University of Bielefeld -  Faculty of technology
Networks and distributed Systems
Research group of Prof. Peter B. Ladkin, Ph.D.
Back to Abstracts of References and Incidents Back to Root




on the draft report

concerning the accident of

China Airlines A300B4-622R, B1816

at Nagoya airport (Japan)

on April 26,1994


The Bureau Enqetes Accidents (France) has appreciated the opportunity that has been given to its representatives to participate in all important phases of the investigation and the cooperative spirit within the commission. The BEA thanks the commission for giving its representatives the opportunity to study and comment the draft report.

The BEA notes that the draft report reflects to a good extent the work performed. However, it points out that some important aspects of the accident are inadequately covered ; this may be misleading for its readers.

Indeed, in the analysis, some established facts either have not been totally taken into account or have not been mentioned in the draft report, and some hypotheses and the related factual elements have not been fully analyzed. Besides, two major statements are not in accordance with the established facts. These shortcomings are found in the conclusion and recommendations, some of them being inappropriate.

The following paragraphs substantiate each of these comments and, accordingly, propose the necessary amendments to the draft report. Indeed, failure to properly highlight all the relevant elements of the accident might come as a disadvantage for accident prevention which is and must remain the sole objective of Japanese AMC and French BEA according to International Civil Aviation Organization Annex 13

(Note: these comments are based on the English version of the draft report as provided by the AAIC).




I-1) Introduction

The report contains several hypotheses. The objective of analysis is to examine the available facts, make all the hypotheses supported by factual elements, and then conclude when facts allow it.

However on three subjects, hypotheses and associated factual elements are not fully analyzed:

I-2) Omitted hypothesis

Concerning the two quoted sentences of the Cockpit Voice Recorder (CVR) "Sir I cannot push it down"(11h14mn51s), and "how come it is like this"(11h15mn11s) the report hypotheses that it refers to the important pitch angle.

This hypothesis is acceptable, but it is much more probable that it corresponds to the very strong and unusual effort on the control column that the copilot and the captain were feeling when they spoke theses words.

The report should thus take into account this very probable interpretation.

I-3) Unjustified elimination of an hypothesis

A part of the analysis deals with the explanation of a phrase from the CVR 11h14mn12s" : Captain : "disengage it".

This presentation is misleading for the reader and leads to conclusions which are not at all supported by facts. Therefore the report must be corrected in this respect.

I-4) Lack of conclusion on hypotheses

Paragraph (4) entitled "Concerning use of AP" submits three hypotheses to explain the engagement of the auto pilot at 11h14mn18s.

The analysis of these hypotheses is based on a single word from the captain at 11h14mnl6s:

"That...", and leads to the conclusion that it is not possible to determine who engaged the auto pilot.

However, there are other less ambiguous sentences in the CVR which are not presently analyzed in the draft report. These sentences support the conclusion that the most probable hypothesis is that the F/Oengaged the auto pilot himself without any instruction from and without advising the captain, who was probably not aware of it.

Indeed, from ca. 10h49mn00s to 1056:00 the captain encouraged the copilot to fly manually several times ; then at 11h14mn20s, 2 seconds after the engagement of the auto pilot , the Captain said: "You watch, watch outside, outside." and at 11h14mn23s: "Push down, push it down. Yeah". These instructions clearly indicate that the Captain thought the F/O was flying manually, and thus, that the captain was not aware that the auto pilot was engaged.


Concerning two points of the report, the available factual elements are neither accurately described nor analyzed.



II-1) Possible recovery actions

Although the report mentions various possible actions allowing which would enable rapid recovery of a normal pitch attitude (paragraph (14)5), it does not indicate that the first officer and then the captain, when he himself was at the control, had time to detect (owing to the very strong and unusual effort on the control column) the pitch up tendency and also to take the appropriate corrective actions.

It would have been useful to state precisely in the report the basic piloting techniques for transport aircraft, corresponding to the following cases:

a)automatic flight system does not behave as expected by the crew;

b)out of trim situation.

The following should then be highlighted in the report:

>a)Automatic Flight System does not behave as expected by the crew

On any aircraft equipped with automatic Systems; when an auto pilot malfunction is suspected by the crew (or when the aircraft behavior -controlled by the Automatic Flight System- is not what the crew expected), the crew must disconnect the automatic Systems and continue the flight manually until they have carried out the necessary checks.

This is part of the basic knowledge of each pilot This is also repeated in the A300-

600 Flight Crew Operating Manual (FCOM) section 2.02.03 p 1 revision 15 in the

Caution on auto pilot override dangers.

b)Out of Trim situation

The pitch trim primary function in every aircraft is to cancel the efforts on the control column so that the pilot does not have to apply a continuous effort on it. As a consequence, each time a pilot flies manually and moves the control column in pitch he instinctively cancels the effort by an action on the pitch trim.

This is basic knowledge learnt during the first hours of flight training.

The pitch trim can be activated either electrically by a button located on one horn of each control column, or manually with the trim wheels located on each side of the central pedestal.

