AIRCRAFT ACCIDENT INVESTIGATION REPORT 96-5

China Airlines
Airbus Industrie A300B4-622R, B1816
Nagoya Airport
April 26, 1994

Aircraft Accident Investigation Commission
Ministry of Transport
[Japan]

Prepared for the WWW by

Hiroshi Sogame
Safety Promotion Comt.
All Nippon Airways

Peter Ladkin
Universität Bielefeld
Germany

Section 3

July 19, 1996


3. ANALYSIS

3.1 Analysis

3.1.1 Establishments of the times recorded on CVR and DFDR

3.1.1.1 CVR Recordings

The times not recorded on the CVR were determined as follows:
The times were established by utilizing the fact that ATC communications recorded on the CVR were correlated with ATC communications recorded on the ATC recorders installed at Tokyo Air Traffic Control Center and Nagoya Airport Office (which follow JST time).
Taking into account possible errors in the correlation process, it is recognized that the CVR stopped recording at approximately 1115:45.

3.1.1.2 DFDR Recordings

Various data are recorded continuously in digital PCM signals onto magnetic tape in the DFDR at various with its sampling rates ranging from eight (8) times per second to every 4 seconds.

Data regarding altitude, speed, and bearing are recorded every second. Data regarding time and engine are recorded every four (4) seconds. Data such as acceleration are recorded at a higher sampling rate of four (4) to eight (8) times per second. All the data are incremented every 4 seconds as one frame.

The cockpit clock times on the panel on F/O's side are also recorded on the DFDR. However, the times are not necessarily synchronized with UTC, so the DFDR times were calibrated as follows: radio communications keying data were recorded on DFDR. By correlating these keying data to the times contained on the CVR transcript for the ATC radio communications, the times recorded on DFDR were established.

Data up until 1115:43 were recorded on the DFDR. Since the DFDR manipulates data in the method that one second of data are temporarily accumulated in buffer (as a subframe) and are recorded on magnetic tape within next 0.5 second, by taking into account possible errors in the correlation process, it is recognized that the DFDR stopped operating at approximately 1115:45.

3.1.2 Flight of the Aircraft

3.1.2.1 Estimation of Flight History

Based on data recorded on the DFDR and CVR, the flight history was estimated to be as follows (see attached Figures 1, 2, 22, 27, 28, and 29):

3-1


0853 (1700 JST) The aircraft took off from Taipei International Airport.
ca.0914 (1800 JST) The aircraft reached FL 330, and flew in accordance with its flight plan toward Nagoya Airport.
1040:17 (1900 JST) The aircraft was cleared by Tokyo Control to PROCEED DIRECT XMC (Kowa VOR/TAC), and the aircraft flew according to this clearance.
1045:45 (1900 JST) The F/O(PF) briefed the CAP (PNF) on the approach procedure to Nagoya Airport, go-around procedure, etc.
1047:35 While cruising at FL 330, the aircraft was cleared by Tokyo Control to descend to FL 210, and began its descent.
ca.1049:00-ca.1056,00 During this period, the CAP (PNF) gave general guidance to the F/O(PF) on flight procedure and control during approach and landing.
1058:18 The aircraft's radio communication was transferred from Tokyo Control to Nagoya Approach, and the aircraft continued its approach.
1059:04 The F/O(PF) said "CHECKLIST". It is considered that he requested the CAP (PNF) to conduct the approach checklist.
1 l00:05(2OOOJST) The CAP (PNF) said to the F/O: "OK, FASTEN LEFT, APPROACH CHECKLIST COMPLETED."
1100:11 The CAP (PNF) instructed the F/O (PF) to control the aircraft at his own discretion and the F/O (PF) responded by saying "YES".
1100:12 The SLATS/FLAPS lever was moved from 0/0 to 15/0.
1102:35 The SLATS/FLAPS lever was moved from 15/0 to 15/15.
1107:22 Until this time AP No.2 had been engaged; at this time AP No.1 was additionally engaged.
1108:26 - 1110:54 Since the F/O(PF) was concerned about wake turbulence, the CAP (PNF) taught him how to deal with it, instructing him to reduce the speed from 180 kt to 170 kt in order to extend the separation between themselves and the aircraft flying ahead.
1110:52 The sound of a seat being adjusted was recorded. This is estimated from the sound spectrum of the CVR recordings which indicate that the CAP adjusted his seat upward in preparation for approach.

3-2


1111:35 With the CAP's (PNF) consent, the F/O(PF) disengaged both AP No.1 and No.2 in order to change from automatic operation to manual operation.
1111:45 The CAP (')NF) called out "GLIDE SLOPE ALIVE".
1111:46 The F/O(PF) acknowledged this and called out "go-around altitude 3,000 ft". And then GO AROUND ALTITUDE was set on ALT SEL before the FMA displayed GS STAR, as a result of which the altitude alert sounded at 1111:55. Both the CAP (PNF) and the F/O(PF) confirmed the above situation.
1112:19 The aircraft passed over the outer marker under manual control by the F/O(PF), and continued ILS approach.
1112:41 The F/O(PF) requested the CAP (PNF) to set "Flap 20", and in response to this the CAP (PNF) moved the SLATS/FLAPS lever from 15/15 to 15/20.
1112:56 The F/O(PF) requested "Gear Down" to the CAP(PNF), and the CAP responded by performing the Gear Down operation.
1113:14 The F/O(PF) called to the CAP (PNF) "30/40, speed V approach 140, landing checklist please"; the CAP moved the SLATS/FLAPS lever from 15/20 to 30/40 and called "Landing check list completed" at 13:27.
1114:05 At approximately 1,070 ft pressure altitude, the F/O(PF) triggered the GO lever. As a result, the engines' thrust began to increase, the aircraft developed a slight nose-up tendency and began to deviate above the ILS glide path. Speed also increased. Engine thrust increase was stopped at EPR 1.21 about 14:08, and was then slightly reduced to EPR between 1.17 and 1.18.
In an attempt to recover the normal descent path, the F/O (PF) performed a nose-down operation by pushing the control wheel (The THS position did not change from -5.3°.)
However, the aircraft did not descend and, around 1114:10, leveled off at approximately 1,040 feet pressure altitude.
1114:09 An aural LANDING CAPABILITY CHANGE WARNING was recorded on the CVR. This sound is considered to have been caused by the change from LAND mode to GO AROUND mode.
1114:10 The CAP (PNF) cautioned the F/O (PF) by saying " You, You triggered the GO lever,", and the F/O acknowledged, saying "Yes, Yes, Yes, I touched a little."
1114:12 The CAP (PNF) instructed the F/O (PF) to "disengage it", and the F/O (PF) answered "AY".
1114:16 The CAP (PNF) said " That" and the F/O (PF) said "AY"
1114:18 During level flight, both AP No.2 and No.1 were engaged in CMD almost simultaneously. As the FD was in GO AROUND mode, the APs were also engaged in GO AROUND. At this time, the elevators were at 3.5° nose-down with the F/O still pushing the control wheel.
1114:20 As the APs were engaged, the THS began to move from -5.3° toward the nose-up direction.
In the meantime, the CVR had recorded a sound that is assumed to indicate activation of the pitch trim control switch.
It is considered that the switch was activated by the F/O (PF) in an attempt to reduce the resistive force on the control wheel. However, trimming of the THS using the pitch trim control switch is inhibited during engagement of the AP(s), so the F/O's actions had no effect.
1114:23 The CAP (PNF) gave the F/O (PF) an order, saying "Push down, push it down, yeah". This is considered to have been an instruction to push the control wheel down in order to correct the descent path that had become too high.
1114:26 The CAP (PNF) told the F/O (PF) "You, that --- disengage that throttle". This is considered to have represented the CAP's (PNF) instruction to the F/O (PF) to manually adjust the thrust by moving the throttle toward its idle position (for the same purpose as in 1114:23, i.e., to correct the descent path that had become too high).
1114:29 The F/O said "Too high". This is considered to have meant that the aircraft was flying above the normal descent path.
1114:30 On noticing that the FMA was still displaying GO AROUND mode, the CAP (PNF) said to the F/O (PF), "You, you are using the GO AROUND mode", and then added "It's OK, disengage again slowly, with your hand on".
There seems to be a possibility that in response to the CAP's instruction, the F/O took some action to change from GO AROUND mode to another mode, but this was not achieved. The words "with your hand on" seem to have two different meanings, the first being that the F/O should keep his hand on the thrust lever and the second that he should keep his hand on the button to change from GO AROUND mode to another mode.
1114:34 The CVR recorded a sound that is assumed to indicate activation of the pitch trim control switch. As at 1114:20, however, this operation had no effect.
1114:37 The THS moved to -12.30°.
1114:39 The CVR recorded a sound that is assumed to indicate activation of the pitch trim control switch. As at 14:20 and 1114:34, however, this operation had no effect.

3-4


1114:45 The CAP (PNF) again pointed out to the F/O (PF) "It's now in GO AROUND mode". The F/O answered, "Yes, sir".
Although there seems to be a possibility that the F/O (PF) took some action to change from GO AROUND mode to another mode, no mode change was actually made. At and around this point of time, the pitch angle and AOA increased and the speed decreased, and to deal with this situation, the F/O increased the thrust slightly.
1114:49 The F/O (PF) said, "Sir, auto pilot disengaged" and disengaged both APs. This action was probably taken at the F/O's (PF) own discretion or under the CAP's (PNF) instruction, but it is not clear which was the case (the conversation in the CVR record just prior to the action had been masked by another ATC communication).
1114:50 The sound of Auto Pilot disengagement was recorded on CVR.
1114:51 The F/O (PF) said, "Sir, I still cannot push it down."
1114:57 With the pitch angle and AOA still increasing, the aircraft continued to its approach with decreasing speed. At approximately 570 feet pressure altitude, the thrust increased suddenly, reaching its maximum level at 1115:03.
This is considered to have been caused by activation of the alpha floor function due to the AOA exceeding the threshold angle of 11.50 for SLATS/FLAPS 30/40 configuration.
1114:58 The CAP (PNF) said "I, that land mode?".
1115:02 The F/O (PF) reported to the CAP (PNF): "Sir, throttle latched again Activation of the alpha floor function displays a symbol "THR-L" on the FMA.
Owing to the thrust increase following activation of the alpha floor function at 1114:57, the aircraft's speed and pitch angle increased; the aircraft stopped descending and began to climb.
At 1115:03, the CAP told the F/O that he would take over the controls. After doing so, the CAP pushed the control wheel to the forward limit, but the aircraft still continued to climb. Around this time the thrust levers were also temporarily retarded, suggesting that the CAP still intended to continue approach.
1115:04 The F/O (PNF) said, "Disengage, disengage." Again at 1115:09, the F/O said "Disengage, dis...". This is interpreted as a request to the CAP (PF) for the A/THR to be disengaged.
1115:08 The CAP (PF) said, "What's the matter with this?". It is considered that the CAP's words expressed his puzzlement that the nose-up tendency was continuing, even though he had pushed the control wheel fully forward and decreased thrust.
1115:11 The CAP (PF) re-increased thrust (which he had earlier reduced) while calling "GO

3-5


lever". At the same time, the CVR recorded the activation sound of the pitch trim control switch, and the DFDR recorded the movement of the THS in the nose-down direction.
The CAP (PF) said "Damn it, how comes like this?". It is considered that the CAP's words expressed his puzzlement that the aircraft pitch angle was still increasing despite his actions to the contrary (pushing the control wheel fully forward and retarding the thrust levers).
Owing to the re-increased thrust, the aircraft began a steep climb with increasing pitch angle. Speed, which had earlier increased, began to decrease.
1115:14 The F/O (PNF) reported go-around to Nagoya Tower.
1115:17 Mode 5 warning of the GPWS sounded "Glide Slope" once. It is estimated that this resulted from a detection of a pseudo-path angle that occurred at an angle 3 times greater than the normal path angle.
1115:18 The sound indicating passage of the SLATS/FLAPS lever through the baulk attached gates was recorded twice (see attached Figure 27).
According to normal go-around procedure, the SLATS/FLAPS lever should be moved from the 30/40 position one step higher to 15/20. However, judging from the numbers of times the STATS/FLAPS lever sound was recorded, it may have been moved beyond the 15/20 position, perhaps to the even higher 15/0, or 0/0 position. Later, at 15:27, a sound presumably indicating the SLATS/FLAPS lever's downward movement passing through the baulk attached gate was recorded on the CVR. Also on the DFDR is a record showing that the SLATS/FLAPS lever was set on the 15/15 position at 1115:28.
1115:20 Both thrust levers were retarded almost simultaneously. At approximately 1115:23, the No.1 thrust lever was retarded to the vicinity of its idle position and the No.2 thrust lever was retarded slightly. At approximately 1115:27, both the levers were back almost to their full thrust positions.
1115:21 "He, if this goes on, it will stall," the CAP (PF) shouted. It is presumed that this remark reflected the CAP's (PF) shock either when he found the aircraft was continuing to climb steeply with increasing pitch angle while reducing speed, or when he noticed the position of the SLATS/FLAPS lever set by the F/O (PNF).
1115:23 The master caution (single chime) sounded. It was probably caused by the tripping of the yaw damper lever.
1115:25 The stall warning sounded for two (2) seconds and then stopped. This was probably due to the following:
The stall warning began to sound at 1115:25 because the AOA reached

