The A320 Thrust Lever Trap
Flight Data Recorder on the Brazilian crash of a TAM A320 at Congonhas Airport has revealed that the pilot neglected to retard the thrust lever on the RH engine, leaving it at 22.5degs (its CLB detent), possibly because it is standard not to select reverse on a locked out (ie. unserviceable) thrust reverser. When the autothrottle clicked off after touchdown this caused that RH engine to accelerate to climb power, causing the wing spoilers not to deploy, the autobrake not to operate and the aircraft to accelerate off the end of the short 6362ft wet runway during a late attempted go-round. Thrust lever non-retardation to idle with a locked-out reverser is a known pilot error automation trap which has occurred in a number of previous A320 accidents. See the detailed explanation at the bottom here.

The A320 Throttle Quadrant
The white arc immediately to left of the No. 1 lever denotes the range from Idle to FLX/MCT (Flex/Max Continuous Thrust), with a mark between them for CLB. For take-off, the levers are usually moved to FLX/MCT. After initial climb out, the levers are normally brought back to the CLB detent, where they normally remain as long as auto-thrust is active. They can remain there until the "Retard" call-out at the flare before landing, when they are brought back to Idle. There are very distinct "clunks" at each detent, which can be felt by the pilot operating the levers.

If one lever is at 50% power and the other is at idle, there would be a difference of several centimeters, which would be noticeable with even a quick glance. With both levers in Idle, the difference might be a millimeter or two, i.e. they would appear "damn near" identical.

A fuzzier area of my understanding is that, after initial climb-out, the levers can be moved between Idle and CLB for manual thrust control, although the chosen setting may only dictate the maximum thrust the system will allow, assuming that the aircraft remains within safe parameters.

The black tabs extending forward from the lower part of the levers are the reverser latches(?). The movements needed to activate reverse aren't much different to those used in a Boeing, logically speaking.

After initial climb out, the levers are normally brought back to the CLB detent. That's what engages Auto-Thrust, assuming it's armed.... and auto-thrust will remain active as long as the levers remain there, i.e. between IDLE (but not "at" Idle) and CLB, but with restrictions on how much maximum thrust the system will produce in normal circumstances.

It's significant that, in all four cases (Bacolod, Phoenix, Taipei, and now presumably Congonhas) the common factor was that one reverser was inoperative, and the misplaced lever was on that same side in each case. That suggests that (for whatever reason) the pilots may have seen the priority as AVOIDING applying reverse thrust on the inoperative side.... Which may - repeat MAY - have led to them being too tentative with the setting of 'Idle' on that side?

Eric Parks, US Airways pilot, in his unofficial A320 notes, makes no less than three references to the need for 'care and attention' in retarding A320 throttles and selecting reverse thrust (see under 'Landing'):-

"Do not carry thrust to the flare as the autothrust will begin to command climb thrust as speed deteriorates if you do not bring the TL back to idle. This will cause a "thrust bump" that will have you floating down the runway with excess energy.

"Be sure you push the thrust lever all the way back through the detent into forward idle. Then retard the lever again against the stop to ensure minimum forward thrust in idle.

"Another misperception is that some folks will reduce the power very slowly. However, remember that autothrust is active until the thrust levers are all the way to idle (assuming autothrust is already active). So once you bring the thrust levers out of the Climb detent you aren't actually reducing thrust until the levers get all the way back to where autothrust has them commanded. You will only be limiting the amount of thrust that can be commanded. If you bring the levers back slowly you are only reducing the maximum amount that can be commanded but not actually reducing the thrust until you get them very far back. If you wait too long, you get the thrust bump we just talked about  - as autothrust is still trying to maintain the speed."

http://www.chipsplace.com/helpful/Airbus/Go%20Around.htm#Landing

The most interesting part in this context (to me anyway) are the phrases '.....all the way through the detent into forward idle. Then retard the lever again against the stop.' 'Through the detent,' and then retard even further, 'against the stop' - not just 'back to'?
The Accident Scenario

The "body English" involved in inadvertently leaving the thrust lever associated with the failed reverser up at CLB goes like this:

a.  The pilot is aware that he shouldn't use reverse on that engine but to a greater extent he is entertaining the need for a very prompt introduction of "reverse" on the engine with the reverser that's "good" (i.e. not locked out).

