helicopters and airplanes with 9 or fewer passenger seats. Previous passenger weight surveys by the CAA of New Zealand showed that the current standard weight at the time—equivalent to about 170 pounds—was out of date. A 1999 study showed that the average weight for passengers traveling aboard airplanes with more than 30 passenger seats was equivalent to about 187 pounds. Also in 1999, the New Zealand Ministry of Health found that the average weight of New Zealanders was increasing by 3 kilograms (6.6 pounds) per decade.

During October and November 2003, 15,414 domestic passengers who were 13 years of age and older were weighed, first with their carry-on baggage and then without the baggage. Most of the passengers were weighed at three major airports—Auckland, Wellington, and Christchurch—and a small number of passengers were weighed at the Queenstown airport. Of the 15,414 survey participants, 8,491, or 55 percent, were male, and 6,923, or 45 percent, were female. The results of the survey are shown in table 11.

Factual Information

93

Aircraft Accident Report

Table 11. Results of the Civil Aviation Authority of New Zealand's weight and carry-on baggage survey.

$^{a}$ For this report, the average weights were converted from kilograms to pounds and were then rounded to the nearest whole number.

The 2003 average weight for male passengers with their carry-on baggage increased from the 1999 figure by about 3 pounds. The 2003 average weight for female passengers with their carry-on baggage increased from the 1999 figure by about 7 pounds.

According to the survey report,$^{124}$ the results of the survey will be incorporated into proposed rule amendments that are expected to become effective in the second quarter of 2004.

Analysis

95

Aircraft Accident Report

manifest indicated a taxi fuel burn of 220 pounds, even though Air Midwest assumes a taxi fuel burn of 110 pounds. These figures resulted in a calculated airplane weight of 17,028 pounds and a CG position of 37.8 percent mean aerodynamic chord (MAC), which were within the Beech 1900D certified weight and CG limits of 17,120 pounds and 40 percent MAC, respectively.

The two ramp agents assigned to the accident flight handled the baggage according to company procedures and interacted appropriately with the flight crew. In a postaccident interview, one of the ramp agents reported that he told the captain that 2 of the 31 bags aboard the airplane had an estimated weight of between 70 and 80 pounds. However, the bags did not have a heavy bag tag attached to them. (Gate agents use these tags to indicate an overweight bag, that is, a bag that weighs between 70 and 100 pounds.) Also, the bags were not identified as overweight on the OF-11E form (the US Airways Express Load Report that is used to account for all passengers, baggage, and cargo loaded on a US Airways Express flight). As a result, the flight crew was not required to account for the extra weight of the reportedly heavy bags on the load manifest form.

Even if the flight crew had (1) accounted for the two reportedly heavy bags or (2) accounted for the two heavy bags, estimated an additional 110 pounds of fuel at takeoff,$^{127}$ and recorded the 12-year-old passenger's weight as 80 pounds rather than 175 pounds,$^{128}$ the Air Midwest weight and balance program would still have indicated that flight 5481 was within the Beech 1900D certified weight and CG limits, as shown in table 12. However, for the second scenario, flight 5481's calculated weight would have exceeded the Beech 1900D weight limit if the 12-year-old passenger's weight had remained 175 pounds.

Table 12. Calculations for Air Midwest flight 5481 using the weight and balance program in effect at the time of the accident.

°

Analysis

105

Aircraft Accident Report

Table 13 compares the weight and CG position for flight 5481 using Air Midwest's weight and balance program that was in effect at the time of the accident and Air Midwest's revised weight and balance program with the actual weights and CG positions that would have been present at takeoff.

Table 13. Weight and center of gravity calculations for flight 5481.

$^{a}$ These calculations assume that two passengers seated in the last row of the airplane and three bags from the AFT1 cargo compartment were off-loaded.

$^{b}$ These calculations are based on a scenario in which the airplane had 50 fewer gallons of fuel at the time of departure and eight checked bags were off-loaded.

The Safety Board concludes that Air Midwest's weight and balance program at the time of the accident was not correct and resulted in a substantially inaccurate weight and balance calculation for flight 5481. The Safety Board further concludes that Air Midwest's revised weight and balance program is also unacceptable because it may result in an inaccurate calculation of an airplane's CG position. Section 2.5.2.4 discusses the FAA's oversight of air carrier average weight and balance programs and other sources of error in weight and balance calculations.