3-6


approximately 16° at 15:22, exceeding the threshold angle of 15° for the configuration SLATS EXTENDED. However, Vc dropped below 75 kts. at 1115:27, so the ADC apparently judged the AOA to be "NO MORE VALID" and terminated the warning function.
It is considered probable that around 1115:25, the aircraft fell into a stall, yet continued to climb until reaching its highest point. The aircraft remained in a stall condition until impact.
1115:26 The pitch angle of the aircraft reached the maximum angle of 52.56°.
1115:27 The THS returned to -7.4° from 12.30° where it had been at 1115:11, and then remained there until 1115:33.
It is considered that this was probably caused by the tripping of the pitch-trim lever at 1115:27.
1115:28 The master caution (single chime) was recorded on CVR. It is considered that this was probably caused by the tripping of the pitch-trim lever.
1115:31 The master caution (single chime) was recorded on CVR. It is considered that this was probably caused by the tripping of the ATS Lever. On the CVR the F/O's words, "Set, set," are recorded. It is considered that these words were probably the F/O's request to reset something when he saw CAUTION MESSAGE displayed on the ECAM. Considering the fact that the THS was moved again after about 1115:35, the F/O probably reset the pitch trim.
After reaching the highest point at approximately 1,730 ft pressure altitude, with a pitch angle of 43.80, the aircraft began to descend, while rolling and yawing greatly to the left and right. There are records showing that corrective actions were taken by the crew by means of the ailerons and rudder during this period.
1115:31 The thrust decreased temporarily.
This was presumably caused by surges that occurred in both engines.
1115:34 From this point until just before the impact , the F/O (PF) shouted "Power" repeatedly. This was linked to his utterance of "Quick push nose down" at 1115:26 and is assumed to indicate a desire to increase thrust and thus recover lost speed.
1115:35 The CAP (PF) performed a nose-up operation using the control wheel.
It is considered that the CAP had until then been applying nose-down input to the elevator in order to decrease the pitch angle, but at this moment he applied a nose-up input to the elevator in response to the decrease in pitch angle and the steep descent of the aircraft.
1115:37 The Mode 2 warning of the GPWS sounded "TERAAIN TERRAIN" once. Computation of the CAS and AOA that had earlier paused, now resumed.

3-7


1115:40 The aural stall warning, which had stopped at 1115:27, sounded again and continued until impact.
1115:45 From the conditions in which the CVR and DFDR recordings ended, it was estimated that the aircraft crashed at approximately 1115:45.

3.1.2.2 Analysis of Flight Conditions

(1) Concerning activation of the GO lever

It is recognized that the F/O (P/F) triggered the GO lever at 1114:05, judging from the following: the increase in engine thrust starting at 1114:05, as recorded on DFDR; the CAP's (PNF) utterance at 1114:06, the sound of LANDING CAPABWITY CHANGE at 1114:09, the CAP's caution at 1114:10, and the F/O's response at 1114:11, all of which were recorded on CVR.
The F/O activated the GO lever, causing the FD to change to GO AROUND mode, and the ATS to be engaged in THE mode.
It is considered that at 1114:06 the CAP said "EH. EH. AH," on seeing the display change on the FMA.
The F/O (PF) seems to have used ATS with his hand on the thrust levers, judging from the smooth transitions of both thrust levers recorded on DFDR until then.
It is considered that the F/O may have mistaken the GO lever for the AT disconnect push button in an attempt to change the ATS into manual thrust, or that he tried to move the thrust levers to control the thrust and thereby inadvertently triggered the GO lever. The reasons why are not clear, but, at any rate, he inadvertently triggered the GO lever.
The GO lever of the A300-600R type aircraft is positioned below the thrust lever knob. The direction that the GO lever is operated in is the same as the direction in which the thrust lever is retarded, or as the same direction that the fingers move when gripping the thrust lever knob With this arrangement, the possibility exists for an inadvertent activation of the GO lever during normal operation of the thrust levers (See attached Photograph 51).

(2) Concerning CAP's direction at 1114:12

The CAP gave an instruction to the F/O, saying "Disengage it". The definite meaning of the word "it" is not found in the CVR records, but there seem to be two possible meanings -- "Auto Throttle" and "GO AROUND mode" -- which it could represent. This is inferred from the following:
         The DFDR records show that activation of GO lever led to a thrust increase; the EPRs stopped at the value of 1.21 at 1114:08. It is considered that the F/O probably pushed the AT disconnect push button while holding the thrust levers -- which were moving forwards at the time -- and then retarded the thrust levers.
After this action the EPRs were reduced slightly. It seems that before 1114:12 the Auto Throttle had already been disengaged by the F/O, and the FMA display had

3-8


         changed to "GO AROUND".
It was most likely that the CAP's instruction in this situation meant that GO AROUND mode should be disengaged, because he must have seen the FMA display.

(3) Concerning the CAP's word at 1114:16

At 1114:16, the CAP said "That", and the F/O said "Aye". The precise meaning of the word "that" is not found in the CVR records, but there appear to be three possibilities:
         (1) The CAP instructed the F/O to engage the AP(s).
(2) Because the CAP's instruction at 1114:12 had not been followed, he repeated it.
(3) The CAP's word did not represent an instruction, because the nuance is ambiguous. In this case, the F/O seemed to have given the CAP only a response.

However, it was not possible to determine which of the above scenarios is the correct one.

(4) Concerning use of AP

According to the DFDR records, both APs were engaged at 1114:18. Around this time, no verbal exchange as to AP engagement was recorded on CVR. However, there seem to be the following possibilities concerning engagement of APs:
         (1) Possibility that the CAP instructed the F/O to engage the APs
If it is assumed that the CAP's word at 1114:16 meant the item (3)-(1) above, the F/O might have engaged the APs in accordance with the CAP's instruction.
According to the positions of the thrust levers (throttle resolver angles) recorded on DFDR from 1114:12 to around 1114:18 (which is recorded every 4 seconds with one second time gap between NO.1 and NO.2), the positions of the thrust levers varied slightly respectively, so taking into account the time allowance for the actions taken by the F/O, it is considered possible that the F/O tried to select the LAND mode at first, then once held the thrust levers after taking actions to change mode, and furthermore took action to engage the APs.
(2) Possibility that the CAP engaged the APs himself
(3) Possibility that F/O engaged the APs himself
If it is assumed that the CAP's words at 1114:16 meant item (3)-02 or (3)-03 above, another possibility is that the F/O engaged the APs without the CAP's consent, or without notifying the CAP. In this case, according to the CVR records, because the F/O had so far operated during the whole flight based on the CAP's instructions, or with the CAP's consent in advance, there seems to be a possibility that the F/O instinctively engaged APs for their assistance.

However, it was not possible to determine which of the above scenarios is the correct one.

3-9


         The reason why either the CAP or the F/O engaged the APs may have been that the crew intended to regain the normal glide path by selecting LAND mode and engaging the APs.

(5) Concerning the operation of Pitch Trim Control Switch

The activation sound of the pitch trim control switch was recorded on CVR at 1114:20, 1114:34, and 1114:39 respectively. If a hypothesis is made that the F/O knew that THS trimming operated by the pitch trim control switch was inhibited during AP engagement, the CAP may therefore have engaged the APs unknown to the F/O. However, from the fact that the F/O actually disengaged both APs at 1114:49, he may finally have recognized that the APs had been engaged by this time.
On the other hand, even though the F/O may have recognized AP engagement, it is still possible that he may have involuntarily operated the switch while pushing the control wheel.

(6) Concerning disengagement of GO AROUND mode

The CAP said "You, you triggered the GO lever" at 1114:10, and alerted the F/O (at 1114:30 and 1114:45) to the fact that GO AROUND mode had been engaged. This leads to the possibility that after the F/O had triggered the GO lever, the CAP, watching the FMA display, intended to disengage GO AROUND mode and instructed the F/O to do so. However, it is inferred from the following facts that GO AROUND mode remained engaged.
         (1) THS moved in the direction opposite to the F/O's input at the control wheel.
(2) Disengagement of GO AROUND mode and engagement of other modes led AP No.2 to be disconnected, but no data concerning this was recorded on DFDR.
(3) CAP continued until 1114:45 to alert the F/O to the fact that GO AROUND mode was still engaged.

In order to disengage GO AROUND mode, both lateral mode and longitudinal mode (except LAND mode) must be selected. Direct access to the LAND mode button cannot disengage GO AROUND mode (by selection of either lateral or longitudinal mode a display of GO AROUND on FMA will turn oft).
However, judging that GO AROUND mode still remained engaged, it is estimated that what the crew's operation on FCU was not correct procedure to disengage it : he must only have pulled LAND mode button. And also, taking into account that the CAP said "I, that LAND mode?" at 1114:58, the CAP seemed to have intended to disengage the GO AROUND mode and select LAND mode.
The procedure for performing an approach by disengaging GO AROUND mode once engaged and then engaging LAND mode is unusual in the final phase of approach. However, the fact that the crew did not change modes as intended seems to have been due to their lack of understanding of the Automatic Flight System (AFS).

(7) Concerning the sequence leading up to the out-of-trim situation

3-10


After the GO lever was triggered, the sequence leading up to the out-of-trim situation was as follows:

(1) Just before 1114:05

         In the landing configuration with landing gear down and SLATS/FLAPS at 30/40, the aircraft continued descent along 30 ILS glide path with a speed of approx. 140 kts., pitch angle of approx. 40, both EPR at approx. 1.1, THS at -5.3°, and the elevator angle (relative to THS) at 0° to 1° nose down.

(2) Just after 1ll4:05

         While crossing approx. 1,070 ft pressure altitude, the GO lever was activated by the F/O, and when the EPRs increased to 1.21 at about 14:08, the thrust levers were manually pulled back slightly.
The aircraft increased its speed and pitch angle slightly, deviating above the ILS glide path. The F/O applied push-down input to the control wheel but it was insufficient and also he did not retard the thrust levers sufficiently. These circumstances led the aircraft to level off about 1 ,040ft pressure altitude around 1114:10. For a while the THS had stayed at 5.3° And pitch trim control switch was not operated.

(3) 1114:18

         While the aircraft continued level flight, both APs were engaged in CMI) with the FD already in GO AROUND mode, and the APs, were brought into GO AROUND mode.
In the meantime, the F/O (PF) had been pushing the control wheel since 14:05, when he had activated the GO lever, in an effort to return to the normal descent path.
At the time when the APs were engaged in CMI), the elevator angle was 3~50 nose-down. The angle decreased to 2.80~2.40 temporarily in the period between 1114:19 and 1114:20, but the nose-down angle gradually increased thereafter.
GO AROUND mode was engaged while the F/O was pushing the control wheel. The AP attempted to move the elevators and THS toward the nose-up direction, but this resulted in the elevators' function being overridden and the THS beginning to move in the nose-up direction from 5.3°
However, the nose-down operation of the elevators performed by the F/O canceled the aerodynamic effect of the THS nose-up which was controlled by the AP, and the aircraft temporarily continued level flight. The surface area of THS (including the elevators) is approximately three times that of the elevators. The aerodynamic effect per unit travel angle of the THS is therefore considerably greater than that of the elevators.

(4) 1114:24

         To correct the descent path, the F/O (PF) began to retard the thrust levers and reduced the EPRs from approx. 1.17 to approx. 1.00 by 1114:31.

3-11


         As a result, the speed began to decrease from 146 kts., causing the nose-up tendency also to decrease. This retard operation of the thrust levers and the push-down operation of the control wheel by the F/O against the movement of the THS in nose-up direction, together caused the pitch angle to decrease, and the aircraft began to descend around 14:26.

