Learning from the past: American Eagle Flight 3379, uncontrolled collision with terrain. Morrisville, North Carolina December 13th, 1994

Reducing to idle a good engine, deviation from standard operating procedures and poor CRM among the causes. Several airline deficiencies in recruiting, training and  monitoring substandard pilot performance, organizational issues, and FAA oversight failures identified.

Flagship 1

Photo Copyright Jim Cain.N918AE. AMR Eagle British Aerospace Jetstream 3201  



On December 13th, 1994 at 18341(all times herein are eastern standard time) in accordance with the 24-hour clock). a Flagship Airlines Jetstream 3201, doing business as (dba) American Eagle (AMR) flight 3379, crashed about 4 nautical miles southwest of the runway 5L threshold during an instrument landing system ( ILS ) approach to the Raleigh-Durham International Airport (RDU). The flight was a regularly scheduled passenger flight under 14 Code of Federal Regulations (CFR), part 135. Thirteen passengers and the two crewmembers were fatally injured and the other five passengers survived. The airplane was destroyed by impact and fire. The weather at the time of the accident was ceiling 500 feet, visibility 2 miles. Light rain and fog, temperature 38º F. and dew point 36º F.

The crew of flight 3379 arrived at company operations about 1300, prior to the scheduled check-in time of 1311, on December 13, 1994. They were scheduled for a 2-day trip which included three flights the first day, an overnight stay in Greenville, North Carolina and five flights the second day, ending at RDU at 1555. The crew had never flown together prior to this trip sequence. N918AE a British Aerospace Jetstream 3201, arrived at RDU at 1213 on December 13. 1994. The terminating crew reported that the aircraft performed normally and there were no write- ups on the aircraft during the four flights they had made in it. At 1411, the accident crew departed RDU on time in N918AE, as flight 3416 and arrived in Greensboro, North Carolina (GSC) at 1449, 2 minutes ahead of schedule. After the passengers deplaned they taxied the aircraft to Atlantic Aero a fixed-base operator (FBO) on another part of the airport, to allow other flights to access the gate. The crew entered the FBO facility at 1530 and remained in the “break room.” About 1620 the customer service agent discussed the fuel requirements for the flight with the captain. He advised that they had about 1000 pounds of fuel on arrival and would take 700 additional pounds for departure. The fueler distributed 50 gallons on each side of the airplane for a total fuel load of 1700 pounds as requested. The crew left the FBO building about 1650, and the airplane taxied from the ramp about 1700.

The gate agent responsible for flight 3379 estimated that the aircraft returned to the gate area about 1715. She gave the dispatch papers to the captain, and 18 passengers boarded the flight. The baggage and cargo were loaded onto the airplane and she gave the captain the load manifest. The captain indicated that there was a problem with the weight distribution and they discussed the options to remedy the problem. Two bags were removed from the aft cargo compartment and the flight taxied out 8 minutes late at 1753.

About 1818 the agent requested the departure times from flight 3379, and the first officer advised her that they used 53 and 03 (taxi out at 1753 and take off at 1803). The delay was reportedly due to the baggage rearrangement. The agent, who had previously met both pilots, reported they were in good moods. She described the captain as typically quiet and the first officer as outgoing.

The flight plan was for a cruise altitude of 5.000 feet and the time en route was 23 minutes. Flight 3379 was assigned a cruising altitude of 9,000 feet (Based on the information from the cockpit voice recorder, the captain was the flying pilot and the first officer the non-flying pilot (Note from the blogger : now named monitoring pilot). The crew contacted RDU approach control at 1814, and advised that had received Automatic Terminal Information Service (ATIS) “Sierra.” The controller advised the crew to expect runway 5L. Following some discussion about the arrival clearance, the controller stated. “Eagle flight 379 reduce speed to uh … one eight zero then descend to six thousand. The crew received continuing vectors and was switched to the final radar control position at 1825. The final controller instructed them to. “…reduce to one seven zero then descend and maintain three thousand.” At 1828, the controller cautioned them about wake turbulence from a B-727 that they were following and assigned them a heading 0f 190º. At 1530, the final controller advised: “Eagle flight 3379 eight from BARRT [the final approach fix) turn left heading zero seven zero join the localizer course at or above two thousand one hundred cleared ILS five left.” …line crew acknowledged the clearance and the subsequent change to the tower frequency.