b.  Consequently it's an act of omission, i.e. that important first stage (of pulling both TL's to IDLE first) is omitted simply because the pilot is seeing the act of reverse as his immediate task destination - and achievable in one staged movement on the engine that's capable of reverse. He's prioritizing reverse as he knows that it's critical on that short wet runway - and very much aware that all else is automated (spoilers and auto-brakes). The critical need to first go to IDLE on both thrust levers is forgotten. Unfortunately, he easily forgets that if you leave that one TL up at CLB, he just won't get autobrake or spoilers.... and that the autothrust will click off, allowing that engine to accelerate itself to climb power.

c.  If that's the mistake they made, and didn't realize it (no time for trouble-shooting), then they would have panicked and gone for the late re-introduction of power (TOGA).... a go-round being very apparently, at that late stage, the only possibility for avoiding a disastrous overrun.

d.  The complication that arises there is that the RH engine (already developing CLB thrust) would readily accelerate. The LH engine needs 5 seconds to come out of reverse and reconfigure (i.e. the T/R(s) being hydraulically stowed and locked). There's every chance that it will compressor stall upon the imposition of TOGA power during that process - in fact that's an Airbus warning and the reason given for pilots not attempting a go-round after reverse is entered.

e.  The first flash seen on the video would have been that left-hand engine's compressor stall

f.   The thrust asymmetry was what took them off centerline into the TAM cargo warehouse.

There are a number of prior instances of this mistake (see this link)

This page is at tinyurl.com/2dpkwa
 
One aspect not mentioned above is the answer to this vital question:
Does the GPWS keep repeating the "Retard" message until the pilots actually retard the/both thrust lever(s), or is it only played once?
If it is not repeated, why not? It could have prevented this accident if the cause was the failure of the pilots to reduce power to IDLE on BOTH TL's.
 
Answer:  The "RETARD" auto-call repeats until touchdown, then it cuts off (whether or not the levers have been fully retarded). This feature was criticised by the Taipei crash investigators (TransAsia accident, October 2004 (A320 with one reverse thrust disabled, runway overrun etc.) - link):
 

Quote:
The ground spoilers and auto braking system were not activated timely to decelerate aircraft due to thrust control lever no.2 not being pulled to the idle position. Though the Flight Warning Computer （FWC） had delivered four "RETARD" aural alerts at 20 ft RA before the aircraft touched down, the pilots did not pull the thrust control lever no.2 back to its idle position.

When auto throttles are used under normal conditions, if either of the thrust control levers is not pulled to idle position, no matter using the automatic landing or the manual landing, at 10 ft or 20 ft RA the FWC will deliver "RETARD" aural alert
and then automatically stop when both thrust control levers are pulled back to idle position. From 1959:23(at 23 ft RA), the FWC began to deliver "RETARD" aural alert. After four times alert, it automatically stopped but not caused by the both thrust control levers at idle position. According to the documents provided by the aircraft manufacturer, the reason for the "RETARD" call-out stopping was due to the thrust control lever no.1 being placed into the reverse position. When the FWC detected an internal signal of TLA inhibition which includes either thrust control lever being at a REV position, then the FWC ceases its aural "RETARD" alert.

During landing, the FWC delivered four aural "RETARD" alerts and then, after touchdown plus two seconds, it ceased. But at this moment the thrust control lever no.2 had not been pulled back to idle or the reverser position. In this situation, one of the thrust control levers is not in a proper position but the aural "RETARD" alert has already stopped and will not warn further about this hazard. The aural "RETARD" alert should continue or there should be other methods for reminding the pilots to pull back the thrust control lever to reduce any probability of an accident being caused by human error.

 

"3.1 Findings Related to Probable Causes

"1. When the aircraft was below 20 ft RA and Retard warnings were sounded, the pilot flying didn�t pull thrust lever 2 to Idle detent which caused the ground spoilers were not deployed after touchdown though they were at Armed position, therefore the auto braking system was not triggered. Moreover, when the auto thrust was changed to manual operation mode automatically after touchdown, the thrust lever 2 was remained at 22.5 degrees which caused the Engine 2 still had an larger thrust output (EPR1.08) than idle position�s. Thereupon, the aircraft was not able to complete deceleration within the residual length of the runway, and deviated from the runway before came to a full stop, even though the manual braking was actuated by the pilot 13 seconds after touchdown.

"2. The pilot monitoring announced �spoiler� automatically when the aircraft touched down without checking the ECAM display first according to SOP before made the announcement, as such the retraction of ground spoilers was ignored.

"3.2 Findings Related to Risk

"1. After touchdown, when the thrust lever 2 was not pulled back to Idle position and the Retard warning sounds have ceased, there were no other ways to remind pilots to pull back the thrust lever.