Aviation Accident Report: American Airlines Flight 383/Analysis and Conclusions
2. ANALYSIS AND CONCLUSIONS
The investigation of this accident disclosed no evidence of any failure or malfunction of the aircraft, its system or components. All applicable aids to navigation, and runway/approach lighting aids were operating satisfactorily during the approach and at the time of the accident. The aircraft and crew were properly certificated and the flight was properly dispatched. All communications during the approach were considered to be normal and revealed nothing to suggest any aircraft or flight crew distress. The last radio transmission, five seconds before impact, indicated that the crew was unaware that they had descended to an altitude below the level of the airport.
After an exhaustive review and evaluation of the available evidence, the Board has become convinced that the cause of the accident is directly related to the manner in which the crew operated the flight and indirectly related to certain specific factors that may have influenced or affected the crew during the landing approach.
On of the factors which undoubtably influenced the conduct of this flight was the weather situation which existed in the Cincinnati area and the specific conditions which were encountered by N1996 during the approach. Based on USWB and witness observations during this time it was reliably ascertained that a line of rain showers and thunderstorms, oriented east-northeast/west-southwest, was moving into the area from the northwest at a speed of 25 knots. Cloud to cloud lightning was reported in the vicinity of the thunderstorms. The surface visibility at the airport was diminishing rapidly due to the precipitation associated with this line. Recorded values showed that between 1902 and 1903 (just after the time of the accident) visibility on runway 18 dropped from between 5-7 miles to 1-1/8 miles, then increased almost immediately to better than two miles. Cloud coverage in the area consisted of an overcast in the process of lowering from approximately 4,000 feet a.f.l. to 2,500 feet a.f.l. with broken clouds based near 1,500 feet a.f.l. A few fracto-stratus or "scud" clouds were located below these layers to the northeast of the field along the Ohio River with light precipitation occurring in this area. Based on surface visibility values it is believed that inflight visibility in the rain showers would have been two miles or less depending on the intensity of the precipitation. Inflight visibility to the east and northeast of the field, along the downwind landing leg, would have been in excess of seven miles.
No inflight turbulence was reported in the area nor was there any significant turbulence indicated on the flight recorder readout. The wind shear of .06 knots per hundred feet and the component headwind change of five knots per minute during the last minute of flight would have had a negligible effect on the flightpath of the aircraft and are not considered to have had any bearing on the cause of this accident.
The weather briefing received by the crew prior to departure from LaGuardia is considered to have been adequate and representative of the conditions existing in the area at the time of their arrival.
The recorder determined ground track was in agreement with the ATC and witness determined tracks and the recorder derived flight profile correlated in all essential elements with the ATC and witness data. A brief review of the flight profile shows that N1996 entered a downwind approach leg approximately four miles east of runway 18 at approximately 2,000 feet m.s.l. (1,110 a.f.l.) and that an approximate 800 feet per minute rate of descent was commenced during the turn onto the base leg. This descent was continued to an altitude of 210 feet a.f.l. at which point a gradual turn toward the final approach course was initiated. Coincident with this turn (20 seconds before impact) the descent rate increased to about 2,100 feet per minute for 10 seconds, during which time the aircraft descended into the valley formed by the Ohio River, below the elevation of the airport. The final 10 seconds of flight showed a decrease in descent rate to approximate 625 per minute until impact.
Analytical studies of the flight recorder data relating to the aircraft's approach configuration were carefully reviewed and evaluated. As was previously stated, only during the descent from 5,000 feet m.s.l. to 2,000 feet m.s.l. could the aircraft's configuration be positively identified; i.e., spoilers extended and engines at idle thrust. However, based on the recorder data and AA operational procedures the other probable configuration sequences were determined relative to the flight profile. This analysis showed that traffic pattern entry was made at an airspeed of approximately 210 knots; that the spoilers were retracted and two degrees of flaps were extended at that time; that airspeed diminished as the aircraft turned toward the base leg; that just prior to the completion of this turn, at an airspeed of 170 knots, flaps where lowered to fix degrees; that near the midpoint of the base leg at an airspeed of 160 knots flaps were extended to 15 degrees; and that the final flap extension to 25 degrees was made during the turn toward the final approach course approximately 20 seconds before impact.