They contacted the tower at 1832, and were told “…cleared to land wind zero one zero at eight traffic three and a half mile final a seven twenty seven.” At1832:25 the crew acknowledged the clearance, “Cleared to land five left 379.” This was the last known transmission from the flight. At 1834:17 an unintelligible noise was heard on the frequency.

Data from the flight data recorder (FDR), cockpit voice recorder (CVR) and the RDU radar plot were correlated for the last minute of flight to reconstruct the approach. There was a change in engine noise similar to an increase in engine RPXI at 1832:28.7, seconds after the captain requested “speeds high.”    This was followed immediately by a “all for gear down and flaps 20”. Flight 3379 crossed slightly right of BARRT, the final approach fix, while descending through 2.100 feet and slowing below 160 knots about this time. At 1833:33.3, the captain asked: “Why’s that ignition light on? We just had a flameout? For the next seconds, the crew discussed the engine anomaly as the airplane heading drifted to the left at approximately 2/3 of a degree per second and eventually crossed the localizer centerline at 1833:45. At this time, flight 3379 was approximately 3.8 miles behind the preceding B727.

For the next several seconds, the airplane remained relatively level at approximately 1,800 feet, as the airspeed decreased from 140 knots to 122 knots then, the captain decided “let’s go missed approach.” In less than 2 seconds, at 1834:05.3, two momentary stall warnings occurred as the captain called “State max power” and the left turn rate increased. The first officer called “Lower the nose, lower the nose, lower the nose.” But, the airplane remained at about 1,800 feet, and the airspeed continued to decay to approximately 119 knots as the left turn rate increased to about 5ºper second.

At 1834:09.4 a stall horn warning started again, and was followed at 1834:09.6 by the dual stall warning horns. At this time the airplane was still at 1.775 feet and the airspeed had sank to 111 knots. The first officer inquired “You got it” and the captain responded “Yeah”. The airspeed decreased to 103 knots at 1834:12, and the first officer said. “Lower the nose.” At 1834:13.2 the first officer said “It’s the wrong, wrong foot, wrong engine.” About this time, the rate of descent increased to more than 10,000 feet per minute. The rate of turn increased about 14º per second at 1834:16, as the airspeed increased rapidly. There were several significant normal accelerations during this period. The airplane finally stabilized few seconds before impact at an airspeed of about 170 knots, an acceleration of 2.5 G absolute and a heading of 290º.

The accident occurred during hours of darkness, at 50’ 5″ north latitude and 78″ 52′ 1″west longitude.

Flagship 2

Photo (C) NTSB from Aviation Safety Network 

Damage to Airplane

The airplane was destroyed by impact and fire. There was an intense ground fire in the area of the forward fuselage and wing center section. There was no evidence of pre-impact fire.


Both flight crew members and 13 passengers received fatal injuries from blunt force trauma, and 11 of them sustained thermal injuries from the post-crash fire. Four of the five survivors were ejected from the aircraft during impact and break-up of the cabin, and sustained blunt force traumatic injuries. The fifth survivor crawled out of the wreckage to a safe distance from the fire. He sustained serious injuries”. Two of the survivor died shortly after the arrival at the local hospital.

Post mortem examination

Toxicological specimens were taken from the bodies of the captain and first officer. Chlorpheniramine was found in the captain’s liver and muscle samples. Chlorpheniramine is an antihistamine, not approved for flying, contained in many over-the-counter medicines. It has the potential effect of reducing alertness, slowing reaction and altering perception.



The flight crew was properly certified in accordance with Federal Aviation Regulations and company requirements.

There was no indication of any preexisting discrepancy or pre-impact mechanical failure of the structure, systems, or flight controls of the airplane that contributed to the accident. The airplane was certificated in accordance with appropriate FAA regulations, except for the improper substitution of the FPA-80 for a GPWS. Although the airplane was maintained in accordance with the FAA-approved maintenance program, the discrepancy in the maintenance manual, and the work cards for propeller removal and installation, resulted in both propellers having incorrect flight idle blade angle settings.