2.5 Organizational and Management Factors

2.5.1 Air Midwest

As previously stated, Air Midwest contracted with RALLC for maintenance work at HTS, and RALLC contracted with SMART to provide mechanics for this work. According to AC 120-16D, air carriers can use contractors to accomplish maintenance as long as the air carrier retains the responsibility for the performance and approval of that maintenance. Thus, Air Midwest was responsible for all of the maintenance that was performed on its airplanes. However, none of the seven maintenance personnel who were on duty at HTS on the night of January 6, 2003, were Air Midwest employees. In fact, Air Midwest's sole representative at the HTS maintenance station, the regional site manager,

Analysis

114

Aircraft Accident Report

deficiencies.”$^{139}$ (Section 2.5.2.2 discusses the results of an investigation conducted by the Department of Transportation Office of Inspector General in response to these findings.)

The Safety Board concludes that Air Midwest's CASS program was not being effectively implemented because it did not adequately identify deficiencies in the air carrier's maintenance program, including some that were found by the FAA before the flight 5481 accident.

2.5.1.4 Detail 6 Inspection Procedures Checklist

The Air Midwest D6 inspection procedures checklist (also known as the D6 work card) was the document that mechanics used for inspecting and adjusting elevator control cables. The work card referred the mechanics to the Beech 1900D AMM for additional details regarding the cable inspection and adjustments. The work card and the AMM contained general instructions to guide the mechanics in performing these tasks.

The D6 work card instructed mechanics to check cable tension according to Beech 1900D AMM chapter 27.$^{140}$ However, the D6 work card did not specifically refer the mechanics to section 27-30-02 of the AMM, which contained the only reference—an elevator cable tension graph—in chapter 27 for determining cable tension. Further, to use the graph, the mechanic was first required to determine cable temperature, but neither the D6 work card nor the elevator cable tension graph described how to prepare the airplane for temperature measurement or how and where to obtain temperature readings. The Safety Board’s January 2003 ground tests showed that temperature readings varied depending on the method used by individual mechanics for measuring cable temperature and that the method used for measuring temperature could affect cable tension.

Interviews with Air Midwest and Raytheon Aircraft Company officials revealed that, when cable tension was found to be outside of acceptable parameters, mechanics were expected to perform the entire elevator control system rigging procedure. Neither the D6 work card nor the Beech 1900D AMM explicitly stated that the entire rigging procedure needed to be performed or that the elevator cable tension adjustment could not be accomplished as an isolated task.

Analysis

116

Aircraft Accident Report

maintenance procedures but fail to share this information with other air carriers that operate the same airplane. If aircraft manufacturers were involved in the process of reviewing and revising maintenance procedures, safety information would more likely be shared among air carriers.

In addition, the flight 5481 accident demonstrated that the usability of aircraft maintenance manuals is as important to safety as the usability of work cards, but Safety Recommendation A-03-31 did not address the added safety benefit of revising procedures contained in aircraft maintenance manuals. As a result, the Safety Board classifies Safety Recommendation A-03-31 “Closed—Superseded.”

The Safety Board concludes that accurate and usable work cards developed jointly by air carriers and aircraft manufacturers would improve the performance of maintenance for critical flight systems. Therefore, the Safety Board believes that the FAA should require 14 CFR Part 121 air carriers to implement a program in which carriers and aircraft manufacturers review all work card and maintenance manual instructions for critical flight systems and ensure the accuracy and usability of these instructions so that they are appropriate to the level of training of the mechanics performing the work.

2.5.2 Federal Aviation Administration

2.5.2.1 Oversight of Air Midwest

The Air Midwest PMI had identified several deficiencies in the air carrier’s maintenance training program during the 2 years before the accident. These deficiencies included inadequate oversight of training, inadequate record-keeping, inadequate training procedures, and lack of adherence to company training procedures. In letters to Air Midwest, the FAA expressed its concerns and urged the air carrier to improve. For example, in an October 16, 2000, letter, the Air Midwest PMI stated that a recent FAA inspection conducted at Air Midwest headquarters could not be completed because the air carrier lacked adequate records to determine whether it was in compliance with 14 CFR 121.371 and 121.375.$^{143}$ Specifically, the PMI noted that required maintenance training records were either not at company headquarters or were not up to date and complete. A followup inspection of Air Midwest’s training files was conducted on December 4, 2000. This inspection revealed evidence of continuing deficiencies in the air carrier’s training records as well as deficiencies in its training manual. The PMI gave Air Midwest a deadline of February 1, 2001, to remedy these deficiencies.