Although the increase in the rate of descent to approximately 2,100 feet per minute appears to correlate with the extension of flaps from 15 degrees to 25 degrees the Board find it extremely difficult to suggest a proper rationale for the pilot permitting such high descent rates to develop at this point in the approach. It would appear, however, that the increased drag and nosedown pitch mode accompanying the 25-degree flap extension was not initially checked either through the use of longitudinal control forces or increased thrust and that the high rate of descent was inadvertently developed. However, airspeed bleed-off from 160 knots to 145 knots coupled with a decrease in the rate of descent to approximately 625 feet per minute indicates that the pilot subsequently utilized the longitudinal control system to arrest the high descent rate. Because of the relatively high airspeed at this point in the approach it would appear that this action was consistent with an attempt to further reduce airspeed. Had added thrust alone been used to decrease the high rate of descent it would not have achieved the apparent desired results of slowing the aircraft.
The traffic pattern airspeeds and aircraft configuration sequences utilized by N1996 were compared to appropriate airspeeds and recommended configuration sequences outlined in the VFR training pattern. From this study it was indicated that the various flap extensions during the approach conformed with the AA normal operating extension speeds. However, because of the comparatively high airspeeds maintained by the flight throughout the approach, these flap extensions were concentrated or "bunched-up" on the base leg.
Based on an evaluation of airspeed bleed-off and descent rates, as well as the metal impingement pattern found in the engine turbines, the Board concludes that, except for the brief level period at 2,000 feet m.s.l., the entire descent was probably conducted at or near idle thrust. In this case, the only logical explanation for conducting the approach at such low engine power settings would have been an attempt by the pilot to expedite airspeed reduction to the appropriate approach/flap extension speeds. However, it can be seen that if the downwind leg altitude had been maintained or if a considerably lesser descent rate had been used while extending flaps in conformance with normal flap extension speeds, airspeed bleed-off would have been more rapid and the appropriate flap extensions could have been achieved further back on the base leg. Thus, with the aircraft slowed down and approach flaps extended, utilization of higher more desirable thrust settings would have been possible earlier in the approach. As it was, a number of aircraft configuration changes and landing checklist items remained to be completed as N1996 was turned onto the final approach course.
The Board recognizes that the VFR landing approach profile can and will vary depending upon the specific operational situation encountered. Since a rather wide downwind leg was flown by N1996 (4-1/2 miles from the runway) as compared to the training pattern downwind leg of 1-1/2 miles, proper technique would have dictated either: (1) remaining at or near the downwind leg altitude (1,100 feet a.f.l.) until much further along the base leg before starting the final descent, or (2) utilizing a considerably lower descent rate throughout the base leg, or (3) leveling off at and maintaining the circling minimum altitude of 600 feet a.f.l. until the aircraft had been turned onto final and a visual landing was assured, or (4) some combination of the single alternatives that would have assured proper terrain clearance. Because none of the above was accomplished and a descent was continued below field elevation it can only be concluded that the crew, possibly distracted by or preoccupied with the exigencies associated with continuing the visual approach into deteriorating weather conditions, did not give proper or sufficient attention to the primary altitude reference instruments during the approach.
It is difficult to reconcile how two experienced captains could spend almost two minutes descending below 1,200 feet a.f.l., at night, under adverse weather conditions and not properly monitor altitude. Even if both pilots were primarily concerned with maintaining visual contact with the airport it would be logical to assume that an occasional cross-check of the flight instruments would be conducted by one or both pilots.