The aircraft services provided to flight 3379 by the RDU approach and tower were routine and performed in accordance with requirements.

All components of the runway 5L ILS were operating properly, based on the successful landing of the preceding B727 at 1834 and the flight inspection of all components the following morning. Similarly, the runway and approach lighting systems were operating properly.

Although the weather at RDU included variable low ceilings and reduced visibility in light rain and fog, it was well above minimums for the runway 5L ILS approach. There were several reports of icing by pilots operating in the RDU area at the time of the accident but none were at approach pattern altitude. In addition, the crew discussed the possibility of ice, and had checked for the presence of any during the descent into RDU area. The Safety Board concludes that there were no problems with airframe or engine ice during the approach.

The wake vortex study revealed that flight 379 never encountered the wake vortices from either of the two aircraft immediately preceding it.

There was a discussion between the pilots regarding an anomaly in the left engine, and the captain stated that it had failed. However, the sound spectral analysis showed that the left engine continued operating. Additionally, examination of the internal components of the engines revealed damage that was indicative of similar rotational velocities of the left and right engines. Finally, damage to the propellers, witness marks, and blade bending were consistent with rotation at high power. During the go-around, airplane performance was consistent with the left engine operating at flight idle, gear down and flaps at 20″. Data show that the airplane could not climb in that configuration. Therefore the Safety Board’s analysis of   the accident concentrated on the crew actions, company training and oversight, and the performance capability of the aircraft it was operated.

Crew Actions and Decisions

The captain was the flying pilot on the GSO-RDU leg, and initially used proper crew resource management techniques in calling for the descent and approach checklists, discussing icing conditions, using positive skills for transfer of control of the aircraft and briefing the approach procedures. He also advised the first officer that he was going to remain at 3000 feet rather than descend to 2,100 feet, which he was authorized to do (there was no obvious reason for this decision, so it was particularly appropriate that he informed the first officer of his intention; further, he actually did not remain at 3,000 feet for long).

The flight tests demonstrated that flight idle power was necessary to match the profile as the airplane descended further. After staring “speeds high and then requesting the first officer to configure the aircraft with 20º flaps and gear down, the captain detected an IGN light. Apparently, the IGN light was the result of a transient negative torque condition caused by the combination of low torque at flight idle and rapid movement of the propeller speed reverse to 100 percent.  At that time he asked “Why’s that ignition light on? We just had a flameout?” The first officer responded in about 5 seconds, “I’m not sure what’s goin’ on with it.” After an additional 5 seconds, the captain announced “We had a flameout.”  Following the 10 seconds of relatively silent evaluation, the captain apparently decided that there was a flame out in the left engine. There was no discussion about the specific parameters that led him to the conclusion, so that the first officer could concur. Significantly, having reached the decision that an engine had failed, there was no attempt to feather the propeller and secure the engine. The first officer did not call this fact to the captain’s attention.

During the next 20 seconds, there was almost continuous dialogue as the first officer queried the captain about his conclusions, and the captain confirmed his conclusion. Finally at 1833:55.9, the first officer asked, “Watta you want me to do, you gonna continue? The captain responded, “OK, yeah. gonna continue. Just back me up.” This demonstrated that even when the first officer asked what the captain wanted him to do, the captain did not follow the company procedures for engine failure.

In this circumstances, it is not clear if the first officer was really thinking of the engine out procedures they should have been following or merely seeking assurance that the captain had a specific plan of action. If he was concerned about the failure to follow engine-out procedures, he should have prompted the captain to implement them. If he was skeptical of the captain’s conclusion he should have either challenged him by identifying specific engine indications that the engine was still operating, or suggested additional tests to confirm that the engine had failed. Additionally the first officer did not report the decreasing airspeed.

The captain reversed his initial decision to continue the approach approximately 4 seconds later and announced “Let’s go missed approach.” This represents another decision that is puzzling. The aircraft was positioned for the approach, and all that was required was minimal differential power to continue the approach. However, the aircraft which had leveled at approximately 1,800 feet when the engine anomaly was detected, continued to drift to the left. The rate of turn increased after the call for, “Set max power,” and the airspeed continued to decrease as he continued to maintain a relatively constant altitude of 1,800 feet. The crew did not properly configure the aircraft for a single engine go-around, leaving the left propeller at flight idle, the landing gear down and the flaps at 20º. During this same time interval, there were two stall warnings, which prompted the first officer to say “Lower the nose, lower the nose, lower the nose.”