(5) 1114:30

         The pitch angle which had decreased to 1.20 again began to increase.
This is considered to be due to the fact that the pitch-up effect generated by the nose-up movement of the THS became larger than the pitch-down effect brought about by the push-down operation of the control wheel from that time on.
The speed continued to decrease slowly. As the pitch angle increased, the AOA also began to increase.

(6) 1114:37

         While crossing approx. 880 ft pressure altitude, the THS reached the full nose-up position of ~12.30, and the elevator was moved to 8.50 in the nose-down direction. Around this time, the descent rate was approximately 1,000 ft/mm.
Although the control wheel was still being pushed, the pitch angle and AOA continued to increase, while the speed continued to decrease. In order to deal with the continuous decrease in speed, the F/O increased the thrust slightly.

(7) 1114:49

         While crossing approx. 700 ft pressure altitude, the APs were disengaged, but THS remained ~12.30, and out-of-trim condition continued.

(8) Concerning activation of Alpha Floor Function (Refer to 3.1.11.6)

         Just after the APs were disengaged at 1114:49, the mobility of the control wheel (being pushed by the F/O(PF)) increased a little, thereby moving elevators in the nose-down direction; and pitch angle and AOA decreased. A few seconds later, forward pressure on the control wheel was loosened a little, and pitch angle and AOA increased again. When the aircraft crossed approximately 570 ft pressure altitude at 14:57, as airspeed was 127 kts., both EPRs were 1.04 and pitch angle was 8.60 and AOA exceeded threshold angle of 11.50 for the configuration of SLATS/FLAPS 30/40, the alpha floor function was activated.
         Although at this point of time, THS was ~l2.30 and the elevator angle was 9~90, the sudden increase of power due to the out-of-trim condition and the activation of the alpha floor function generated a pitch up moment. As for the fact that the F/O (PF) loosened forward pressure on the control wheel a little several seconds after AP was disengaged, it is considered likely that he did so in order to correct the pitch angle. However, even if the F/O (PF) had not loosened forward pressure on the control wheel at this point of time, the AOA, sooner or later, must have been exceeded threshold angle of 11.50 due to the trend of speed,

3-12


         pitch angle and the AOA if the aircraft had continued to approach under the above-mentioned out-of-trim condition.

(9) Concerning continued approach

         The CAP (PNF) had had the F/O perform the PF duty while making an WS approach. Judging that although the aircraft once deviated above the glide path after the F/O triggered the Go lever, it began to return to the normal glide path due to the F/O's fully forward pressure on the control wheel and reducing the thrust following the CAP's instruction, and the runway was visible to the crew due to good weather condition and sufficient visibility, the CAP probably intended to have the F/O continue the approach. It is considered that the CAP paid his attention outside to assess the aircraft position and the descent flight path from the view of the runway and the CAP would have instructed the F/O only to have the aircraft recover the normal glide path.

(10) Concerning override of the AP

The CAP (PNF) instructed the F/O (PF) repeatedly to push the control wheel. There seem to be the following possibilities as to why he did so:

(1) The CAP did not recognize that the APs were engaged.
(2) The CAP recognized that the APs were engaged, but he believed that the F/O had disengaged GO AROUND mode either when the CAP instructed him to do so or when he had pointed out that GO AROUND was still engaged.
(3) The CAP recognized that the APs were engaged, but he thought that he could manually override the AP, based on his flight experience of B747-200 and 400 aircraft.

The aircraft incorporates a supervisory override function which allows pilots to assist the AP by applying a force on the control wheel when capturing the Glide Slope, the Localizer or the VOR course. There seems to be a possibility that the crew's experience in using this function led to their mistaken belief that they could override the APs during all phases of approach. This could have led them to override the APs while in GO AROUND mode.

The hazard of overriding the elevator by operating the control wheel while the APs are engaged in GO AROUND mode is described as a "CAUTION" in the FCOM. The reason why the crew took actions which nevertheless resulted in an out-of-trim condition, is presumed to be that they had not properly understood the contents of these cautions, and of other related descriptions in the FCOM. As mentioned later, the fact that descriptions in the FCOM are not easy for pilots to understand, and functions to alert pilots of THS movement are not properly incorporated, probably affected this outcome as a background factor (Refer to 3.1.11.3 and 3.1.11).

3-13


(11) Concerning the CAP's remarks at 11 14:58

At 14:58, CAP (PNF) said," I, that LAND mode?". This may be interpreted as follows:

         (1) At 14:12, CAP (PNF) instructed the F/O (PF), saying, "Disengage it." (Disengagement of GO AROUND mode was very likely).
(2) Twice after this, the CAP (PNF) cautioned the F/O (PF) that GO AROUND mode was still engaged.
(3) CAP instructed the F/O (PF) repeatedly to push the control wheel.
It is considered likely that the CAP said "I, that LAND mode?" in puzzlement on realizing that the aircraft was still not adopting the proper attitude for descent, in spite of his above-mentioned instructions and cautions.

(12) Concerning timing of control take-over

After the APs had been engaged, the F/O attempted to recover the normal descent path, but could not maintain airspeed and aircraft altitude to do so.
To deal with this situation, the F/O (PF) disengaged both APs at 14:49, saying "Sir, auto pilot disengaged."
At 14:51, the F/O (PF) reported to the CAP (PNF), saying "Sir, I still cannot push it down, yeah", probably because the pitch angle was still high, and the aircraft was still not responding to his actions. Again at 15:02, the F/O (PF) reported, saying "Sir, throttle latched again."
Until then, the CAP (PNF) appears not to have fully grasped the flight situation. Hearing the F/O's (PF) report above, the CAP (PNF) seems to have decided to take over the controls to deal with the unusual situation. At 15:03, CAP took over the controls. However, even at this point, the CAP (PF) still seems to have been unaware that the THS was at the nose-up limit.
Although the CAP (PNF) would have been unable to experience directly the unusually strong resistive force of the control wheel until he took the controls, in view of the points described below, he could still have recognized to some extent that an abnormal flight condition had arisen.

         (1) GO-AROUND mode continued to be displayed on the FMA.
(2) The CAP had earlier instructed the F/O to push the control wheel and retard the thrust levers in order to regain the normal glide path. However ,the aircraft did not respond as the CAP had intended when he issued his instructions.
(3) The CAP had had to give directions and cautions (such as item 02 above) to the F/O, one after another. This fact itself suggests that the F/O must no longer have been in a condition to perform PF duty adequately.

3-14


However, it is considered that the CAP's situational awareness as PlC for the flight was inadequate, control take-over was delayed, and appropriate actions were not taken.

(13) Concerning GO AROUND after CAP took over controls

Immediately after the CAP took over the controls, he retarded the thrust levers to reduce the power before calling "GO lever" at 15:11.
It is considered that when CAP took control, although he was aware of an unusually strong resistive force on the control wheel, he still intended to make a landing; so he pulled the thrust levers to try to reduce the pitch angle which was increasing. However, judging that the CAP was unable to stop the pitch angle (which was increasing in nose-up direction), it is estimated that he gave up landing, uttering "How come like this?", decided to go around and then called out "GO LEVER" while increasing the thrust, which had earlier been reduced, to full thrust.
In normal go-around procedure, PF calls "go around flap" as he operates the GO lever, PNF moves the SLATS/FLAPS lever one step up, and after calling "positive climb" PNF performs a gear up operation following PF' 5 order.
In this case, however, the correct procedure was not followed as stated. After "GO lever" was called, it took about seven (7) seconds before initial movement of the SLATS/FLAPS lever in the retract direction took place.
While the SLATS/FLAPS lever should be moved from 30/40 to 15/20, from the CVR record it is considered possible that the lever was moved to 15/0 or even higher, to the 0/0, before being lowered again to the 15/15 position. The landing gears were left in the down position (See attached Figure 27.).

(14) Concerning operations performed to deal with increasing pitch angle and steep climb

(1) 1114:57

In a pitch-up side out-of-trim condition with the THS at -l2.3° and the elevators at 9.9°, the alpha floor function was activated, suddenly increasing the thrust and which caused a large pitch-up moment to be generated.

(2) 1115:03

The pitch angle did not stop increasing despite the CAP's (PF) efforts who, after taking the controls, pushed the control wheel to the forward limit and retarded the thrust levers.
Around 15:04, the aircraft which had, until that point, been descending, began to climb from approx. 500 ft pressure altitude (approx. 360 ft radio altitude).

(3) 1115:11

When the CAP(PF) increased the thrust again and called, "GO lever", the aircraft was climbing through approx. 600 ft pressure altitude with pitch angle at 21.5°.
The pitch angle was further increased by a large pitch-up moment generated by the increase of thrust under the pitch-up side out-of-trim condition.
Speed began to decrease from 137 kt around 15:08.

3-15


(4) 1115:12

It is considered that the CAP (PF) continued to push the control wheel fully to reduce the pitch angle, and intermittently operated the pitch trim control switch in the pitch down direction, as indicated by the slow return of the THS from the limit angle of -12.3° to -l0.9° by 15:19. (Intermittent use of the switch does not generate the "Whooler" tone.)
Thereafter, the THS moved again from -10.9° at 15:21 to -7.4° at 15:27. It is considered possible that the alpha trim function activated because the AOA at 15:23 was approx. 18°, which exceeded the threshold angle of 17° for SLATS/FLAPS 15/20 and 15/15 configurations.
It was not determined whether or not the manual trim had been operated during the above period.
During the period from 15:27 to 15:33, the THS remained at -7.4°
The CAP (PF) operated the pitch-trim only intermittently during the go around. Consequently, it is considered that he was not aware of the THS state.

(5)1115:20

There are three feasible ways to reduce increasing pitch angle: to push the control wheel, to regain trim, and to reduce the thrust. Under the conditions of steep climb and continued decrease of speed, it seems that the CAP(PF) hesitated to reduce the thrust.
However, at this point of time, when the speed had decreased to 115 kts., the pitch angle had increased to an abnormal 40.3°, it is considered that the CAP (PF) retarded the thrust levers to reduce pitch angle.
At approximately 15:23 No.1 thrust lever was retarded to a position near idle and No.2 thrust lever was retarded slightly. This is probably because, although the CAP was hastily attempting to retard the thrust levers in an effort to correct the aircraft attitude which continued to climb steeply, the CAP's (PF) hand came off the thrust levers at the above-mentioned position while continuing to push the control wheel with the aircraft's steep nose-up attitude.
Around 14:27, the thrust levers were moved to a position close to full thrust. This seemed to result from the fact that either the CAP or the F/O pushed the thrust levers forward in an attempt to recover lost speed.
It is also presumed that the aircraft's nose-up pitching moment was further increased as a result of the SLATS/FLAPS retracting from 30/40 to 15/15.

(15) Concerning crew coordination between the CAP and the F/O (See Appendix 2-1)

(1) At 1059:04 and 1113:14, the CAP (PNF) read out the approach checklist and the landing checklist at the request of the F/O (PF), but these were not performed in the proper manner because the CAP (PNF) read the items only to himself, including those to which the CAP and F/O (PF) should responded together.

(2) At 1114:18, both APs were engaged, but nothing was said that expressed definite

3-16


instructions or intentions to make the F/O (PF) change mode or engage the APs.
The CAP's words when instructing the F/O to disengage GO AROUND mode were only "Disengage it". These words did not describe a definite operation.

(3) At 1114:12, the CAP said, "Disengage it." At 14:30, he said, "You, you are using the GO AROUND mode." At 14:45, he said, "It's now in the go around mode." By these words, the CAP pointed out the current mode and instructed the F/O to change the mode.
In response to this, it seems that the F/O took some action to change modes, but was unable to successfully engage LAND mode in the end. In this case, the F/O did not immediately report to the CAP (PNF) that he could not change modes (or that did not know how to change modes).
In the meantime, after the CAP pointed out the current mode and instructed the F/O to change modes, it is considered that the CAP did not check the FMA display properly on each occasion to see whether the mode had actually changed or not.

(4) When the F/O was instructed by the CAP ("Push more.") he did not report the abnormally strong resistive f6rce of the control wheel to the CAP. As a result, it is probable that the CAP was not fully aware of the situation and that his instructions to the F/O were therefore inadequate.