It is therefore not reasonable to believe that the failure to properly monitor the altimeters and continued unchecked descent can be attributed to one or two factors, but is more properly enveloped within a number of significant, complex, and highly inter-related conditions. The following are considered to be the most prominent, consequential areas and although discussed individually must be evaluated within the frame work of their total effect upon the crew's conduct of the approach:
Deterioration Visibility Conditions
Prior to commencing the turn onto the base leg better than VFR conditions existed along the aircraft's flightpath. After the turn onto the base leg the flight began to encounter light rain showers and scud clouds which rapidly reduced inflight visibility. In order to maintain VFR conditions and remain clear of these clouds it may have been necessary for the flight to initiate a descent from the downwind leg altitude. As the flight progressed along the base leg rain shower activity became increasingly heavy with inflight visibilities dropping to two miles or less. Towards the latter part of the base leg most probably the only airport lights visible to the crew would have been the sequenced "flashers" associated with the approach lighting system. In view of the rapidly deteriorating visibility as well as the rain being reported at the airport it is likely that the pilots were concerned with and perhaps principally occupied in maintaining visual contact with the airport.
Terrain Features
It is noted that the particular terrain (see Attachment #1) in the vicinity of the aircraft's flightpath may well have provide the crew with an illusion of proper altitude or adequate terrain clearance. The Ohio River basin directly to the left of the flightpath is approximately 400 feet lower than the terrain directly under or to the right of the aircraft's flightpath. Also the terrain south of the river in the direction of the airport is featured by a wooded unlighted hillside which rises steeply from the river to the approximate elevation of the airport. Except for the lights of residences located near the riverbank there are no lights which would provide terrain definition. At night under poor visibility conditions it is quite conceivable that the elevation of the lights in the river valley could be associated with the elevation of the terrain in the vicinity of the airport.
It should be remembered that at all times after turning onto the base leg, and throughout the descent, the airport remained well to the left side of the aircraft. In order to keep the airport in sight all observations by the pilots would have been to the left of the aircraft through the left side cockpit windshield or left side window. Then, the only lights visible to them would have been the ground lights along the riverbank which could have provided the pilots with a sensory illusion of adequate altitude over the immediate terrain.
It is realized that both pilots should have been familiar with all aspects of the terrain having flown into and out of the Greater Cincinnati Airport many times previously. However, at night in limited visibility conditions it is possible that the terrain, as viewed by the pilots, would not have triggered any immediate awareness of low altitude. Conversely, it is believed if the ground lights directly under or to the right of the aircraft's base leg flightpath had been observed an immediate low attitude situation would have become apparent to the pilots.
Misreading/Misinterpretation of Altimeter Presentations
Examination of the pilot's and copilot's altimeters revealed no evidence of an pre-impact failure or malfunction and both were set properly to indicate altitude above field elevation. The functional capabilities of the No. 3 altimeter could not be determined because of extensive impact damage to this instrument. However, based on the fact that the last ATC clearance altitude of 2,000 feet m.s.l. was maintained for approximately two minutes on the downwind leg prior to the initiation of the descent, it is believed that this altimeter was properly set and functioning normally throughout the approach.
All events leading up to impact, including the final radio transmissions from the flight, indicate that the crew was not concerned with and totally unaware of a dangerously low altitude situation. In examining the many potential ramifications involved in the aircraft's deviation from the desired altitude profile one obvious possibility is a misinterpretation or misreading of the information being presented on the altimeters. It is noted that the drum altimeter has been in operation for many years and is considered by all segments of the industry to be an accurate, highly reliable instrument. While the Board has no reason to believe the drum altimeter is deficient in any way whatsoever, it does believe that improper monitoring of the instrument could possibly result in misinterpretation, and this in turn could help to explain the inappropriate descent involved in this accident.