At this point the captain had responded inappropriately to indications of an apparent engine anomaly, failed to follow company procedures to engine go-around, and stall recovery and was about to lose control of the aircraft. The first officer asked the captain “You got it?” At this time the aircraft was approximately 30º off course, and the captain has not responded to the stall warning o the first officer comments to lower the nose. The captain failed to cope with what was actually a transient anomaly. Good crew resource management that he, as the pilot in command should have assured that control of the airplane was maintained while the problem was analyzed. He had the option of sharing either function with the first officer, or retaining both. He could have transferred control to the first officer, SO that he would be free to analyze the problem, and decide on the proper course of action. Instead he tried to do both and failed.

He continued to attempt to fly the aircraft, unilaterally decided that there was an engine failure, and neither ordered not performed the immediate action items associated with the engine failure checklist. Subsequently, his decision to go-around neither was nor followed by the correct flight procedures. The left turn indicates that he failed to advance both power levers, did not command flaps 10º or gear up and did not maintain adequate airspeed.  If he had advanced both power levels, both engines would have responded, and the perceived emergency would have been resolved. Finally the captain did not follow company procedures for stall avoidance or recovery. He not only failed to control the aircraft, he did not request help from the first officer. Therefore, the Safety Board concludes that the captain’s improper conclusion that the left engine had failed, and his failure to follow established procedures, led directly to the accident.

The exact motivation for some statements by the first officer are unknown, but, based on his reputation, it is assumed that he was applying some crew resource management skills to the situation, in an effort to assist the captain. For               he asked, “K,you got? when the captain decided the engine had failed. He questioned this assessment twice in the next seconds, “We lose an engine?”  “We Iose that en’ left one?” but he never did directly questioned this assessment. He also made two  suggestions to facilitate their situation. He announced that he was going to turn OR engine ignition switches and then asked, Watta you want me to do, you gonna continue?” If he had suggested that they either advance the left power lever to test the engine response or perform the engine failure checklist, there could have been a more positive result. The first officer may hare been about to suggest one of these actions but he was interrupted in mid -entence. “All right I’m gonna…” by the captain’s statement. Let’s go missed approach.” At this point the stall warnings occurred and he focused on trying to get the captain to lower the nose.

It is impossible to determine what control inputs were being made by either crew member, but they had little or no lateral or directional control of the aircraft for the next 13seconds. During that interval the first officer asked “You got it? And made the following prompts:   “Lower the nose;” and “It’s the wrong foot, wrong engine.”   The dual stall warning horns and positive G values recorded by the FDR indicate that the captain induced repeated stick pusher activations with excessive nose-up control column inputs. Finally the first officer said “Here”. This could have signaled his decision to help with rudder input because they were 110ºoff heading. It could have indicated that he was adding power on the left engine, or it could have signaled his decision to take control of the airplane himself. Whatever the meaning, it was too late to recover from the extreme descent rate that developed during the loss of control.

Although the first officer asked the captain twice if they had lost an engine, he did not challenge the captain’s erroneous conclusion with specific information (RPM, EGT, oil pressure, etc) that indicated it was still operating. More importantly, he should have suggested the captain advance the left power lever to see if the engine was operative. Nonetheless, he did continue a supportive role by prompting the captain to lower the nose as they encountered the stall warnings during the early stages of the go-around. Finally, the evidence suggests that he resorted to direct control inputs and power lever movement when he said, “…wrong foot…” and “Here”. Unfortunately, these actions occurred too late for recovery. The Safety Board believes that the first officer’s actions did not directly lead to the accident, but his delayed assertiveness precluded an opportunity to avoid it.

