NTSB Factual Report

HISTORY OF FLIGHT

On June 7, 2003, approximately 1400 central daylight time, a Beech (Raytheon) 1900D, N901SK, operated by Skyway Airlines Inc. as flight 1233, experienced lateral control problems shortly after takeoff from General Mitchell International Airport (MKE), Milwaukee, Wisconsin. An emergency was declared and the flight returned to MKE, landing without incident at 1410. The two crewmembers and six passengers reported no injuries. Visual meteorological conditions prevailed at the time of the incident. The flight was conducted under 14 CFR Part 121 and was on an instrument flight rules (IFR) flight plan. The flight departed MKE at 1345 with an intended destination of Sawyer International Airport (SAW), Gwinn, Michigan.

According to the captain's written statement, immediately after takeoff the first-officer, who was the pilot flying, announced that approximately 45-degrees of right control wheel deflection was required to maintain straight flight. The captain reported they climbed to 7,000 feet mean sea level in order to evaluate the problem. He stated: "We then elected to make several maneuvers at altitude to determine whether we had full flight control authority. No anomalies were evident." He reported that after notifying air traffic control and Skyways operations of the situation, they elected to return to MKE for a precautionary landing. Concerning the approach and landing, the captain noted: "We elected to perform a no flap landing, since we had no idea what was causing the anomaly. Approach and landing were normal other than the required yoke deflection needed the entire time."

Airline maintenance personnel conducted a post-incident inspection of the aileron system. According to their statement, the aileron system rig pin was installed in an attempt to isolate the problem. The aileron control surfaces were aligned at the neutral (zero deflection) position indicating the problem was between the rig pin and the control wheels. The mechanics reported that closer inspection of the system revealed that the captain's (left side) aileron control column sprocket assembly was cracked at the silver braze between the aileron drive sprocket and the shaft. This allowed both control wheels to be rotated without corresponding movement of the aileron control surfaces.

PERSONNEL INFORMATION

The pilot-in-command (PIC) held an Airline Transport Pilot (ATP) certificate with single and multi-engine land ratings. Single-engine operations were limited to commercial pilot privileges, according to Federal Aviation Administration (FAA) records. The certificate listed a type rating for the BE-1900, with a limitation stating "BE-1900 Second in Command Required."

The PIC held a First Class medical certificate issued on May 27, 2003. This certificate listed a restriction stating: "Must wear corrective lenses." He had accumulated approximately 3,400 hours total time. His most recent 14 CFR Part 121 training event was on May 7, 2003.

The second-in-command (SIC) held an ATP certificate with single and multi-engine land ratings. Single-engine operations were limited to commercial pilot privileges, according to FAA records. The certificate listed a type rating for the BE-1900, with a limitation stating "BE-1900 Second in Command Required."

The SIC held a First Class medical certificate issued on October 21, 2002. This certificate listed a restriction stating: "Must wear corrective lenses for near and distant vision." He had accumulated approximately 3,920 hours total time. His most recent 14 CFR Part 121 training event was on July 16, 2002.

AIRCRAFT INFORMATION

The incident aircraft was a 1994 Beech (Raytheon) 1900D Airliner. Skyway Airlines was operating the aircraft in 14 CFR Part 121 revenue service at the time of the incident. The 1900D is a twin-engine, turbo-prop, pressurized aircraft certified to 14 CFR Part 23 Commuter Category regulations through Amendment 34.

The aircraft was issued an airworthiness certificate on March 29, 1994. According to the operator, it had accumulated 20,720 hours total time over 25,472 cycles. The most recent maintenance inspection under the operator's continuous airworthiness program was completed on May 28, 2003, at 20,678 hours.

According to maintenance records, the night before the incident flight a flight data recorder (FDR) functional check was completed. The mechanic who conducted the functional check reported the procedure was routine. He stated that no discrepancies were noted, the FDR performed per the specification and the aircraft was approved for return to service.

The incident flight was the first flight of the day. According to the MKE chief pilot, the same crew assigned to the incident flight preformed the first flight of the day checks. He reported no anomalies were found during those checks.

The 1900D lateral control system consists of a direct mechanical connection via control cables between the Captain's and First Officer's control wheels and the aileron control surfaces. The primary elements of this system are the control wheels, sprocket assemblies, connecting cables, bellcranks, pushrods and ailerons.

METEOROLOGICAL CONDITIONS

Weather conditions at MKE, recorded at 1352, were: scattered clouds at 6,000 feet above ground level (agl), scattered clouds at 7,500 feet agl, 4 statute miles visibility in haze, and winds from 130 degrees at 11 knots.