It can be seen that in reading this altimeter the pilot must first look at the number below the index on the "drum" to identify the thousand-foot level, and then to the radial pointer to determine the hundred feet indication. Two separate readings are required to assess altitude. At stabilized altitudes, or at low rates of climb/descent where the drum is moving slowly or not at all, the pilot must be certain to associate the proper thousand feet indication in relationship to the hundred feet reading. For example, an altimeter presentation of 900 feet would show the "1" (thousand feet) slightly above the thousand feet (drum) index and "zero" below the index. The radial pointer would be pointing to the nine on the outside dial. A misinterpretation could occur if a pilot mistakenly associates the radial pointer reading with the "1" slightly above the index rather than the "zero" below the index for the thousand feet indication. The result would be a reading of 1,900 feet rather than the 900 feet actually portrayed–or, and error of 1,000 feet. It was also noted that AA altimeters have a crosshatched band adjacent to the numbers from plus 1,000 feet to minus 1,500 feet as a means of emphasizing these lower altitude values. In a descending situation to below zero elevation the radial pointer, rotating counter-clockwise, does not point to the actual hundred of feet below zero. For instance, 100 feet below zero elevation would be portrayed with the radial pointer on the outside dial nine and with the zero on the drum slightly above the index. This requires the pilot to interpret the nine hundred feet indicated as actually meaning 100 feet below zero. Additionally, when reading negative values the number above the drum index rather than the number below the index gives the correct thousand feet determination. In other words the drum presentation and reading reverses at below zero altitudes.
It is realized that in these cases or in any altitude presentation a pilot knowledgeable in the altimeter should have no problem whatsoever in quickly determining the correct altitude reading. However, it can also be seen that under certain conditions compounded by infrequent, fragmented, or distracted monitoring, a misreading/misinterpretation of altitude could occur.
Cockpit Workload
Another factor which must be considered in completely evaluating the apparent improper monitoring of altitude is the cockpit workload and crew coordination involved in this approach. It was shown that higher than normal airspeeds were maintained by the aircraft throughout most of the approach and that flap extension sequencing was concentrated on the base leg. As late as the turn to the final approach course, 2-1/2 miles from the runway, only 25 degrees (approach) flaps had been extended. In accordance with recommended operating procedures, the completion of this turn to final would have required the aircraft to be in a landing configuration, with airspeed and rate of descent stabilized at or near appropriate values. However, it can be seen that even in the later phases of the turn the airspeed was still above 40-degree (landing) flap extension speed, the landing gear had not been extended, and the landing checklist obviously was not completed. Because of the number of essential landing items still incomplete this late in the approach there can be little doubt that the cockpit crew would have been extremely busy at this time and more than likely throughout most of the base leg.
Consideration must be given to the fact that these two qualified B-727 captains had flown together on seven previous flights, knew each other well, and most probably had established a high degree of reliance on one another's operational capabilities.
It is possible that the check captain, confident in the other pilot's ability to operate the aircraft safely, would assume that the altimeters and other flight instruments were being monitored and could therefore concentrate on maintaining visual contact with the airport. Moreover, it is possible that the captain being observed was secure in the knowledge that a well qualified check captain was in the right seat performing copilot duties. He could therefore in view of the rapidly decreasing visibility, concentrate on keeping the airport in sight, depending upon visual reference to ground lights for altitude guidance, with the assurance that the check captain was monitoring the flight instruments and would alert him to any unusual contingency.
American Airlines operational procedures require that the pilot not making the landing call out the airspeed, altitude, and rate of descent at 500 feet a.f.l., and also the rate of descent any time it exceeds 700 feet per minute below an altitude of 500 feet a.f.l. The flight recorder readout shows that at 1900:45, about 42 seconds before ground impact, the aircraft was descending through 500 feet a.f.l. near the midpoint of the base leg and that the rates of descent remained in excess of 700 feet per minute throughout the remainder of the approach. It is evident that either the AA altimeter monitoring procedures were not being followed or that a misreading/misinterpretation of the altimeter occurred which indicated higher values to the check captain (pilot not making the landing). It should be noted that if the check captain was concentrating on maintaining visual contact with the airport the limits of his line of vision would have been between 80 and 45 degrees left of dead ahead. It can be seen that with his heavy workload; extending flaps, performing the landing checklist, making all radio transmissions, and trying to keep the field in sight, he would have had little time to swing his gaze back to his own instrument for altitude reference. Aside from, but compounded with any other altimeter misread/misinterpret possibility, the probability of error is enhanced when reading the instrument from a side angle which in this case would have been approximately 55 degrees left of forward. It is doubtful that the center (No. 3) altimeter would have been used on the approach as it would not have been indication absolute altitude over the airport as were the Nos. 1 and 2 altimeters.