The F/O must have perceived the abnormally strong resistive force of the control wheel, but he, who was under high stress from the following factors, probably delayed reporting the situation properly to the CAP:

(5) During approach, the CAP had instructed the F/O to perform PF duty, but after the F/O triggered the GO lever, the CAP gave a series of instructions and cautions to the F/O about control and operation of the aircraft, one after another. As a result the CAP made the F/O lose his autonomy and disregard their duty assignment, namely, that the CAP was the PNF and the F/O was PF.

(6) The FCOM 2.03.18 (page 3) stipulates under the title of "STANDARD OPERATING PROCEDURES - STANDARD/APPROACH" that if the speed exceeds VAPP +10 Kts. or becomes less than VAPP -5 Kts. ,or if the aircraft deviates a dot or more from the glide slope during an approach, the PNF should call out the fact.

3-17


    At 1114:17, the aircraft deviated more than a dot upward from the glide slope, and speed decreased to less than -5 Kts. from the VAPP of 140 KCAS as the aircraft continued approach. Despite this, the CAP did not call out these facts as PNF.

(16) Concerning the tones of MASTER CAUTION record on CVR.
     There is a possibility that audible tones of MASTER CAUTION (SINGLE CHIME) recorded on CVR at 1115:23, 15:28 and 15:31 were set off respectively by the tripping of the yaw damper levers, pitch trim levers and ATS lever.
These CAUTIONs were triggered because the input sensor data were judged as "INVALID" and the systems relating to the above mentioned levers were disconnected.
Because the aircraft's attitude and speed changed rapidly during this phase of the flight, the possibility that the MASTER CAUTIONs were generated by a different cause cannot be ruled out.
However, if the conditions on which these CAUTIONs sounded , the relations among these occurrences around the time when they sounded, and the analysis of the sound spectrum of the CVR recordings are considered together, it would be highly possible that these CAUTIONs are the same as the ones described at each of the times in paragraph 3.1.2.1.

3.1.3 Estimation of crash time

As described in 3.1.1, it is estimated that the CVR and DFDR stopped recording around 1115:45. This was probably caused by the breaking of cables on impact. The time of the crash is estimated to be around 1115:45 (when the CVR and DFDR stopped recording).

3.1.4 Attitude of Aircraft at the Time of Crash and Damage to Aircraft

3.1.4.1 Attitude of Aircraft at the Time of Impact

From the DFDR records, it is estimated that the aircraft stalled, then 4escended steeply with wildly changing roll angle, and impacted the ground.

The spot where the aircraft hit the ground was an unpaved, flat landing area. There were marks left on the ground surface that clearly identified those portions of the aircraft which had hit the ground. From the shapes of the marks and these positional relationships as well as the condition of destroyed landing gears, it is inferred that on impact, the aircraft was in a somewhat left-wing down, nose-up attitude, and was in an almost level attitude.

3.1.4.2 Investigation of Broken Landing Gears

The broken nose and main landing gear were investigated in order to analyze the conditions of the landing gear, the aircraft attitude, and other associated conditions of the aircraft at the

3-18


time of the crash.

All of the broken landing gears had signs of compressed oleo struts with buckled cylinders, which implies that the gears received an upward thrust, and so that the aircraft had impacted the ground with all landing gears extended.

The rear bogey beam of the LH main gear, which presumably touched the ground first, was sheared off in a ring shape at a relatively thin portion beside the strut attaching part. The rear inner wheel (with tire still inflated) of the LH main gear and the accompanying brake assembly, were flung away furthest to a point approximately 190 meters from the point of impact. It seems that this occurred because of a rupture that occurred at the moment of impact, when the kinetic energy was greatest and its loss was minimum, and also by a high rebound force brought about by the tire.

The damage to the two (2) front tires of the LH main gear was extensive, with outer tires burnt and inner tire burst owing to impact. An assumption from these conditions is that breakage occurred at the front and rear of the bogey beam, with subsequent impact transmitted directly from the ground to the strut. The direction of lacerations on the tires suggested that they skidded to the right.

3.1.4.3 Crash Circumstances and Damage to Aircraft

The crash process from when the aircraft first hit the ground to when it was destroyed is estimated to be as follows:
(1) The LH main gear of the aircraft impacted the ground first, and at this point of time, there were no other parts in contact with the ground.
Compared with the wheelbase of the aircraft, the measurement between the scars on the ground was greater. This implies that the aircraft was moving forward in a slightly nose-up attitude.
At this point in time, the aircraft's magnetic heading was approximately 15 degrees (015°) and its side-skid angle was approximately seven degrees (7°) to the right, judging from the aircraft's attitude, that was almost level, and the direction of motion of the aircraft deduced from the marks left on the ground.
Since the LH main gear impacted the ground first, the aircraft's began to turn counterclockwise as viewed from above.
(2) The RH main gear impacted the ground a little later than the LH main gear.
The fact that the ground scars of both main gears were not long in comparison with the track of the wheels suggests that the aircraft was descending at a steep flight path angle. The pitch angle at this time was approximately four degrees (4°) nose-up, as calculated from the geometrical characteristics of the aircraft, the conditions of the broken LH main gear, and the position of the nose gear.
(3) At the moment the LH engine impacted the ground (receiving the strongest shock), the RH

3-19


engine was beginning to receive the impact from the ground. Both the main gears were being destroyed in that while.
The roll angle at this time was approximately three degrees (3°) to the left as determined from the positions of both main gears, nose gear, and LH engine.
(4) By the time the nose landing gear impacted the ground, and was under maximum shock, both main gears had already been destroyed and the LH engine was in the process of destruction.
(5) At the above point in time, the LH wing-tip also hit the ground.
Judging from the positions where the marks of the LH wing tip and RH engine were found, all the landing gears were destroyed and the bottom of fuselage started breaking. The aircraft received an additional counter-clockwise moment after touching the ground and the entire airframe was distorted.
The recovered LH wing-tip had a deformation showing an impact it had received from contact with the ground at its lower, slightly inboard section. On the other hand, the LH wing tip was damaged on its upper part, indicating that it was damaged when the RH wing was destroyed.
(6) After the LH wing-tip had been destroyed, the LH flap track touched the ground. Destruction of the fuselage progressed to the lower part of the floor, and shortly afterwards, the horizontal stabilizer was flung onto the ground almost in a level attitude.
The direction of the scratch marks left on the lower access panel of the THS corresponds roughly to the direction in which the debris were strewn (22 degrees (022°) in magnetic heading).

From extrapolation using the DFDR records, the trajectory angle at the time of impact was estimated to be approximately 32°.

3.1.4.4 Condition of Wreckage

After the aircraft had impacted against the ground, the major parts of the wreckage are estimated to have been in the conditions described below.

From the scattered condition of wreckage, it is estimated that the momentum vector of the aircraft in the level plane when it crashed was approximately 22 degrees (022°) from magnetic north and approximately 42 degrees (42°) to the right of the centerline of the runway 34.

(1) The LH engine, having dropped from the wing pylon, tumbled forward; the lower skin of the aft fuselage remained in the vicinity of the spot where it had first contacted the ground, and the horizontal tail plane and APU compartment were ruptured and had separated.
(2) The outer flap, center flap, and outer wing of the LH wing were ruptured and had separated from the wing.

3-20


(3) The outer flap, center flap, inner flap, spoiler, outer wing and other components on the RH wing were ruptured and had separated from the wing, and were strewn as far as the irrigation ditch.
(4) The upper portions of the forward and aft fuselage sections, along with the portion of wing that remained attached to them, tumbled to the vicinity of the irrigation ditch, together with the ruptured and separated vertical tail plane and upper portion of the aft fuselage, all in broken state.
(5) The cargo loaded on board was scattered in the area between the spot where the aircraft hit the ground and the vicinity of the irrigation ditch, and almost all ruptured seats were found near the irrigation ditch.
(6) The fuel that had leaked from the broken LH wing when the aircraft crashed into the ground splashed over the area from the vicinity of the spot where the LH wing first hit the ground to where the center of the wing had come to rest, and fuel from the RH wing was scattered widely, together with debris of the RH wing, as far as the vicinity of the irrigation ditch.

It is estimated that among the wreckage strewn forward, by the forward-acting inertial force generated when the aircraft crashed, were items such as the fuselage section aft of the central wing section, functional components, cabin furnishings, seats, and cargo burned when the fuel ignited, and were destroyed by expanding fire.

3.1.5 Investigation of Engines and FADEC

Investigation of the dismantled engines revealed that damage to components of both engines were indicative of their rapid destruction. Rotors blades were torn and deformed in the direction opposite to that of rotation. All these conditions support the assumption that the engines had been running at high speed until the aircraft crashed.

The data recorded in all the channels of both FADECs show that surges occurred in both engines, indicating a sudden drop of combustion chamber pressure during the flight.

Also, engine data recorded by the DFDR show that the engine pressure ratio, fuel flow rate, and high-pressure shaft rotating speed dropped for short periods at 1115:31 for the No.1 engine and at 1115:32 for the No.2 engine, to values lower than those that should normally result in response to the thrust lever operation performed. This could have been due to the FADEC counteracting the engine surge. Subsequently these engine parameters returned rapidly to values that normally correspond to the thrust lever movements, and no abnormalities caused by surges of the engines were subsequently detected.

When engine surging occurs, flames sometimes shoot out from the front and rear of the engines. Both engines may therefore have emitted light as the aircraft fell into an abnormal

3-21


condition during stall.

It is estimated that the engine surge which occurred when the aircraft fell into a stall condition was due to a phenomenon called "inlet distortion" in which uniform air flow through the engine air inlets is not available owing to a high AOA.

It is estimated that the AOA at this point of time was far greater than the range for normal operation, exceeding the engine air intake airflow angle limit permitted for the aircraft.

3.1.6 Results of Investigation of Computers and Other Equipment

3.1.6.1 Computer Memories

The memories of one FAC, one FMC, one SGU-EFIS, two SGU-ECAMs, one FWC, two ADCs, three IRUs, two GCUs, and one ILS receiver could be read out, but there nothing was recorded that might have been of relevance to the accident.

3.1.6.2 Investigation of Disassembled Components and Others

Of the components recovered from the crash site, 54 items (128 Units) including the AP pitch actuator, elevator actuators, trim actuator gear box, and center pedestal and so on, were analyzed by means of disassembly and other methods. Nothing abnormal was found except the damage inflicted at the time of the accident.

3.1.6.3 IRS Mode Selectors

As shown in 2.15.2, IRS mode selector No.3 was in the NAV position, while the No.1 and No.2 selectors were in the OFF and ATT positions, respectively. The IRS is an essential system for engaging the APs, and the DFDR records indicate that both AP No.1 and No.2 were engaged. Also after that, the CVR records did not contain any data that suggests failure of the IRS s, and further, there were no evidence showing changeover of the selector in the Left side Instrument Switchings record. These facts imply that the No.1 and No.2 IRS mode selectors were moved by the impact at the time of ground impact, or later.

3.1.6.4 SLATS/FLAPS Positions and THS Angle

(1) SLATS/FLAPS positions

By comparing the screw jack nut positions of the SLATS/FLAPS actuators of an aircraft of the same type with those of the crashed aircraft, the nuts of the crashed aircraft were to be at the positions corresponding to 15/15.
These positions are almost in agreement with the last SLATS/FLAPS angles of 17.05°/8.25° (15/15 position) recorded on the DFDR.
Also, the shaft of the broken SLATS/FLAPS command sensor unit was found seized up at the

3-22


15/15 position.
Judging from the above findings, it seems that the SLATS/FLAPS lever position of 15/20 shown in paragraph(2), Section 2.15.2, was a result of a movement that might have occurred as a result of impact, or some time afterward.

(2) THS angle

The THS angle was determined to have been approximately 8° as a result of a comparison made between the THS screw jack nut position of the crashed aircraft and that of an aircraft of the same type. This angle roughly agrees with the last THS angle of -7.4° recorded on the DFDR.
It was also found that the pitch trim control cable system was broken. The conclusion from the above is that the THS position indicator reading of -9.4° / 9.5° shown in paragraph(2), Section 2.15.2, is a result of a movement in the indicator that might have occurred at impact or thereafter.

3.1.7 Seated Positions of CAP and F/O

3.1.7.1 Analysis Based on CVR Records

(1) Analysis of data recorded on channels 2 and 3 of the CVR, could not determine where the CAP and F/O had been seated, since the voices and sound records, including radio commutations on channel 2 (on the F/O's seat side) are exactly the same as that on channel 3 (on CAP's seat side), owing to the cockpit intercom communication system.
(2) From the record of the CAP's call out, "SHOULDER HARNESSES" (1100:05), to confirm the wearing of the shoulder harnesses as he read the approach checklist, the record of the F/O's response, "FASTEN RIGHT", and the record of the CAP's confirmation of approach check list completion, "OK, FASTEN LEFT, APPROACH CHECKLIST COMPLETED" (1105:05), on the CVR, it is estimated that the CAP was seated in the left seat and the F/O in the right seat.
(3) From the conversation between the CAP and F/O, recorded on CVR, about control of a light, it is recognized that the CAP was adjusting a light (refer to attached Figure 23).