Flagship 3

Photo (C) NTSB from Aviation Safety Network  

AMR Eagle Selection and Hiring Practices

AMR Eagle’s application process required prospective employees to complete employment history forms and to sign civil releases giving AMR Eagle permission to contact previous or present employers. Such an employment practice is not uncommon in the industry and is intended to check past job performance as to predict future performance. Contacting former employers has been shown one of the best methods for evaluating prospective employees. The accident had signed a release permitting his previous employer to respond to AMR Eagle’s inquiries, but a request was apparently not sent by AMR

By not following the intent of its own hiring  procedures that were established to gather information on an applicant’s background, AMR Eagle precluded the possibility that it could learn that the pilot possessed questionable aviation abilities. If Flagship had asked for, and Comair had provided the captain’s performance history while at their company, it is likely that the deficiencies in captain’s skills would have been specifically addressed prior to his being offered employment. This might have resulted in a decision not to hire him. But, even if AMR Eagle had decided to make an offer of employment, a complete employment history, in the possession of his immediate supervisor, should have made the subsequent complaints regarding his abilities far more meaningful.

Three times previously the Safety Board has recommended that air carriers be required to conduct substantive background checks of prospective airmen/employees before they are hired.” Each time the FAA has essentially rejected this recommendation and the Safety Board has classified all three “Closed–Unacceptable Action.”

The stated policy of the previous employer of the accident captain -the nondisclosure of employee performance information- illustrates the common perception that the release of such information (especially unfavorable information) may lead to civil liability, position that is typical within the industry.

The Safety Board acknowledges the concerns within the industry about potential legal actions and other issues regarding the retention and use (especially the provision to a third party) of records containing pilot performance evaluations. However, it should be recognized that a major portion of airline pilot mining records involves check rides given by designated pilot examiners. The designated examiners represent the FAA during such check rides, so the records of their work are technically FAA records. The Safety Board believes that many of the industry concerns about the provision of records to a third party can be alleviated by having the performance/training and checking records for airline pilots forwarded to the FAA, similar to the manner in which airman’s records are currently retained by FAA. This system would permit airlines to request pilots records directly from the FAA and would resolve the problems faced by airlines in providing previous employee records.

The Safety Board continues to believe that airlines and the traveling public would benefit from more availability of pertinent information on the quality of the previous performance of applicants for pilots positions. Therefore the Safety Board concludes that the FAA should require all airlines operating under 14 CFR parts 121 and 135 and independent facilities providing training to the airlines to provide to the FAA, for incorporation into a storage and retrieval system, pertinent standardized information on the quality of pilot performance in activities that assess pilots skills, abilities, knowledge, and judgment during training, check flights, initial operating experience, and line checks. In addition, the NTSB believes that FAA should require all airlines operating under 14 CFR part 121 and 135 to obtain records of this information from the FAA, for the purpose of evaluating applicants for pilots position during selection and hiring process.

AMR Eagle Training

Training Records

However, before the system discussed above can be effective, appropriate records on training and performance of pilots must be developed and maintained. For example, the computer-based records generated by the AMR Eagle training center, provided to Flagship Airlines, contained an annotation of the dates when specifically required activities were accomplished, but there were no amplifying comments regarding performance or strengths/weaknesses for reference by subsequent instructors, check airmen or others. Information concerning specific problems experienced, if any, was either not recorded, or was destroyed once training was completed. There was not even a record to indicate when extra training sessions were required. This is not only eliminated the ability to evaluate the individual’s performance, it also prevented management from evaluating the effectiveness of its syllabus. Further opportunity to evaluate both the training and the individual pilot was lost because AMR Eagle/Flagship did not require written comments during a pilot’s IOE or probationary year.

By contrast, the Flagship training records compiled during the captain’s training by Flagship personnel, prior to transfer of all training to AMR eagle in September 1993, reflected a cause of possible concern. The records not only documented the captain’s unsatisfactory progress, thy reflected the maneuvers involved (single engine non-precision approaches- March 1992 and crosswind takeoffs and landings, engine failures, and single engine missed approaches on April 1992) Although the records were not available at the RDU base, they could have been reviewed by BNA management for the RDU Base manager, or sent to RDU via company mail for his own examination.