It is reasonable to assume that the flight engineer would have been engaged in duties attendant to completing the landing checklist and would not have had time to devote his full attention to the monitoring of flight instruments during the descent along the base leg.
With respect to crew coordination it appears significant that the entire flight was conducted in a manner that would expedite the arrival and approach into the Cincinnati Airport. A brief review of the operational phases of the flight shows that; (1) en route clearance changes were obtained which provided direct and shorter routings to Cincinnati, (2) the average ground speed within the terminal area (between 30 miles and 6 miles of the airport) was in excess of 325 knots, a departure from FAR 91.85(c)(1) which restricts terminal area speed to 250 knots IAS within 30 miles of destination below 10,000 feet m.s.l., and (3) despite the forecast for, and observation of approaching adverse weather, the flight elected to conduct a visual approach rather than utilize standard instrument approach procedures.
These phases of the flight do not, nor are they meant to, suggest any hazardous operational practices but they do indicate operational decisions pointed toward arriving at the Cincinnati Airport in the shortest possible time. It is strongly believed that the pilot being observed would normally be more inclined to maintain airway routings, more closely comply with terminal area speed restrictions, and conduct an instrument approach to the field, unless action to the contrary was being suggested or decided upon by the check captain. Further, decisions or suggestions of this nature would tend to indicate to the pilot being observed that all operational aspects of the flight were being closely monitored and would strengthen any possible over reliance situation of one pilot on the other which may have developed in the cockpit. Again, it is realized that this situation alone should not have precluded proper monitoring of aircraft altitude and can only be considered as a possible contributing factor to the apparent inattention to flight instruments.
It was noted that at 1901:14 in response to an inquiry from the tower if they still had the field in sight the flight replied, "just barely we'll pickup the ILS here." Although one of the control heads on the radio frequency selector control panel was found set at the approximate Cincinnati ILS frequency it could not be determined if the ILS had been tuned in throughout the approach or if it had been selected at the time of the above transmission. In either event, a full "fly-up" indication should have been displayed on the glide slope indicator at this point which if observed by the pilots could have alerted them to the low altitude situation or at least could have provided the stimulus for an altimeter cross-check.
It can not be determined to what extend the lightning flashes associated with the storm would have affected cockpit visibility or otherwise distracted the crew or, if the windshield rain repellent had been initially actuated in the light rain showers and had possibly deterred forward visibility. These and other indeterminable factors may or may not have presented additional distractions to the crew but must be considered in the final evaluation. It is clear, however, that a rapidly deteriorating visibility situation and increasingly heavy cockpit workload compounded with all or any combination of the other influencing conditions may have distracted from the full attentiveness of flight instruments and the proper monitoring of the aircraft's altitude.
Review of B-727 Design as Related to Accident Events
The Board would be remiss if it did not take cognizance of the concern that existed in the minds of many elements of the aviation community and traveling public regarding the three Boeing 727 fatal accidents that occurred within a short period of time. As indicated earlier in the factual portion of the report the B-727 flight characteristics were not a factor in the cause of this accident. Moreover, the pilot groups have indicated that in general the Boeing 727 is a reliable, versatile, and highly responsive aircraft and that no major difficulties are being experienced in actual line operations.