Of the adjustable lights, if the one that the CAP (PNF) was adjusting was the captain and center instrument light, or the main instrument panel flood light, it is considered that the CAP was seated in the left seat (the adjusting knobs of these lights are located at the left end of the instrument panel, on the same side as the left-hand seat).
However, if the light which the CAP (PNF) was controlling was either the glare shield light, pedestal and overhead panel light, or the dome light, the adjusting knobs of these lights could be operated from either the right or left seat.
As inferred from the above, it is therefore not possible to determine which seat the CAP was occupying.

3-23


3.1.7.2 Analysis Based on Wreckage of Seats (See attached Figures 24 and 25)

(1) The recovered seat wreckage had damage marks caused at the time of the ground impact. Analysis of the seat-setting positions from these marks is as follows: for the left seat (CAP's seat), the vertical position was set approx. 70 mm above the lowest position, and its longitudinal position was near the foremost position, whereas for the right seat (F/O's seat), the vertical position was set approx. 30 mm above the lowest position, and the longitudinal position was set approx. 33 mm rearward from the foremost position.
As inferred from these settings, the right seat was set approx. 40 mm lower than, and approx. 33 mm rearward of the left seat. Since the crew positions are adjusted by means of an eye locator, such that the eye levels of both the right and left seat occupants will be about equal, it is considered that the person in the right seat was taller than the one in the left seat.
(2) According to the airman medical certificate, the CAP was 162.5cm tall and the F/O 178.1cm tall. From this data, and the damage marks to the seats discussed in (1), it is estimated that the CAP was in the left seat and the F/O in the right seat.

3.1.7.3 Seated Positions

From the implication of the CVR record concerning the wearing of the shoulder harnesses, and by comparison of the vertical and longitudinal positions of both seats estimated from the damage marks to the seat wreckage, as described in sections 3.1.7.1 and 3.1.7.2, it is considered that the CAP and the F/O were seated in their formal positions: the CAP in the left seat, and the F/O in the right seat.

3.1.8 Injury to Passengers and Seat Assignment (See attached Figure 26)

3.1.8.1

The number of survivors of this accident was seven (7), all being seriously injured. As noted in section 2.12.2, 16 passengers were taken to several hospitals by rescue workers.
(1) Among passengers hospitalized, six (6) persons were found dead on arrival. The cause of death of four (4) of the six (6) was whole body contusion and fractures; the other two (2) died of whole body contusion and thermal injuries.
It is estimated that four out of the six (6) were seated in the forward section of the cabin and the other two (2) in the aft section.
(2) Three (3) out of the (10) seriously injured passengers died on April 27, April 28, and May 1, respectively, at the hospitals to which they had been admitted. The cause of death of these three (3) was whole body contusion and fractures.
The estimated seat assignment for two (2) of the three (3) was in the forward section of the cabin and one (1) in the aft section.

3-24


(3) The remaining seven (7) seriously injured were diagnosed as suffering from traumatism and various external injuries, primarily bone fractures.
All of these seven (7) passengers were seated in the forward section of the cabin, in front of the wings.

3.1.8.2

According to the post-mortem report, the cause of death of those passengers who died before hospitalization was determined to be whole body contusion, fractures, and thermal injuries.

The positions where the passengers suffering whole body contusion and fractures were seated extended over the whole cabin area, from the front to the rear of the cabin, while thermal injury was noted in many of the passengers who are estimated to have been seated behind the main wing where the fire started.

3.1.9 Detection of Ethanol in Remains of CAP and F/O

The remains of CAP, F/O, and purser were stored in the hangar after the accident, and underwent autopsies at different colleges. During the autopsies, samples were collected from the remains, and taken to the Scientific Investigation Labs of Aichi Prefectural Police Headquarters, where they were stored in refrigeration. On the following day, alcohol reaction tests were performed at the Labs. The results are summarized below.
Time elapsed after death and before sample was taken SampleEthanol concentration
CAPApprox. 24 to 25 hoursPleural fluid13 mg/ 100 ml
F/OApprox. 19 hoursPleural fluid55 mg/ 100 ml
PurserApprox. 19 hoursBlood in heartNot detected

Ethanol was detected in the samples from the remains of the CAP and the F/O, which is considered to be due to one or more of the following three causes.

It is also considered that two or three of the following causes were combined.
(1)Post-mortem ethanol production
(2)Alcohol ingestion before death
(3)Mixture with alcoholic drinks on board

Each of the three causes were analyzed as described in the following sections.

3.1.9.1 Possibility of Post-Mortem Ethanol Production

According to legal medical documents, there are confirmed cases where ethanol has been

3-25


detected in the body of a person who never ingested any alcohol before death. It is understood that progress in decomposition of the body after death results in microbial fermentation, which produces ethanol.

There is a possibility that the ethanol detected in the remains of the CAP and the F/O was due to a post-mortem ethanol production for the following reasons:
(1) The remains of the cockpit crew were stored in a hangar after the accident, and approximately 18 to 22 hours elapsed before they underwent autopsies. During this period, special measures, such as placing the remains in refrigerated storage, were not taken.

The lowest and highest atmospheric temperatures at Nagoya Airport area during this period were approximately 1 0°C and 23°C, respectively. The temperatures in the hangar where the remains were kept is assumed to have been somewhat higher than the above-mentioned temperatures.

Furthermore, it was noted that the remains of CAP and F/O had deep open wounds. In particular, the body of F/O was significantly damaged.

The long time interval from occurrence of the accident to the autopsies, the environmental temperatures, and the existence of open wounds are considered to satisfy the conditions for post-mortem ethanol production.
(2) The concentrations of ethanol detected in the bodies of the CAP and the F/O were 13 mg/1 00 ml and 55 mg/i 00 ml respectively; these concentrations are considered to be comparable with those normally detected as a result of post-mortem production.
(3)Regarding the difference in concentration between the ethanol found in the body of the CAP and that in the body of the F/O, when the difference in temperature resulting from the difference in location within the same hangar and the difference in severity of open wounds between them are taken into account, the variation in concentration is considered to be within a conceivable range. On the other hand, ethanol was not detected in the pursers body which had a few open wounds.

3.1.9.2 Possibility of Alcoholic Ingestion before Death

Collected samples from the bodies of the CAP and the F/O in this case consisted only of pleural fluids, probably because damage to the bodies was extremely extensive. If the crew drank any alcohol before death, ethanol may have been detected in their pleural fluids (as ethanol in the blood is absorbed into other internal organs some time after alcohol is ingested). From the fact that the open wounds of the bodies of both the CAP and F/O were extensive, it is considered possible that blood became mixed with their pleural fluids.

If this is the case, the concentration of ethanol detected in the samples should be the concentration of ethanol produced post-mortem described in section 3.1.9.1 plus the concentration of ethanol resulting from alcoholic ingestion, but the possibility of alcohol ingestion before death and the extent thereof could not be determined.

3-26


3.1.9.3 Possibility of Mixture of Alcohol from Scattered Liquor Containers

Adjacent to the cockpit and behind the F/O's seat, a galley (No.1 galley, see attached Figure 26) was located in which liquors (about 40 bottles of whiskey and other liquors) were stored for passenger service. Also on board were bottles of alcoholic liquor carried by passengers.

It seems possible that destruction of the partition between the cockpit and galley by the impact caused liquors and other alcoholic drinks from broken bottles to be scattered over the bodies of the crew (who had open wounds). However, it could not be confirmed.

3.1.10 China Airlines' Operation and Training Rules and Handling of Service Bulletins

3.1.10.1 Operation

China Airlines has an Operations Policy Manual and an Air Crew Manning and Dispatch Manual that were prepared according to the requirements stipulated by the Taiwanese civil aviation authorities, and operate their aircraft in compliance with those manuals.
(1)Qualifications for Flight Crew

According to an Operations Policy Manual, the flight crew shall consist of personnel of good character, clear-cut features, and sufficient technical knowledge, and shall in addition satis~ the following conditions.
It is recognized that the CAP and F/O of the aircraft held valid licenses and satisfied the required number of hours' flight experience.
(1) CAP
a.
Shall have an ATR license (airline transport pilot license) and a type rating certificate issued by the Taiwanese civil aviation authorities and an airman worker card (identification card) issued by China Airlines.
b.
Shall have a valid airman medical certificate issued by the Aviation Medical Center.
c.
Shall have not less than 1,000 hours' flight experience with China Airlines.
d.
Shall have not less than 3,500 hours' flight experience overall.
e.
Shall have passed the captain training tests.
(2) F/O
a.
Shall have an SCP license (commercial pilot license) and a type rating certificate issued by the Taiwanese civil aviation authorities and an airman worker card (identification card) issued by China Airlines.
b.
Shall have a valid airman medical certificate issued by the Aviation Medical Center.
c.
Shall have passed the flight officer training tests.
(2)Requirements for Flying Aircraft by F/O in Revenue Flights
According to "AIR CREW MANNING AND DISPATCH MANUAL", requirements for

3-27


flying aircraft by F/O in revenue flights are stipulated as follows. (item (1) through (4) )
(1) China Airlines shall have F/Os fly a certain aircraft type in take-off and landing phases at least three times every three months to maintain their flying skills.
(2) When an F/O is to control an aircraft type in a revenue flight, the F/O shall be seated in the right seat.
(3) When an F/O is to control an aircraft type in a revenue flight, the CAP shall strictly supervise the F/O's operation, shall assume all responsibilities for safety, and shall observe the following:
a.
CAP shall have not less than 1,000 hours' flight experience on the aircraft type concerned.
b.
The weather conditions at the origin and destination airports shall be VMC, the runway shall be in good and dry condition and be sufficiently long, and the cross wind velocity shall be not more than 15 kts.
c.
Both take-off and landing weights shall be not more than 85 % of the maximum design weights of the aircraft type concerned.
d.
When the F/O performs operations in critical phases of flight such as take-off climb, and approach and landing, the CAP shall always strictly supervise the operation and keep his hands and feet in position on the control wheel, rudder pedal, and thrust lever. In cases of an abnormal or emergency condition, the CAP shall immediately take controls and call, "I HAVE CONTROLS", and F/O shall call, "ROGER".

(4) When flying, if 1) the weather conditions do not meet VMC criteria, 2) a fault occurs in the aircraft, 3) a clearance given by ATC is inappropriate, 4) the F/O carries out a procedure that exceeds the safety limits of the aircraft, or 5) if an emergency arises, the CAP, in the interest of safety, shall make a quick decision to take control.
It is recognized that items O,(c) and (c)-a, b and c above were satisfied in this case, i.e., that the F/O was flying the aircraft in revenue flight.
Considering item (c) above, whether the CAP kept his hands and feet in the appropriate positions during the approach and landing phases could not be determined. However, from the CVR record, it is recognized that he took over the controls to deal with the abnormal situation, saying "I have got it" at 1115:03.
Concerning item (4) above, as described in Paragraph 3.1 .2.2.(12), it is considered that the CAP's judgment situational awareness was inadequate, and that he was delayed in taking over the controls.

(3)Utilization of Operation Technical Reports

China Airlines used the technical report on the incident of the A3 10 aircraft that had occurred on February 11, 1991 in Moscow for the training of the crew members concerned. However, there are no records indicating that they used the reports on other cases for training purposes. The CAP and F/O involved in the accident did not attend the training that utilized the above technical report, as at the time they were not yet been assigned to A300-600R aircraft.

3-28


3.1.10.2 Training

(1) Training

China Airlines has established a training program according to the requirements stipulated by Taiwanese civil aviation authorities to conduct the following training courses in the classroom, simulator, and aircraft.

The CAP and F/O are recognized to have finished the training established by China Airlines. In the case of the A300-600R aircraft training program, the company basically employs the syllabus established by Airbus Industrie, the manufacturers of the aircraft, as their syllabus with the training standard times set themselves.
(1) New qualification acquisition training
Training performed to give experience, knowledge, and skill required for obtaining flight crew qualifications.
(2) Promotion training Training performed to give experience, knowledge, and skill required for obtaining higher level flight crew qualifications.
(3) Type transfer training
Training performed for flight crew who are to serve in an aircraft of a type different from the one in which they are currently serving or in which they served in the past to give experience, knowledge, and skill required for obtaining identical class flight crew qualifications.
(3) Periodic training
Training performed periodically for flight crew to maintain and improve their knowledge and skill.