The captain has demonstrated adequate skills in routine operations that may have masked his deficiencies in some checking and oversight situations. However, his line flying performance caused several line pilots to speak to the Base Manager about the accident captain. In fact, the captain had even approached the Base Manager to discuss the situation on his own initiative. Although the Base Manager addressed the issues raised by the individuals making the comments and offered the captain additional training/instructor time, there was no evidence that he attempted to review the captain’s records. If the Base Manager had reviewed the AMR Eagle computerized training records of the captain, he would not have found the annotation of the failed SD3-60 training periods (March 1992- April 1992). Also, he would not have found any record of the failed J-3201 upgrade type rating of October 6, 1992. However, these failures were documented in records available in the Flagship training records at Nashville and might have prompted additional discussion/action by management. Rather the relying on a report from a first officer, the events calling the deficient performance of the accident captain to the attention of his Base Manager should have prompted some form of records review, discussions with other company personnel, and possibly a line check or check airman assessment. In short, the lack of accessibility of and sufficient detail in the pilot records apparently prevented Flagship management from reviewing the captain’s performance history, even when complaints from others and self-initiated comments from him were received. Moreover, the deficiency in the AMR Eagle/flagship training record prevented Flagship management from ensuring that pilot problems were kind addressed in training and from adequately monitoring substandard pilot performance trends.

The NTSB, as a result of an accident investigation, have previously issued a safety recommendation to the FAA whom in response issued Flight Standard Information Bulletin 8FISB) 94-16A, January 22, 1995, directing POIs to review their assigned operator’s airman training recordkeeping procedures which weres considered ineffective for NTSB.

The accident involving 3379 demonstrates a continuing need for positive FAA action to enhance the quality of information that airlines retain on each pilot. Therefore, the Safety Board believes that the FAA should require all airlines operating under 14 CFR Parts 121 and 135 and independent facilities that train pilots for the airlines to maintain pertinent standardized information on the quality of plot performance activities that assess pilot skills, abilities, and line checks and to use this information in quality assurance of individual performance and of training program.

Engine-out training

Flagship line pilots currently receive all ground and flight training at the AMR eagle flight training center in DFW. Examination of the syllabus indicated that both ground school and simulator training addressed the auto-relight system and the ING light, the engine torque/NTS system, engine failure recognition, go-around procedures, and stall recognition/recovery. Various ground and flight instructors interviewed responded properly to questions about these subjects.

However, several line pilots, by contrast, gave varying responses regarding engine failure recognition. The confusion represented in the line pilots’ answers reflected unfavorably on the training effectiveness, and at least, in part, prompted Jetstream customer support to issue Notice to Operators that emphasized RPM as the single unequivocal indication of engine failure. It stated that low torque and low EGT are not necessarily indications of flameout or failure. If RPM is above 90 percent, then the engine is running. The availability of power should be assessed by advancing the power lever and checking whether the torque responds normally.

The captain apparently did not advance the power ever to test the operating condition of the left engine, and this was possibly reinforced by inappropriate simulator training on the combined NTS/engine failure. This simulator demonstration allowed the RPM to remain at about 60 percent on the failed engine. The fine pitch condition of the unfeathered propeller was feathered manually. This exercise alerted the pilot that the NTS had failed. It also established the misconception that any NTS condition and the associated IGN light, were connected with an engine failure. The actions of the captain and the answers of the line pilots interviewed indicated that they associated the illumination of the IGN light with and NTS/ flameout condition. The Safety Board considers this “negative training” situation because the training taught a concept that was incorrect and that could adversely affect pilot performance in a real emergency. Although the training scenario concludes with feathering the propeller, the captain did not follow this procedure in the accident flight.

Another indication of “negative training” is that during single engine missed approaches in the simulator, most pilots stated that they advance only one power lever. This may be a direct reflection of previous training in airplanes in which a zero-thrust condition (for safety reasons) had been established on one engine in the emergency scenario, and consequently, only one power lever was used by the pilot receiving the training. Apparently, this practice was perpetuated in the simulator training because the instructors did not enforce the company procedure, described in the Aircraft Operating Manual, to advance both power levers to maximum power.

The CRM training provided by AMR Eagle was thorough and consistent with current industry standards and practices. Both crew members had received this training. However, the captain failed to apply it to this perceived emergency situation. The first officer, by contrast, appears to have been at least attempting to assert himself in the various questions and suggestions he made, if not in actions he took or initiated. However, when corrective action was not commanded in memory items for engine failure and go-around procedures, he did not verbally advise the captain of the appropriate company procedures.