TESTS AND RESEARCH

The captain's (left side) control wheel torque tube was connected to the Aileron Sprocket Assembly (Raytheon Part No.: 100-524120-1) through a universal joint. The sprocket assembly consisted of a shaft and two sprockets. Each sprocket was silver brazed to the shaft.

Raytheon engineering drawing, Sprocket Assembly - Control Column, Aileron, No.: 100-524120, dated 2/28/1972, specified the braze operation to be in accordance with MIL-B-7883. A general note required that the completed assembly be proof loaded prior to acceptance. Specifically, a 3,619 lbs. axial load was to be applied to the shaft and resisted at the sprocket. After testing a proof stamp was required to be applied to the part. The proof stamp was present on the incident part.

An engineering change order was issued which revised the braze note to read: "Silver braze per spec MIL-B-7883 except omit paragraph 4.4.2 radiographic inspection." A general note was added to the drawing at that time which stated: "Visually inspect to see that the braze material has filled the joints between shaft and all mating silver brazed parts 100%." According to the order, the change was initiated because the assembly configuration "does not lend itself to radiographic inspection."

Military Specification MIL-B-7883B which specifies requirements for brazing steels states: "The filler metal shall be introduced at one edge ... and shall flow by capillary action to fill the interstice." The specification also requires removal of flux residue immediately after brazing and cooling. No residual flux is permitted on the surface of the brazed joint per the specification. Any lack of penetration is cause for rejection of the part.

The specification also states that internal defects that do not exceed 15% of the joint area, either collectively for multiple small defects or singly for one large defect, are permitted.

The specification denotes two grades for joint quality. Grade A is applicable for "critical fittings and structural applications." Critical fittings are defined as ones in which "the single failure of which would cause significant danger to operating or other personnel or would result in a significant operational penalty." This is further clarified to specifically include loss of control of an aircraft. The quality assurance inspection required for a Grade A joint includes visual and radiographic examination, as well as a dimensional inspection.

Raytheon Process Standard PS 35010D, Silver Alloy Brazed Joints in Primary Structures (12/05/1952), concerning the placement of filler metal, stated: "If no [joint filler] grooves are specified, the filler metal shall be placed at the edge or end of the joint, or manually fed into the joint during the brazing operation."

The sprocket assembly and the Flight Data Recorder (FDR) were removed and sent to National Transportation Safety Board (NTSB) laboratory facilities for examination. The complete Materials Laboratory and Vehicle Recorders Laboratory Factual Reports are contained in the docket material associated with this report.

NTSB Materials Laboratory examination of the sprocket assembly removed from the incident aircraft revealed a 360-degree fracture on both sides of the braze joint between the larger aileron cable sprocket and the shaft. When the sprocket was removed from the shaft it separated freely, with no ligaments of the braze still intact.

Visual examination noted that filler material had been introduced from both sides of the joint and that the two braze regions did not meet at any point around the circumference of the joint. A lack of penetration of braze filler material was observed over approximately 20% of the joint area. Separation appeared to be at the braze-to-shaft and braze-to-sprocket interfaces. Minimal fracturing through the braze material was observed. Dimensional inspection of the components did not reveal any discrepancies when compared to the drawing requirements.

In addition to the sprocket from the incident aircraft, three additional sprocket assemblies were reviewed for comparison. These assemblies were obtained from: (1) Air Midwest Beech 1900 accident aircraft at Charlotte, North Carolina (NTSB No.: DCA03MA022); (2) Colgan Airways Beech 1900 accident at Yarmouth, Massachusetts (NTSB No.: NYC03MA183); and (3) CommuteAir Beech 1900 in-service aircraft.

The sprocket assembly from the Air Midwest Beech 1900 appeared cracked at the aft sprocket-to-shaft braze joint over approximately 350-degrees of the circumference. After the sprocket was removed from the shaft, a lack of penetration of braze filler material extended over the entire circumference (360-degrees) of the joint and covered approximately 50% of the total joint area.

A grease-like substance covering the braze joint extended from the aft fillet radius to portions of the forward fillet radius. Analysis of the substance revealed a composition consistent with braze flux O-F-499, which was the flux material specified in the assembly drawing.

The sprocket assembly from the Colgan Airways Beech 1900 exhibited complete penetration at the braze joint along the plane sectioned. The shaft portion of the assembly was bent. A crack on the aft side of the braze joint was observed around the entire circumference.

When the sprocket was sectioned a "thumbnail-shaped" region consistent in appearance to copper corrosion was observed. Location of the region was consistent with the approximate location of maximum tensile loads on the shaft due to bending from impact forces. (This component was submerged in sea water for several days following the accident.)

The sprocket assembly from the CommuteAir Beech 1900 exhibited complete penetration of filler material through the braze joint and was not subjected to further examination.