(2) Simulator training
(1) China Airlines did not have simulators for the A300-600R aircraft. Therefore, they conducted simulator training for the said aircraft type by using simulators owned by Thai International Airways of Thailand and Aeroformation of France (China Airlines contracted part of pilot training to Airbus Industrie, and Airbus Industrie subcontracted this to Aeroformation.).
(2) The CAP underwent simulator training for the A300-600 aircraft type in the Thai International Airways' simulator approved by the Thai aviation authorities. The F/O underwent this same training for the said aircraft type in the Aeroformation simulator approved by the French aviation authorities, and periodically in the Thai International Airways' simulator.
(3) The simulator training manual used by China Airlines was prepared by Aeroformation of France. However, it had not been updated.

3-29


(4) When the F/O underwent simulator training in October through November, 1992 in the Aeroformation simulator, a item of" GO-AROUND DEMONSTRATE AP MISUSE IN GO-AROUND "was included in its check sheet which the instructor used.
A mark "+" was placed in the box indicating completion of the item. Airbus Industrie said that the item was added after the incident which had occurred in Moscow airport in February 1991.
However a check sheet, which the F/O was previously given as a part of training materials by Aeroformation before training, was not yet revised and did not contain the above item.
But how the F/O underwent training for the item could not be clarified.
When the CAP underwent the training in June through July, 1992 in the Thai International Airways' simulator, the check sheet, which China Airlines had obtained, was used. But this check sheet was not revised and did not contain the above item.
(5) Concerning that Airbus Indstrie did not provide China Airline with up-to-date training materials, it is considered that the agreement on dealing with up-to-date training materials was not made clearly between the two companies which had contracted the crew training.
(6) According to French and Taiwanese persons concerned, the Thai International Airways simulator does not simulate the AP overriding function in GO AROUND mode for the A300-600R aircraft's AP, but whether this had any bearing on the accident could not be determined.

(3) AFS Training
(1) The descriptions in FCOM for the AFS are not easy for crews to understand.
(2) The crew was not given sufficient technical information with regard to similar incidents.
(3)Up-to-date training materials were not properly obtained.
(4) CVR transcripts show that crew understanding of the AFS was probably not sufficient.

From the above items it is concluded that the training required to understand the sophisticated and complicated AFS was insufficient.

3.1.10.3 Handling of Service Bulletins

Service Bulletins (hereinafter referred to as "SBs") are issued by the manufacturers to notify each operator of the inspection and modification to aircraft and their equipment.

SBs are generally divided into four compliance categories: Mandatory, Recommended, Desirable, and Optional. Upon receipt of an SB, operators, referring to the compliance category described on the SB, determine whether or not it applies to their airplanes and, if so, how they should implement it.

3-30


Mandatory SBs are usually implemented on earliest possible occasion. If the SB is "Recommended" or any of the remaining categories, operators plan to implement it on the most suitable occasion, taking into consideration their operational experience, maintenance schedules, and type of operation.

In China Airlines, the Maintenance Headquarters first receive SBs, and then the engineers belonging to the Chief Engineer Office determine the way to implement each of the SBs as well as the applicable airplane numbers after evaluating and examining it. The determined results are entered in a form called TIPS (Technical Instruction Processing Sheet) which is then forwarded to the Department of Maintenance Control after being examined by the Department of Quality Assurance. The SBs are implemented under the supervision of the Department of Maintenance Control.

SB A300-22-6021 issued by Airbus Industrie dated June 24, 1993 with compliance "Recommended" specified, concerned a modification to the AFS, which disengages the AP when a force in excess of 15kgf is applied to the control wheel in pitch axis during a flight in the GO AROUND mode above radio altitude 400 ft (See appendix 2.). To implement this SB, it was necessary to modify the two FCCs on each aircraft to which it applied.

According to China Airlines, the actions they took after receipt of the SB were as follows:

China Airlines received this SB on July 29, 1993. The SB B470AAM-22-007 of Sextant Avionique, the manufacturer of FCCs, which is specified in the Airbus SB, was issued on July 12, 1993. China Airlines, after receiving the Sextant's SB, issued on September 1, 1993, a TIPS (A300-6153y which contained instructions on handling of the Airbus SB.

Since the compliance category of the SB A300-22-602 1 was "Recommended", China Airlines judged its implementation not urgent, and decided to carry out the modification at the time when FCCs needed repair.

As of August 1993, China Airlines possessed six (6) Airbus airplanes of the same type and the number of spare FCCs in stock was six (6). The number of the FCCs removed from the company as a result of failure, and sent for repair to Sextant Avionique Asia PTE Ltd. (hereinafter referred to as "Sextant Singapore"), was four (4) in 1991, eight (8) in 1992, and one (1) as of this time in 1993.

Since the maintenance facility of China Airlines was not sufficiently equipped for implementing the modification specified in the SB, China Airlines had to send their FCCs to Sextant Singapore which can carry out the modification. Consequently, China Airlines planned to carry out the modification at the same time as the repair of those FCCs removed owing to failure, while taking into consideration the time required for the modification as well as keeping spare FCCs necessary to maintain operations.

3-31


However, no FCCs were removed from China Airlines' aircraft as a result of failure, and sent to Sextant Singapore for repair in the period between June 24, 1993 when the SB was issued and April 26, 1994 when the accident occurred at Nagoya Airport.
According to Sextant Avionique, the implementation of the modification of FCCs had been arranged as follows:

The repair facilities of Sextant Avionique were located in France, America and Singapore.

The SB A300-22-6021 was not "Mandatory" when it was issued. The system to make an acceptance of the modification available was established in September, 1993. But it was in December, 1993 that Sextant Singapore started the modification job at the request of airlines.
Likewise, the Sextant's repair facilities in France and America started the modification of the FCC in April, 1994.

At the beginning as stated before, China Airlines adopted the SB A300-22-602 1, but planned to accomplish the modifications at the time when FCCs needed repair (because the modification was not considered urgent). Since no FCCs had been sent to Sextant Singapore for repair before the Nagoya accident, the modifications in accordance with the SB A300-22-6021 were not made.

3.1.11 Automatic Flight System (See Appendix 1)

3.1.11.1 Pitch control system

Concerning the A300-600, it was possible for a pilot to override the elevators while the AP was controlling the THS in GO AROUND mode and LAND mode. Therefore, two control inputs for two different objectives could be allowed simultaneously in the pitch axis. The aircraft was not equipped with a warning device which would alert the pilot to two simultaneous control inputs. Such a design might have contributed to the accident as one of the factors of the abnormal out-of-trim.

3.1.11.2 AFS Operation

In order to verify the status of the THS, a computer simulation was conducted to demonstrate THS movement using the recorded parameters in DFDR such as airspeed, attitude, etc. The analysis revealed that the calculated THS movement history showed good correlation with the recorded THS data. From the analysis, it is recognized that FAC and FCC were functioning normally per design concerning with THS movement.

3.1.11.3 Modification to AFS (See Appendix 3 and 4)

3-32


(1) With regard to the A300-600 aircraft incident of March 1, 1985 that involved an out-of-trim condition triggered by the switching of AP mode to ALT HOLD mode, Airbus Industrie established Mod.7 187 in order to prevent the recurrence of similar incidents, and on March 18, 1988 it was approved by DGAC.
The Mod. 7187 was to add a function to allow AP disengagement by applying a 1 5Kgf force on the control wheel in pitch axis in modes except LAND track (below 400ft radio altitude) and GO AROUND mode.
After that, Mod. 7187 was included in SB A300-22-6009 dated June 1, 1989, but the SB did not contain any mention of the Mod.71 87.
After that, in view of further incidents which involving out-of-trim conditions triggered by AP mode switching to GO AROUND mode -- on the A300-B4-203FF aircraft at Helsinki Airport on January 9, 1989 and on the A3 10 aircraft at Moscow Airport on February 11, 1991 -- Airbus Industrie issued SB A300-22-6021 dated June 24, 1993 which recommended operators to accomplish a modification to the APS, as a measure against recurrence of similar incidents, namely to introduce a function that disengages the AP when a force greater than 15kgf is applied on the control wheel in pitch axis at a radio altitude higher than 400 ft in GO AROUND mode.
(2) Although the causes which triggered the above incidents are different, all the incidents were similar in that the operation of the control wheel by the crew and operation of the AFS conflicted with each other, the THS ended up in an out-of-trim condition, and the crew had to deal with a rapidly changing aircraft attitude, without having time to grasp the full extent of the situation.
Such serious incidents occurred in March 1985, January 1989, and in February 1991 respectively.
Airbus Industrie informed operators of the summary of the these incidents, but did not present a systematic explanation on the technical background sufficiently.
(3) As described in (1), three to four years elapsed before, in response to the incidents, Airbus Industrie introduced the modifications to the AFS. Considering the importance of the incidents, it is considered that the modifications were not introduced promptly enough.
The system to make acceptance of modification available to operators was completed at the FCC manufacturer in September, 1993 after Airbus Industrie issued the SB on June 24, 1993, as mentioned in section 3.1.10.3.
(4) The China Airlines' A300-622R B-1 816 aircraft which crashed at Nagoya Airport on April 26, 1994, had incorporated Mod.7 187 when manufactured, and implementation of SB A300-22-602 1 was planned. However this had not been done by the time the accident occurred.
It is considered that in the accident at Nagoya Airport, activation of the GO lever by the crew changed the mode to GO-AROUND mode, and since the crew later engaged the APs, the control wheel push-down operation by the crew, who seemed to have intended to continue approach, conflicted with the motion of the THS controlled by the AFS, resulting in an abnormal out-of4rim condition. If the modification prescribed in SB A300-22-602 1 had been incorporated, it is considered that the APs would have been disengaged under a

3-33


force greater than 15kgf applied at the control wheel in the nose-down direction, preventing the aircraft from entering such an abnormal situation.
However, China Airlines who received the SB judged its accomplishment not urgent and decided to implement the modification on an occasion when their FCC(s) needed repair.
Since operators had hardly grasped and understood the technical background and detailed information with regard to the three serious incidents described in the items of (1) and (2) above, it is considered that this decision was affected considerably by the fact that the SB was issued as "Recommended", and not "Mandatory", and the reasons and technical background for issuance of the SB were not explained clearly and in detail.
(5) In view of the significance of those three incidents, it is considered proper that the French airworthiness authorities pertaining to aircraft design and manufacture, at an earlier stage, should have urged Airbus Industrie to establish the modification promptly to preclude the recurrence of similar incidents, and issued an airworthiness directive so that each operator could implement promptly the SB pertaining to the modification. It is also considered necessary for the French airworthiness authorities to have requested Airbus Industrie to provide each operator with technical information describing each incident systematically.

3.1.11.4 FCOM of A300-600 Type Aircraft

(1) Revision of FCOM Based on Mod.7187 (See Appendix 2-2 and 2-3)

After the incident involving an A300-600 aircraft on March 1, 1985, Mod.7187 (rearranged into SB A300-22-6009 in June 1989) was established on March 18, 1988, introducing a function allowing AP disengagement in pitch axis in modes except LAND track (below 400ft radio altitude ) and GO AROUND mode.
The condition of "LAND mode" was not clearly described in the FCOM issued in June 1988. It is ambiguous whether this meant a phase after LAND mode was selected on the FCU or a phase when "LAND mode" is displayed on FMA (LAND track mode). However, when SB A300-22-602 1 dated June 24, 1993 (addition of function of allowing a pilot input in pitch axis to disengage the AP above 400ft radio altitude in GO AROUND mode) was implemented, the description in the FCOM concerning the above mentioned "LAND mode" was revised and clarified such that LAND mode meant a phase when "LAND mode" is displayed on the FMA.
(2)Addition of CAUTION (See Appendix 2-2 and 2-4)
After the incident involving an A300B4-203FF aircraft at Helsinki Airport on January 9, 1989, FCOM 1.03.64 P 3/4 and 2.02.03 P 1 were revised in January, 1991, adding a CAUTION against a hazardous out-of4rim condition that may lead the hazardous situation if the AP is overridden in pitch direction during the LAND and GO-AROUND modes.
The FCOM describes that this override was concerned in order to protect the pilot against AP abnormal behavior. On the other hand the CAUTION in the FCOM prescribes that pilots are prohibited from overriding the AP when it operates normally.