Company Maintenance

The investigation discloses several administrative errors involving maintenance records. Items included incorrect aircraft registration numbers and a location where work was performed, which are considered isolated incidents, and the improper blade angle value entered on work reference which was corrected.  Four days before the accident flight, the aircraft had passed a functional check flight (FCF), the result of the right propeller assembly having been replaced. After the aircraft returned to service it flew 24 flights prior to the accident. There were no comments on either asymmetric torque indications or directional control difficulties on landing.

Although the accident captain made comments about directional control problems experienced at GSO en route to RDU, there was no direct connection between those comments and the condition observed on the FCF. Accordingly, the Safety Board concludes that the torque split condition identified on the FCF was most likely an error in indication only.

AMR Eagle/Flagship Management Structure

The Safety Board examined the nature of the oversight of Flagship by AMR Eagle and the management of Flagship itself, to determine what role if any, the organizational structure may have in the accident. The evidence indicates that most, if not all, of the critical decisions governing the conduct of Flagship operations, were made at AMR Eagle headquarters by persons employed either directly or indirectly by AMR Eagle. These decisions addressed such areas as pilot selection, pilot training, route selection, flight scheduling, recordkeeping procedures, aircraft operating practices, payroll, profit and loss determinations and other key elements critical to managing the airline. Nevertheless, Flagship (like the other AMR Eagle carriers) operated under its own certificate in accordance with FAA requirements, therefore was responsible for assuring that flight operations were conducted safely and in compliance with FAA regulations.

The fact that the major decisions affecting Flagship operations were made by AMR Eagle personnel at DFW who were not directly involved in Flagship operations did not adversely affect safety or line operations at Flagship. For example, the ineffectiveness of Flagship management in it is oversight of the captain does not appear to have resulted from any action taken or decision made by AMR Eagle. The evidence suggests that the decisions and actions of the RDU base manager with regard to this captain were independent of MAR Eagle management. Consequently, the Safety Board does not believe that the organizational structure of Flagship and its relationship to AMR Eagle was a factor in this accident.

Flight Profile Advisory System

The AMR Eagle training was inadequate with respect to the FPA-80 system. Information required by 14 CFR 135.153 was not available in the airplane flight manual, and only marginal system information was included in the ground school. More importantly, the system as installed on the Flagship fleet, did not meet the requirements of 14 CFR 135.153.

The Safety Board does not believe that the absence of a GPWS or the improper installation of the FPA-80 system contributed to the cause of this accident. However, this situation raises questions about management of Flagship Airlines and the oversight of Flagship by the FAA.

Physiological Factors

Although the captain had taken sick leave for 3 days prior to the accident there were no indications suggesting that the captain was sick. The presence of a small amount or chlorpheniramine in the toxicological analysis indicated that he had taken some antihistamine in the recent past. This medication has the potential to reduce alertness, increase reaction time, and adversely affect perception, the variation in individual metabolic rates precluded the safety board from estimating either the time of ingestion or the effect, if any, it may have had on his performance.

The safety board remains concerned about the use and misuse of medications, both prescribed and over the counter, by pilots, air traffic controllers, dispatchers and others involved in aviation operations who may be unaware of the potential hazards many medications present. Moreover, many in the aviation community lack knowledge about these hazards and the fact that medications can remain hazardous following ingestion. With the number of medications that were available exclusively by prescription now being distributed over-the-counter, accompanied by extensive media marketing campaigns, the Safety Board believes that an already potentially hazardous situation may become worse.(Note from the blogger : It is even worse now. Love this paragraph. All those knowing my work during the las 24 years know this is an issue which I have been very insistent with, and it is a recurrent topic in my classes. I’ve taught this to all kind of flight crew from students pilots to major heavy airliners commanders.)

This accident, involving AMR Eagle flight 3379, suggests that the FAA programs to educate and inform those holding airmen certificates about the potential hazards of medication may not be fully effective. Additional effort may be needed to educate those in the aviation community on the need to avoid all but a handful of approved medications for several days before flying, controlling air traffic, or being involved in other critical aspects of the air transport system.