It should also be noted that, apart from the Visual indicator of pitch trim position, the effort on the control column and the full forward control column position (leading to a stretched forward position of the pilot arms) are clear indications of an out of trim situation which are common to all types of aircraft.



The A300-600 FCOM recommends in the chapter "Abnormal and emergency procedures", as an immediate action in case of 'Abnormal Pitch Behavior":







The action on the trim wheel (manual trim) disconnects the pitch trim levers and as a consequence the auto pilot. It thus cancels the cause of the trim movement and at the same time corrects the consequence (out of trim). This corrective action does not necessitate any preceding analysis.

It can be noted that this solution was successfully used by the crew during the 1989 incident

11-2) Action of authorities after previous incidents

The draft report describes three incidents on A300-600 and A310 which have some similarities with the Nagoya accident (important pitch up with an out of trim situation due to an auto pilot override). They took place, respectively, in 1985, 1989 and 1991. The draft report carries judgments on the actions the DGAC undertook in order to improve safety after these incidents. The statements made in this context require the following commentaries.

It is inappropriate to deal with the actions of the DGAC in isolation after these three incidents, as the decision making process resulted from multiple factors, and from the actions of several organizations (for specific action taken by the manufacturer, please refer to paragraph III-1):

1.For the first (1985) incident no official organization participated in the internal investigation, and the available information was not fully disclosed.

2.The second (1989) and the third (1991) incidents (separated by two years) were investigated by two different accident investigation authorities Finland and Germany).

In both cases the reports concluded that the causes were operational.

Moreover, for the 1989 event, the report indicated, among the findings, that the Captain was ill.

For the 1991 event, there was a finding concerning crew coordination and cockpit resource management. It stated that "flying and managing the aircraft with a 2-man crew "glass cockpit" had put them (the crew) under excessive pressure". It should be noted that both pilots had all their flight experience (before A310) on Ilyushin 18's, with at least 4 crew members.



The report on the 1989 event recommended improving the crew training program and the information on the dangers of overriding the auto pilot. There was no recommendation asking for a modification of the Auto Flight System.

The report of the 1991 event included no recommendation.

DGAC also conducted its own analysis of the events and agreed with the conclusions of the investigating authorities.

After the 1989 incident, in accordance with the recommendations1 the DGAC and the manufacturer amended the training program and Airbus Industrie revised the FCOM.

At the date of the third incident which occurred on an A310, the FCOM revision was implemented on A300-600, but was in progress on the A310 (the revision was carried out one week later).

In line with the investigators, the airworthiness authority and the manufacturer considered that;

Consequently, the modification proposed in Service Bulletin SB 22-6021 was not made mandatory.

The 1989 incident was presented during the 1990 annual seminar of the International Society of Air Safety Investigators and the 1991 incident was described in the January 1992 magazine of the Flight Safety Foundation. To our knowledge, no question was raised and nobody disputed the analyses and conclusions of the investigators.

The corrective operational measures had been defined after the 1989 incident and were fully implemented one week after the 1991 incident. Therefore it is not correct to suggest that the airworthiness authority and the manufacturer did not react to these incidents and did not take positive measures to cope with the situation.




Ill-1) Information to crew on auto pilot override

The report states that the crew's reaction was inadequate partially because their information on auto pilot override and its dangers was not insufficient. The BEA disagrees with this statement which is the result of the following shortcomings in the report:

- the information provided to the pilot is not fully described;

- moreover, the information produced is spread over several sections including the appendices. In order to provide a better overview, all the available information should be stated very precisely within one paragraph. This includes the information given in the training and presented in the manuals and the specific information sent to the airlines.

1.During initial training in simulator session n° 1 (Aeroformation qualification program) there is a demonstration of the consequences of auto pilot override in Go Around mode. The copilot performed this exercise during his A300-600 qualification training in 1992.

2.The manufacturer provided the airlines with the following elements dealing with auto pilot overriding:

The FCOM includes information on Systems design and procedures to be applied in each case.

After the March 198S incident, Airbus Industrie issued in June 198S an Operation Engineering Bulletin (OEB n° 29/1) on the consequences of an auto pilot override and in March 1988 proposed the Modification 7187.

After the June 1989 incident, Airbus Industrie sent an Operator Information Telex (OIT n° ST/999. 037/89) reminding crew how to use the Auto Pilot.

During the 6th Airbus Industrie Operational Conference that took place in Cairo in May 1990, this subject was also treated Two fully competent representatives of China Airlines attended this conference.

-In January 1991, the FCOMs were amended to include a Caution about auto pilot overriding in section 1.03.64 page 3/4 and section 2.02.03 page 1 "CAUTION saying:


Working on the pitch axis against the auto pilot in CMD may lead to hazardous situation in LAND and GO AROUND mode.

So if abnormal flight control behavior is encountered during these flight phases:

- check AP status (FMA, FCU),

- if AP engaged, disconnect it and take over.

- After the February 1991 incident, Airbus Industrie issued two OIT (n° ST/ 999. 0036/91 et n° ST 999.0048/91) reminding crew how to use the auto pilot.