3-34


Therefore, pilots may be confused and may receive two contradictory meanings, such as recommendation and prohibition. For example, if a pilot had suspected that the APs were malfunctioning, he might miss taking proper action because no criterion had been written as to the situations in which one should override. Accordingly, the technical information, examples of possible situations, and the corresponding confirmation and operation procedures should be written in the FCQM systematically in order to encourage crews' further understanding of the AP overriding function.
(3) FCOM 1.03.67 P-8 REV 17 (A3 1 0/A300-600) (See Appendix 2-3)

(1) The FCOM section dealing with GO AROUND mode disengagement procedure is as follows:

* When a longitudinal mode is engaged(V/S, ALT, LVL/CH, ALT*, or PROFILE mode); - GO-AROUND mode disengages, and - HDG mode engages as lateral mode.

* When a lateral mode is engaged (HDG SEL, VOR CAPTURE or TRACK phase, NAV CAPTURE, or TRACK phase); the GO-AROUND mode disengages. However, the SRS mode, a longitudinal mode of GO-AROUND, remains active.

(2) According to the procedure in FCOM, if V/S mode is selected as a longitudinal mode when GO-AROUND mode is active, "GO AROUND" (a common mode) disappears from the FMA display and "V/S" and "HDG" (a lateral mode) both appear instead. Actually, the GO AROUND function is not completely disengaged, even though it appears from the FMA display that it has.
In order to completely disengage GO-AROUND mode, the pilot is required to select another lateral mode (HDG SEL, VI, or NAV mode), making the FMA display change from "HDG" to the display corresponding to the selected lateral mode.
For example, if HDG SEL mode is selected, the FMA display changes from "HDG" to "HDG/S" and GO-AROUND mode is completely disengaged.
GO-AROUND mode is a common mode that combines both longitudinal and lateral modes, and is disengaged completely when both lateral and longitudinal modes are changed.
The FCOM description, however, does not mention that GO-AROUND mode does not disengage completely when only a longitudinal mode is selected. Readers, therefore, are misinformed as to the precise relationship between how the various modes are selected, how they are displayed on FMA, and how they actually work.

(4) Notice to each operator issued by Airbus Industrie after the Accident in Nagoya
After the accident involving a CAL Flight 140 at Nagoya Airport, which occurred on April 26, 1994, Airbus faxed all A300/3 10 and A300-600 aircraft operators cautionary information to be applied when a pilot moves the elevators in conflict with the APs while

3-35


the APs are in LAND or GO-AROUND mode (See Attachment 2-6).
In view of this information, Airbus Industrie recommend that the best way to disengage the GO-AROUND mode is to disengage the APs by means of the AP instinctive disconnect button, or to select another mode, and the subject information also states that if the AP is then disengaged, the aircraft is left in an out-of-trim situation which might be hazardous if not trimmed back.
This technical information describes the basic AP override function, examples of operations leading to an out-of-trim situation and the measures to be taken if this situation arises. It is considered that such cautions, which are specified definitely and clearly therein, need to be promptly included in adequate chapter or subject in FCOM.

3.1.11.5 AP Override Function

When the AP is in CMD, the AP actuators move the roll, pitch, and yaw control surfaces in response to commands from the FCC. The THS moves according to commands from the FAC.

The override function mechanically disconnects the AP actuator from the control surface and allows the pilot to manually control the aircraft by applying a force greater than a threshold on the control wheel or rudder pedals. When the force applied to the control wheel or rudder pedal is released, the AP actuators are reconnected to the corresponding control surfaces. However, the THS remains under control of the AP even while the AP is being overridden and continues to operate as commanded by the AP.

Airbus Industrie define the AP override function in their FCOM and FCOM Bulletin as a safety device to allow the flight crew to regain control from the APs in the event of AP anomalies. Airbus Industrie also recommend in the above bulletin that the pilot should disconnect the APs immediately, upon suspicion of any abnormal aircraft behavior when AP is in CMI) (See Appendix 2-4 and 4.).

When the AP is in CMD, if the pilot overrides the AP's pitch command for some reason, the AP activates the auto-trim function and moves the THS so as to maintain the aircraft on the scheduled flight path. If the pilot disengages the AP without noticing this, the aircraft is left in an out-of-trim situation which might be hazardous if not trimmed back (See Appendix 2-6.).

In the case of a post SB A300-22-6021 aircraft, if the crew carry out an operation to hold the control wheel in an effort to decrease an excessive pitch angle at a radio altitude lower than 400 ft during a go-around started from a low altitude, the result will be the same as an AP override operation, causing the THS to move toward the nose-up direction. If the crew is not aware of the THS movement and does not make a trim-back operation, the aircraft could enter an out-of-trim situation, which is potentially hazardous.

3.1.11.6 Operation of Alpha Floor Function

3-36


As the aircraft continued descent in an out-of-trim condition, the pitch angle and the angle of attack (AOA) increased. The AOA exceeded the threshold angle of 11.5 degrees, corresponding to SLATS/FLAPS 30/40, while crossing approximately 550 ft pressure altitude at 14:57. The alpha floor function was activated, increasing power. Although, immediately after this, the thrust was reduced for a while, the go-around thrust was set again, and a rapid increase of the pitch angle continued.

In the case of this accident, the Alpha Floor function -- a safety device which is designed essentially to prevent stall and to protect aircraft within the flight envelop -- activated and increased the engine thrust when the THS was in full nose-up position. This generated a pitch up moment. Immediately after this, although the thrust was manually reduced for a short period, the pitch angle increased by 9.5°, up to 18.0°. It is considered that the automatic increase of the thrust and the accompanying increase in pitch angle resulted in a narrowing range of selection for subsequent recovery operations, and a reduction in the time allowance for such operations.

3.1.11.7 Alpha Trim Function

It is considered probable that, after the CAP called "GO lever" at 1115:11, the alpha trim function came into operation about 15:21. The THS continued to move, and by 15:27 had reached -7.4° from -l0.9° (where it had been at 15:21). Movement then appears to have stopped owing to the pitch-trim tripping.

Tripping of the pitch-trim is considered to occur when one of a certain number of conditions is met, such as when the AOA cannot be calculated correctly owing to low speed and unstable aircraft attitude. The alpha4rim function is designed to stabilize the longitudinal aircraft attitude by trimming the THS automatically in the nose-down direction (maximum 40 nose down), in conditions of high pitch and low speed.

3.1.11.8 THS-In-Motion Warning I Recognition Function

The A300-600 is equipped with the following systems, for the purpose of THS motion awareness.

(1) Visual trim indicator: two indicators located on the center pedestal, on which the current position of the THS is displayed.
(2) Manual pitch trim control wheels: two wheels with white strips located on both sides of the center pedestal. They turn according to the THS motion.
(3) THS motion warning: continuous trimming by means of the electric pitch trim switch activates an aural warning (whooler).

In this event of the accident involving the THS, systems (1) and (2) above are not always in pilots' field of view, and cannot alert them actively to the THS motion. Moreover this accident occurred in night and the cockpit was dark, so it is considered that these two systems did not provide pilots with effective information as to the status of THS-movement.

3-37


Furthermore, system (3) above was not active during the approach phase of this flight, because the APs were engaged in CMI).

During the development stage of the A300-600 aircraft, the motion warning had been designed to provide an aural warning "Whooler" when THS motion occurred in either automatic or manual flight, but the warning function was later eliminated from automatic flight by a design change.

On this matter, from a statement by the British aircraft accident investigation authorities the circumstance is as follows;

The UK CAA in its evaluation summary for this aircraft type required that "Excessive operation of the trim in motion warning which occurs when trim is being applied at the high rate (flap extended) during autoland flare should be reduced."
Airbus Industrie chose to address this issue by deleting the trim in motion warning completely when the autopilot was in a command mode.

It is considered that, if the THS-in-motion warning had sounded continuously during an automatic flight, the crew would have recognized a significant change in flight configuration or suspected some anomaly in the AFS, and confirmed the operating conditions of the system.

A characteristic of the AP override function of A300-600 type aircraft is that a prolonged override of the AP acting on the pitch axis via the control wheel leads to an out-of-trim situation. Accordingly it is considered necessary for Airbus Industrie to have maintained the function of THS motion warning in the AP CMI) or, if eliminated, to establish another warning and recognition function which alerts pilots directly and positively to know the condition of the THS.

3.1.12 Fire Fighting and Rescue

3.1.12.1 Fire Fighting and Rescue Service System

The fire fighting and rescue services at airports are specified in "Level of Protection to Be Provided" of Annex 14 "Aerodrome" to the Convention on International Civil Aviation and in the "Airport Services Manual" (hereinafter referred to as "ICAO Level") pertaining to it. Incidentally, the stipulations contained therein are considered desirable for safety, exactitude, and efficiency, and are categorized as "Recommended Practices", but not "Standards" at the present stage. In Japan, however, the fire fighting and rescue service is understood in principle to conform with the "ICAO Level", deploying and operating required vehicles etc. accordingly.

Nagoya Airport is managed by Nagoya Airport Office. The airport is used for international scheduled flights. Next to the airport is Komaki Base, Japan Air Self-Defense Force.

The Nagoya Airport Office is in charge of fire fighting and rescue services for civil aircraft.

3-38


At the time of the accident, the office had an emergency medical services transport vehicle on standby which was loaded with medical supplies. The office did not have chemical fire vehicles and other fire/rescue vehicles on standby at that time, but was in the process of equipping itself with these vehicles and other necessary equipment. Even though equipment acquisition was not complete, the office was capable of providing fire fighting and rescue services that conformed to "Level of Protection to Be Provided" for Category 9 airports recommended in Annex 14 to the Convention on International Civil Aviation, by utilizing fire fighting and rescue vehicles assigned to Komaki Air Base, Air Self-Defense Force, based on an agreement with them. The fire extinguishing foam solution discharge rates, however, did not completely satisfy the specified level.

Additionally, agreements were signed with neighboring fire fighting organizations, permitting utilization of their support. The fire fighting and rescue standards at Nagoya Airport were as follows:

(1) Airport Category

Airports are ranked in categories from Category 1 to 9, based on the length and width of the fuselage of the largest aircraft which usually use the airport. Nagoya Airport is ranked Category 9, for Boeing 747 used the airport frequently.

(2) Required Amounts and Discharge Rates of Fire Extinguishing Foam Solution

The Airport Services Manual uses the concept of "critical area" as the basis for calculating the necessary amount and discharge rate of fire extinguishing foam solution, seeking to control only that area of fire adjacent to the fuselage of the aircraft involved in an accident, the area being determined with the overall length of the fuselage taken into consideration. (See Appendix 9.)
(1) Required Amount of Fire Extinguishing Agents
According to the ICAO Level, the minimum requirements for fire-extinguishing agents are 24,300 liters for water for foam production and 450kg for auxiliary fire-extinguishing agent. The total amount of water for foam production, including water available from the six fire fighting and rescue vehicles (five chemical fire vehicles and a water supply vehicle) assigned at Komaki Air Base, Air Self-Defense Force, was 33,600 liters and that of the auxiliary fire-extinguishing agent was 550kg, conforming with the amounts specified in the ICAO Level.
(2) Discharge Rates
According to the ICAO Level, the minimum foam solution discharge rate is 9,000 liters/mm. The discharge rate of the five chemical fire vehicles was 7,500 liters/mm (1,500 liters/mm x 5) which was considered short of the rate recommended by the ICAO

3-39


                    Level.
This was due to the low discharge rate of each chemical fire vehicle.
The chemical fire vehicles assigned to Komaki Air Base had a discharge distance of approximately 30m.

(3) Response Time

According to the ICAO Level, the target time from when first notification of an accident is made to the rescue and fire fighting detachment until the rescue and fire vehicles arrive at the runway end, and actual activities are started, should not exceed three minutes. At the start of their fire fighting activities, the fire vehicles are required to discharge fire-extinguishing foam solution at a rate of at least 50% (4,500 liters/mm) of the specified 9,000 liters/mm rate; vehicles following the first ones should arrive subsequently at the crash site within one minute.
In the case of the accident, the ICAO Level was conformed to, since the rescue and fire vehicles arrived at the runway end within three minutes, and in the initial fire extinguishing activity the discharge rate of the preceding three vehicles was in excess of 4,500 liters/mm.