The FAA's March 2001 RASIP inspection found that deficiencies still existed in the maintenance training manual and that training was not being conducted in accordance with company procedures. Specifically, the inspection found that Air Midwest did not have adequate procedures for verifying the experience of new hires,$^{144}$ information in the

Recommendations

136

Aircraft Accident Report

BY THE NATIONAL TRANSPORTATION SAFETY BOARD

ELLEN ENGLEMAN CONNERS

Chairman

JOHN J. GOGLIA

Member

MARK V. ROSENKER

Vice Chairman

CAROL J. CARMODY

Member

RICHARD F. HEALING

Member

Adopted: February 26, 2004

Appendix B

Aircraft Accident Report

Appendix B

142 Aircraft Accident Report

Appendix B

143 Aircraft Accident Report

INTRA-COCKPIT COMMUNICATION

AIR-GROUND COMMUNICATION

Appendix B

145

Aircraft Accident Report

Appendix B

Aircraft Accident Report

Appendix B

Aircraft Accident Report

Appendix B

150 Aircraft Accident Report

Appendix B

153 Aircraft Accident Report

Appendix B

155 Aircraft Accident Report

Appendix B

156 Aircraft Accident Report

Appendix B

Aircraft Accident Report

Appendix B

160 Aircraft Accident Report

Appendix B

Aircraft Accident Report

Appendix B

Aircraft Accident Report

INTRA-COCKPIT COMMUNICATION

AIR-GROUND COMMUNICATION

TIME and SOURCE CONTENT

Appendix B

169 Aircraft Accident Report

Appendix B

170 Aircraft Accident Report

INTRA-COCKPIT COMMUNICATION

AIR-GROUND COMMUNICATION

Appendix B

175 Aircraft Accident Report

Appendix B

176

Aircraft Accident Report

INTRA-COCKPIT COMMUNICATION

AIR-GROUND COMMUNICATION

INTRA-COCKPIT COMMUNICATION

AIR-GROUND COMMUNICATION

Appendix B

183

Aircraft Accident Report

INTRA-COCKPIT COMMUNICATION

AIR-GROUND COMMUNICATION

Appendix B

186 Aircraft Accident Report

TIME and SOURCE CONTENT

Appendix B

190 Aircraft Accident Report

Appendix C

196

Aircraft Accident Report

Raytheon Aircraft

BEECH 19000 AIRLINER MAINTENANCE MANAUL

〔图〕。00£00LEFTELEVATORCONTROL HORNLEFTELEVATORBSTOP BDLTSDETAIL ALEFT ELEVATORCONTROL HORNSUPPORT20°+1°0°UPDEFLECTION FROMTH

Figure 201 - Elevator Control Horn Stop Bolts

Appendix C

197

Aircraft Accident Report

Raytheon Aircraft

BEECH 1900D AIRLINER MAINTENANCE MANAUL

〔图〕BELLCRANK STOPBELLCRANKSTOP0.37 ± 。O6 INCH0.37D6INCHSTOPBOLTELEVATOR FORWARDBELLCRANK\n

Figure 202 - Elevator Forward Bellcrank Stops

Appendix C

198

Aircraft Accident Report

Raytheon Aircraft

BEECH 1900D AIRLINER MAINTENANCE MANAUL

〔图〕插图

Figure 203 - Elevator Cable Tension Graph

Appendix C

199

Aircraft Accident Report

Raytheon Aircraft

BEECH 1900D AIRLINER MAINTENANCE MANAUL

〔图〕。000000220。10BDETAIL121.UPFER BOLT2.WASHER (2 PLACES)3.UPPER ROLLER4.WASHER5.NUT6.LOWER SCREW7.WASHER (2 PLACES8.LOWER R

Figure 204 - Control Column Rollers