Flagship 4

Photo from PlaneCrashInfo. com http://www.planecrashinfo.com/w19941213.htm



  1. The flight crew was properly certificated in accordance with Federal Aviation Regulations and company procedures
  2. The airplane was certificated and maintained in accordance with existing regulations, except for the improper installation of the FPA-80 as a substitute for a GPWS
  3. Air traffic control services were properly performed.
  4. The Weather was not a factor in the accident.
  5. The captain associated the illumination of the left engine IGN light with an engine failure
  6. The left engine IGN light illuminates as a result os a momentary negative torque condition, when the propeller speed levers were advanced to 100 percent and the power lever were at flight idle.
  7. There was no evidence of an engine failure. The CVR sound spectrum analysis revealed that both propellers operated at approximately 100 percent RPM until impact, and examination of both engines revealed that they were operating under power at impact.
  8. The captain failed to follow established procedures for engine failure indication, single engine approach, single engine go-around, and stall recovery.
  9. The flight crew failed to manage resources adequately; specifically, the captain did not designate a pilot to ensure aircraft control, did not invite discussion of the situation, and did not brief his intended actions, and the first officer did not assert himself in a timely and effective manner and did not correct the captain’s erroneous statement about engine failure.
  10. Although the first officer did perform a supportive role to the captain, his delayed assertiveness precluded an opportunity to avoid the accident.
  11. Flight 3379 did not encounter any wake turbulence during the approach to runway 5l, or during the departure from controlled flight.
  12. AMR Eagle did not adequately address the recognition of engine failure at low power, the aerodynamic effects of asymmetric thrust from a “windmilling” propeller, and high thrust on the other engine.
  13. AMR Eagle provided “negative simulator training” to pilots by “associating the ING light with engine failure and by not instructing pilots to advance both power levers during single engine go-arounds as required by the operational manual.
  14. AMR Eagle and Flagship Airlines crew training records do not provide sufficient detail for management to track performance.
  15. Flagship airlines management was deficient in its knowledge of the types of crew records available and in the content and use of such records.
  16. Flagship Airlines did not obtain any training records on the accident captain from Comair. Further, Comair’s standard response for employment history would not, had it been obtained, have included meaningful information on training and flight proficiency, despite the availability of such data.
  17. The FAA did not provide adequate guidance for, or ensure the proper installation of the FPA-80 as a substitute for a GPWS on Flagship’s fleet.
  18. The structure of the FAA’s oversight of MAR Eagle did no provide for adequate interaction between POIs and AMR Eagle management personnel who initiated changes in flight operations by the individual Eagle carriers.

Probable Cause

The National Transportation Safety Board determines that the probable causes of this accident were: 1) the captain’s improper assumption that an engine had failed, and 2) the captain’s subsequent failure to follow approved procedures for failure engine, single-engine approach and go-around, and stall recovery.

Contributing to the cause of the accident was the failure of AMR Eagle/Flagship management to identify, document, monitor and remedy deficiencies in pilot performance and training.

Excerpted from National Transportation Accident Report, NTSB/AAR-9507DCA95MA006: Uncontrolled collision with terrain. Flagship Airlines, inc. doing business as American Eagle Flight 3379. BAe Jetstream 3201, N918AE. Morrisville, North Carolina December 13,1994

Additional source: Commuter Captain Fails to Follow Emergency Procedures After Suspected Engine Failure, Loses Control of the Aircraft During Instrument Approach (Flight Safety Foundation – Accident Prevention April 1996)

Further reading on this blog:

  1. Shutting down the wrong engine
  2. TransAsia Airways Flight GE235 accident Final Report
  3. Lessons learned from British Midland Flight 92, Boeing B-737-400, January 8, 1989


Posts in this blog:

Human Factors in Aviation

Accidents/Incidents investigation

Follow me on facebook Living Safely with Human Error and twitter@dralaurita. Lots of Human Factors information updated almost every day.

15 thoughts on “Learning from the past: American Eagle Flight 3379, uncontrolled collision with terrain. Morrisville, North Carolina December 13th, 1994

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s