(4) Numbers of Fire Fighting and Rescue Vehicles and Personnel

According to the ICAO Level, at least three fire fighting and rescue vehicles are required. Assigned to Komaki Air Base, Air Self-Defense Force were five chemical fire vehicles loaded with rescue materials and one water supply vehicle. The number of these vehicles was in excess of the number specified in the ICAO Level.
An ordinary fire vehicle was also assigned to Komaki Air Base, Air Self-Defense Force. According to the ICAO Level, persons to be counted as fire fighting and rescue activity personnel are those who can get into rescue and fire vehicles immediately after an accident and operate the relevant equipment with maximum competence.
The number of appropriate personnel on duty at Komaki Air Base, Air Self-Defense Force at the time of the accident was six.

3.1.12.2 Fire Fighting and Rescue Training for Aircraft Accident

Fire fighting and rescue training for aircraft accidents are required to be conducted periodically, in order to maintain the competence of personnel who are to engage in fire fighting and rescue activities in the event of an emergency. Annex 14 "Aerodrome" to the Convention on International Civil Aviation has a "Standard" which requires all organizations concerned with fire fighting and rescue services to conduct pertinent training at intervals of less than two years. On May 24, 1993, organizations such as Air Self-Defense Force, neighboring fire fighting organizations, Airport Police and Medical Association of Aichi Prefecture, etc. participated in the fire fighting and rescue training for aircraft accidents under the auspices of Nagoya Airport Office.

3.1.12.3 Fire Fighting and Rescue Activities (Times are JST.)

3-40


Fire fighting and rescue activities were conducted as follows:

(1)
As for the response time of the fire fighting and rescue services, the fire platoon was notified of the occurrence of the crash at approximately 2016; the first group of three chemical fire vehicles (10,800 liters) was mobilized at approximately 2017, and commenced fire fighting and rescue activities immediately on arrival at the crash site, at approximately 2019.
However, the three succeeding vehicles, consisting of two chemical fire vehicles and one water supply vehicle, (22,800 liters in total) which were driven by the fire fighters who responded to the emergency call, arrived at the crash site at approximately 2030, a little later than the first group.
The total discharge rate of the first three chemical fire vehicles was 4,500 liters/mm, and even after arrival of the second two chemical fire vehicles, the discharge rate was not more than 7,500 liters/mm.
It is inferred therefore that the necessary amount of foam solution was not discharged at the initial stage of the fire fighting activity.
(2)
It is acknowledged that two ambulances from Komaki Air Base, Air Self-Defense Force arrived at the crash site at approximately 2019 and 2023, and mobile cranes, a light wrecker, etc., driven by personnel who responded to the emergency call, arrived at the crash site, and joined the rescue activities at approximately 2030.
(3)
It is acknowledged that fire vehicles, ambulances, etc. dispatched by neighboring fire fighting organizations, police, and Aichi Medical Association at the request of the Airport Office, began arriving one after another at the crash site from approximately 2027, and commenced rescue services such as confirmation of survivors, first aid and to transport of the injured.
(4)
It is recognized that the Aviation Safety Association controlled the traffic of emergency vehicles of the neighboring fire fighting organizations, guarded against entry of unauthorized personnel at the No.2 West Gate after the accident, had an emergency medical supply vehicle on standby with its engine running in front of their office at approximately 2030, and, on request from the neighboring fire fighting organization, at approximately 2115 sent the vehicle to the crash site, where it arrived at approximately 2122.
It is recognized that at the time of the crash, a large quantity of fuel (approximately 22,000 lbs) remained on board. It spilled around the airframe at the impact site, and ignited almost immediately. The fire propagated around the airframe, covering an area of approximately 70 square meters. Owing to wreckage scattered at the crash site well beyond the fire area, along with the irrigation water channel and revetment, access by the chemical fire vehicles to the crash site was restricted.
It is considered that the ICAO Level had been stipulated for the case of a fire breaking out in an aircraft which had been moderately destroyed and grounded on the runway. With regard to this accident, it is considered that fire fighting and rescue activities had been hampered by the above-mentioned conditions and, furthermore, by the fact that it occurred at night.

3-41


3.2 Summary of Analysis

3.2.1 General

3.2.1.1

The flight crew had valid airmen proficiency certificates and valid airman medical certificates.

3.2.1.2

The aircraft had a valid airworthiness certificate and had undergone maintenance and inspection as specified.

3.2.1.3

From the results of the investigation, it is estimated that the aircraft did not suffer any failures or malfunctions that contributed to the accident.

3.2.1.4

It is estimated that the weather around the time of the accident had not contributed to it.

3.2.2 Flight sequence of the Aircraft

The flight sequence is estimated to have been as follows:
(1)
While the aircraft was on ILS approach to Runway 34 of Nagoya Airport, under manual control by the F/O(PF), the F/O inadvertently triggered the GO lever.
(2)
This led the FD to change to GO AROUND mode, and thrust increased. The F/O(PF) applied nose-down input to the control wheel and reduced the thrust which was increasing, but these actions did not have sufficient effect. Consequently the aircraft deviated above its glide path and then leveled off
(3)
The CAP (PNF) was most likely to have instructed the F/O(PF) to disengage GO AROUND mode.
However, the crew did not perform an adequate operation to change GO AROUND mode into LAND mode. Consequently the GO AROUND mode was not disengaged.
(4)
There is a possibility that the AP was engaged either by the CAP himself, by the F/O(PF) in accordance with the CAP's (PNF) instructions, or by the F/O without the CAP's consent, (or without notifying the CAP).
(5)
The F/O(PF) continued pushing the control wheel forward, in spite of its strong resistive force, in an attempt to recover the normal glide path above which the aircraft had deviated. He did so in accordance with the CAP's (PNF) instructions. The THS moved to its full nose-up position, leading to the abnormal out-of-trim condition.
(6)
Subsequently the APs were disengaged, but the out-of-trim situation still remained.
(7)
The AOA increased and the alpha-floor function was activated. This led to a thrust increase. As a result, a large pitch-up moment was generated because THS was still in

3-42


out-of-trim condition at this point of time.
(8)
At the time when the CAP took over the controls after being informed by the F/O that the thrust was latched, the CAP (PF) still seemed to intend to continue approach. Subsequently the CAP retarded the thrust levers once, but the nose-up tendency was strong and the aircraft led to a condition from which it was unable to recover its normal glide path. So the CAP probably decided to go around.
(9)
Because thrust was increased for go around and a flap-up operation was performed, the aircraft climbed steeply with the pitch angle increasing. Consequently speed decreased and the aircraft stalled.

3.2.3 Control and Operation by the Crew

(1)
It is considered that the decision by the CAP and the F/O to change from GO AROUND mode to LAND mode, as well as their subsequent actions to do so, was due to their inadequate understanding of the aircraft AFS.
(2)
With regard to the APs being engaged by either the CAP or the F/O: it is considered possible that they were attempting to recover the normal descent path by selecting LAND mode and using the assistance of the APs.
(3)
It is probably that the CAP did not recognize that the APs were engaged, or that although he recognized it, he believed he could continuously override the APs. His belief may have arisen either from confusion with regard to the supervisory override function of the A300-600R, or from his flight experience in B747.
In this regard, the fact that the aircraft was not equipped with a warning function which would alert the crew directly and actively to the THS movement, when the AP was engaged in CML), is also considered to have had an effect on their judgment and actions.
(4)
The F/O did not report to the CAP either that he could not change modes or that the aircraft was not responding as desired (owing to a strong resistive force on the control wheel). Furthermore after the CAP had given further instructions and cautions to the F/O with regard to the mode change, he (the CAP) did not verify whether they were being properly followed.
(5)
During approach, the CAP had instructed the F/O to perform PF duty, assigning himself PNF duty. However, after the F/O triggered the GO lever, the CAP disregarded their duty assignment. It is considered that the CAP's judgment of the flight situation as PlC was inadequate, that control take-over was delayed, and that appropriate actions were not taken.
(6)
It is considered that the CAP intended try to continue the approach when he took control, but that he probably decided to go around when he found he could not stop the pitch angle increasing. Although the aircraft was climbing steeply with pitch angle still rapidly

3-43


increasing, the CAP seems not to have recognized, even at this time, that the aircraft was in an abnormal THS out-of-trim situation. This could be the reason why the CAP operated the Pitch Control Switch only intermittently, and did not reduce the excessively high pitch attitude.

3.2.4 Crash and Destruction of Aircraft

It is estimated that, after diving steeply without recovering from stall, and while rolling considerably, the aircraft impacted the ground almost in a level attitude. The aircraft was destroyed, and separated into forward fuselage, wings, aft fuselage, horizontal tail plane and vertical tail plane.

3.2.5 Investigation of Ethanol

There is a possibility that the ethanol detected in the remains of the CAP and F/O was due to a Post-Mortem ethanol production. The possibility of alcoholic ingestion before death could not be determined. The possibility that liquor loaded onboard splashed the bodies of the CAP and F/O could not be confirmed.

3.2.6 Operations and Training of China Airlines and Handling of Service Bulletins

3.2.6.1 Operations

It was recognized that China Airlines had an Operations Policy Manual and an Air Crew Manning and Dispatch Manual prepared in accordance with Taiwanese civil aviation laws, that the aircraft was operated according to these manuals, and that both the CAP and the F/O held valid qualifications for their respective duties.
The fact that the CAP had allowed the F/O to operate the aircraft on this flight is considered to satisfy the requirements of their crew qualifications, aircraft weight, weather conditions and airport. As described in Paragraph 3.2.3.(4), however, it is considered that the CAP's situational awareness of the flight conditions was inadequate and that control take-over was delayed.

3.2.6.2 Training

It is recognized that the CAP and the F/O completed classroom, simulator and flight training based on the training syllabus prepared by China Airlines in accordance with Taiwanese civil aviation laws.
However, it is recognized that this training was not necessarily sufficient to understand the sophisticated and complicated AFS system.

3.2.6.3 Handling of Service Bulletins

China Airlines received service bulletin SB A300-22-6021 issued by Airbus Industrie, on July 29, 1993. Since application of the service bulletin was categorized as "Recommended",

3-44


they judged its implementation not urgent and decided to implement the modifications specified in this SB when FCCs needed repair. Therefore, this modification had not been incorporated in the aircraft by the time the accident occurred.

3.2.7 Automatic Flight System

3.2.7.1 Modification of AFS

(1)
Before this accident, there had occurred three incidents in 1985, 1989, and in 1991 which also involved an out-of-trim condition. It is recognized that Airbus Industrie had not provided operators with systematized and sufficient explanation about technical background with regard to these incidents, and the proposal for implementation of the AFS modification had been delayed.
In addition, implementation of SB A300-22-6021 pertaining to the modification was issued as "Recommended", not as "Mandatory", which prescribes the highest priority.
(2)
Despite the importance of these three incidents, the airworthiness authority, pertaining to the aircraft designed and manufactured by Airbus Industrie, did not issue promptly an airworthiness directive for implementation of the SB A300-22-602 1.

3.2.7.2 FCOM

The contents of "Cautions" added to the FCOM, the descriptions in the revision to FCOM associated with the AFS modification, and the procedures for disengagement of GO AROUND mode are not easy to understand. In addition, FCOM does not specify systematically the primary purpose of the AP override function, the way to detect the out-of-trim situation, and the procedure by which crews can recover from it.
In addition to this, it is considered that of the technical information distributed by Airbus Industrie to each operator after the accident, the necessary and important items should be reflected in the main body of FCOM.

3.2.7.3 Activation of Alpha floor function under out-of-trim condition

The activation of alpha-floor function under the abnormal out-of-trim condition caused an sudden increase in the aircraft's pitch angle and contributed to its steep climb and subsequent stall.

3.2.7.4 THS Motion Warning and Recognition Functions

Airbus Industrie, during the redesign phase, eliminated the aural whooler function which had been provided in the original design as THS motion warning when the AP is in CMI). It is considered that Airbus Industrie did not conduct sufficient studies as to whether to maintain the function to provide crews with THS motion awareness and attract their attention to continuous THS movement when the AP in CMI), or to incorporate an alternative device which

3-45


can alert crews to the THS out-of-trim situation.

3.2.8 Fire Fighting and Rescue System

It is recognized that the Nagoya Airport generally had a fire fighting and rescue system almost in conformity to the "Level of Protection to be Provided", recommended by the Convention to International Civil Aviation, except that the discharge rate of fire-extinguishing foam solution did not completely satisfy the specified level.