Yet, the Board's investigation did uncover certain aspects of operational practices in the Boeing 727 that warrant industry attention in the interest of preventing future accidents. It was noted in the NASA-VGH studies that close-in, high descent-rate, unstabilized approached are being conducted more often in the 727 than in any of the other jet transports studied. Why this is true is not evident from the preliminary review and any realistic evaluation will have to await a final NASA report on this matter. However, from an analytical viewpoint it can be seen that the principal application of this aircraft has been geared to a short/medium haul of "up-down" type operation as compared to the longer range (time-at-cruise) type operation of the older and larger jet transports. It may be that the very nature of this (short range) operation is engendering deviations from the hitherto standard jet operating principles. Also, consideration must be given to the fact that the 727 does have highly responsive and versatile flight characteristics and that these favorable characteristics may be misleading to the pilot, or are presenting the impression that greater liberties may be taken with the aircraft in normal operating situations, especially in the approach/landing regimes. It should be stressed that because of the high drag characteristics of the aircraft in the 40-degree flap configuration, high descent rates and airspeed reduction, in response to decreased thrust, can develop more rapidly than in the less sophisticated models and that this flight regime should be avoided close to the ground. It was noted that most of the 727 operators have included in their flight manuals information concerning the avoidance of high descent rates during the landing approach and also, that the FAA has instituted a requirement for the demonstration of a low altitude high descent rate maneuver as part of the air carrier training programs.
The Board does not know which of the above described factors ("up-down" type operation, sophisticated-responsive aircraft) is causing the results indicated in the NASA-VGH studies, or if some combination of these factors is involved. However, the Board believes it is of sufficient importance to warrant a thorough review by industry to resolve this matter.
The entire investigation and in particular the specific examination of the Boeing 727 flight performance and characteristics revealed no design deficiencies or unsatisfactory operating characteristics. Furthermore, the latter review of the aircraft design aspects uncovered no factors that would have had a significant effect on the events leading up to the accident.
It is concluded that the 20-minute delay incurred by the flight in departing LaGuardia, combined with the forecast and inflight observations of thunderstorm activity in the Cincinnati area, prompted the crew to conduct the flight in a manner so as to greatly expedite their arrival at the Cincinnati Airport. Once in the Cincinnati area there can be no doubt that the crew was aware of the rapidly deteriorating weather situation both through visual observations and radio communications with the tower as they proceeded on the downwind and base legs. In view of the total weather picture known to the pilots and despite the fact that VFR conditions existed at the airport, it is believed that a more prudent judgment would have been either for the flight to have conducted an instrument approach, or to have delayed the approach until the storm had moved beyond the airport.
It is further concluded that after the flight turned onto the base leg inflight visibility was sharply reduced and that both pilots became preoccupied in maintaining visual contact with the runway, resulting in inattention to, and improper monitoring of, the altitude reference instruments.
The last point at which this accident could have been averted was approximately 13 seconds prior to impact. At this time the aircraft was descending below the level of the airport and any visual contact with the field or approach lights would have been lost immediately. The flight then should have, and could have accomplished a pull-up and go-around within the operating capabilities of the aircraft. However, at that point, as previously stated in the analysis, the accumulation of many factors either delayed or precluded proper recognition of the situation.
Regardless of the possible mitigating circumstances there can be no valid excuse for such a gross deviation from safe operating altitude as was presented in this accident. The approach procedures outlined in the carrier's manuals, the number of cockpit altitude reference instruments and related monitoring procedures, the approach aids available, and the skills expected of airline pilot, are all protective devices design to prevent the development of such a situation. However, it is apparent that these devices were not fully utilized in the conduct of this approach.
While the circumstances and conditions of the accident are greatly different from those involved in another Boeing 727 accident which occurred at Salt Lake City, Utah, and for which the Board has already issued a report, there are some elements in common relative to pilot/crew judgment. For this reason the Board must re-emphasize that the responsibility and authority committed to an airline captain requires the exercise of sound judgment and strict adherence to prescribed practices and procedures. Any deviation can only result in a compromise of aviation safety. Airline management, too, has a heavy responsibility for devising, developing, and implementing methods and procedures designed to insure that all of their pilot personnel constantly exercise a conservative, prudent, approach to their daily work.
The Board determines that the probable cause of this accident was the failure of the crew to properly monitor the altimeters during a visual approach into deteriorating visibility conditions.