- The FCOM bulletin 05/1 (subject 10) dedicated to auto pilot overriding was issued in June 1991.

- Airbus Industrie issued on 24 June 1993 Service Bulletin 22/6021 which includes a modification to the Flight Control Computer (FCC) "to provide auto pilot disengagement by applying a 15 daN force on the control column in Go Around mode above 400 ft Radio Height".

III-2) Statement made concerning the aircraft Auto Flight System

In several parts of the draft report, the Auto Flight System is qualified as "complicated" without any factual or analytical argument to support this statement.

For example in paragraph (3), four arguments are provided to support the conclusion that ((the training required to understand the sophisticated and complicated Auto Flight System was insufficient.

The first two arguments are

"1 the description in FCOM for the Auto Flight System are not easy for crew to understand

2 the crew was not given sufficient technical information with regard to similar incidents"

However, we have shown in the previous paragraph that the FCOM and the documentation provided are perfectly clear concerning auto pilot override in Land and Go Around mode.

The third argument is

"3 up to date materials were not properly obtained"

As the copilot really performed the exercise of Auto pilot override in Go Around mode during his qualification, this argument is not relevant to this accident.

The fourth argument is

"4 CVR transcripts show that crew understanding of the Auto Flight System was probably not sufficient"

There is no precise quotation on the CVR to support this allegation.

Defining the Auto Flight System as "complicated" has, in no way, been justified.

In conclusion, the BEA asks for the deletion of this statement.


IV-1) Comments about the recommendations directed to China Airlines

The provisions contained in the recommendations addressed to China Airlines, concerning in particular the general aspects linked to experience gaining and skill maintenance of crew, are probably also applicable to other airlines. The report could thus suggest that all airlines study these recommendations, determine those which are applicable to them and verify that those particular ones are indeed implemented within their company.



IV-2)Recommendation n° 2-(1)1 and 2-(1)2 on "Improvement of Auto Flight System functions on A300-600R"

The first recommendation proposes an automatic disengagement of the auto pilot under effort on the control column.

The BEA concurs with that recommendation. It is consistent with the one issued by the BEA in June 1994 after consultation with the AAIC, which was implemented through French Airworthiness Directive CN 94-185-165(13). This Airworthiness Directive is being supplemented by another one for radio-height lower than 400 feet.

The first part of the second recommendation covers the same idea, but in more general terms and is therefore not useful.

IV-3)Recommendation n° 2-(1) 3 on "Improvement of warning and recognition functions for Trimmable Horizontal Stabilizer movement"

When auto pilot is engaged, it is no longer possible to be in an out of trim situation after the implementation of Airworthiness Directive CN 94-185-165(13).

In manual flight the Trimmable Horizontal Stabilizer movement is highlighted by an aural indication (whooler) in addition to the trim wheel motion. The airworthiness authority and the manufacturer have indicated that this is an adequate indication of the Trimmable Horizontal Stabilizer movement and that it is not useful to add a supplementary warning to the existing indications (refer to paragraph 111-3 b).

IV-4)Recommendation 2-(3) on "Positive dissemination of technical information to operators"

This recommendation is not supported by the facts. Indeed after each significant "in service event", Airbus Industrie, just like all the other aircraft manufacturers, provides its operators with the relevant information. So, after the Nagoya accident, Airbus Industrie provided its operators with factual information on the circumstances of the accident as well as with proposals for reminders on the possible consequences of auto pilot overriding.

It must be noted that France considers that positive dissemination of technical information on the circumstances of any accident/incident is most important for Safety. This is why the French Accredited Representative requested on May 33 1994 that the authorities and operators be informed of the circumstances of the accident and of safety reminders, either by the investigation commission or by the State of Design. The investigation commission did not agree with this request.

lV-5) Recommendation 3 on Auto Flight System

The assumption that the Auto Flight System of the aircraft is complicated is not supported by any of the facts established during the investigation.



The commentary in recommendation number 3, on human behavior in emergency situations is of course fully valid. This has been known for a long time and is taken into consideration in the design of all aircraft and in basic training of at! pilots in the following way

This corresponds to the fundamental basis that design cannot be dissociated from training and procedures.

The BEA proposes the following recommendation, valid for all types of aircraft and operators:

highlight to pilots in the FCOMs and during their training (both initial and recurrent) the importance or disconnecting automatic Systems (auto pilot and auto throttle ) in case or lack of understanding or doubt concerning their operation.

IV-6) Proposal of supplementary recommendation:

Annex 8 to the Convention on International Civil Aviation states that (paragraph 4.2.4) the State of Registry shall transmit to the State of Design all mandatory continuing airworthiness information.

This notion is restrictive because it specifies that the State of Design is informed only when a State has taken a mandatory measure.

It should be widened to include all cases where a State judges that information concerning an "in service event "is useful for safety improvement.

Therefore, France proposes the following recommendation:

"International Civil Aviation Organization study an amendment to Annex 8 to request that a State forwards to the State of Design any information in its possession, which it considers to be useful to maintain or improve flight Safety.



Peter B. Ladkin, 1999-02-08
Last modification on 1999-06-15
by Michael Blume