The NTSB Vehicle Recorders Laboratory downloaded the FDR data. The incident aircraft FDR contained over 99 hours of data, which encompassed over 84 flights.

The FDR roll control parameter is designed to measure left aileron movement through a transducer, which measures movement of an aileron cable that runs through the fuselage. The data is transcribed as raw decimal counts, which are then correlated to aileron position.

The relationship between decimal counts and aileron position were: 580 decimal counts at -24 degrees aileron deflection (up); 1644 decimal counts at 0 degrees aileron deflection; and 2593 decimal counts at +17 degrees aileron deflection (down). At the left aileron, positive (+) deflection is down and negative (-) deflection is upward movement of the aileron trailing edge.

The roll control parameter data recorded during the incident flight appeared similar to the roll control data recorded by the FDR on previous flights.

The entire data set was scanned to review the history of the roll control parameter. During the approximately 83 flights prior to the FDR functional check the night before the incident flight, the maximum values reached were +14.6 degrees (down), decimal count 2433, and -23.4 degrees (up), decimal count 620.

The FDR functional check, as it related to the roll control parameter, was to be accomplished by deflecting the control wheel and measuring the degrees of surface deflection on the left aileron using a travel board. The corresponding values recorded by the FDR were then compared to the established limits.

During the functional check the roll control parameters recorded by the FDR attained maximum values of +16.5 degrees (down), decimal count 2558, and -24.4 degrees (up), decimal count 555.

Acceptable parameters for the FDR functional check were outlined in the operator's test procedure. At a left aileron neutral position (0 degree), the acceptable decimal count range was 1599 - 1689. At -24 degrees (up), the acceptable decimal count range was 515 - 645. And at +17 degrees (down), the acceptable decimal count range was 2553 - 2633.

The maximum position of +14.6 degrees (down), decimal count 2433, recorded during the flights prior to the FDR functional check, would not have been sufficient to satisfy the test requirement of 2553 - 2633 decimal counts. The only time the FDR recorded values in this range was for the 4-second time period during the functional check.

Testing related to the application of additional force on the control column and its relation to FDR roll parameter data was conducted at Raytheon Aircraft facilities under NTSB supervision. With the ailerons positioned at the control surface stops, additional force applied to the control wheel will be reacted by slight stretching of the control cables. Because the FDR transducer measures movement of the cable, cable stretching will affect the FDR roll parameter readings.

Aileron cable tensions were set to the estimated tensions of the incident aircraft. When the control wheel was positioned so that the ailerons were resting against the control surface stops, the FDR recorded aileron deflections of 16.0 degrees down and 23.5 degrees up. With a force of 80 lbs. applied at the control wheel, the recorded aileron deflections were 16.7 down and 24.8 up.

The manufacturer provided documentation related to the assumed loading and structural analysis of the sprocket assembly. The assembly was originally analyzed for installation in the Model 200 King Air airplane. In this installation an 88.5 lbs. ultimate load applied at the control wheel was assumed in accordance with FAA airworthiness standards. This resulted in a load on the braze joint of 3,540 lbs. and a margin of safety of 6% with a 50% efficiency factor applied.

In the case of the 1900D Airliner, which was certified in accordance with Commuter Category airworthiness standards, an ultimate load of 4,344 lbs. was assumed applied to the braze joint. According to the manufacturer's data, this resulted in a -15% margin of safety with a 50% efficiency factor applied. The assembly was structurally tested for installation on the 1900D, which resulted in it meeting the ultimate load requirement without failure.

Airworthiness Standards for Commuter Category airplanes (14 CFR 23.397) require the aileron control system to safely sustain an applied moment of 50 times the control wheel diameter -- "50D", where D is the control wheel diameter. In addition, for design weights above 5,000 lbs., the specified values must be increased linearly with weight to 1.35 times the specified values at a design weight of 19,000 lbs. The regulations also require a safety factor of 1.5 to be applied to the limit loads (14 CFR 23.303).

ADDITIONAL INFORMATION

Silver brazing is a process for joining two metals which involves heating the parts to be joined and feeding molten filler metal into the gap between the parts. The filler metal flows to fill the joint b

NTSB Probable Cause

Failure of the aileron sprocket assembly at the sprocket-to-shaft braze joint. Contributing factors were the improper braze procedure used by the part manufacturer at the time of fabrication which resulted in an inferior quality joint and the inadequate quality control (inspection) criteria which failed to identify the improperly brazed joint. An additional factor was the improper inspection procedure utilized by the operator's maintenance personnel, during which excessive force was applied to the control wheel in order to obtain acceptable flight data recorder readings during the functional check.