C-FWGH 738 slow take off BFS : AAIB
 

https://www.gov.uk/government/news/a...737-86j-c-fwgh

sounds like it could have been a lot worse

Yet another example of a Pilot/Automation interface failure...

That's 'ok'...

However... what we continue to 'miss' is the back to basics of being a 'Pilot' of an aircraft.

What should be happening in this situation is clear... A 'thinks bubble'... 'Hey... why are we accelerating so slowly? Why is the runway end coming up? Mmmm, something is not right...

Reaction... before, or certainly past V1, we are now going flying. Lets apply full-power, manually (Firewall the trust levers) and get this aircraft into the air and climbing away. Sort out/discuss the cause later.

30+ years of check and training experience says one thing to me. This is a training issue... even probably a 'pilot selection' issue. And fundamentally, a regulator issue.

It starts with more than 'I have £100k, please make me a pilot' and should absolutely ensure that at 'ab-initio' and 'basic' flying training, commonsense and 'natural' pilot aptitude is followed up right through a pilots career/training.

Yes, we need more pilots and the demand for pilots is ever increasing. But this incident is a hairs breath away from yet another totally avoidable hull loss...

History is littered with similar incidents/accidents. The time will come where we reach tipping point on 'dumming' down selection and training versus front page 'newspaper' reports of yet another accident.

Do we have to wait for this? The 'threat' is clear. Lets do something about it now. There are more and more threat signals every year. That response must also include by-passing commercial pressures and e.g.: share and stakeholder budget cutting 2% YOY, in training and regulation. The industry at large, will be much better off as a result.

Another Friday night incident like the Air Canada 759 taxiway approach at SFO where the feds were unable to get the CVR (and FDR in this case) due to reporting delays and lack of weekend office hours.

From the AAIB Special Bulletin Introduction:

Were any laws or regulations violated?

What were the corrective actions? were they airworthiness related or presumed crew or operator error?

The report (which can be read on the link in post #1) strongly suggests that the primary cause was the crew entering the wrong (TOC) OAT into the FMS, rather than the airfield OAT.

The error was compounded by the fact that the aircraft didn't have the latest FMS software version, which would have trapped that particular error.

a. The expected top-of-climb outside air temperature (OAT) was entered into the OAT field on the n1 limit page instead of the OAT at the airport (a figure of - 52°C as opposed to +16°C); and
b. The correct assumed temperature of 48°C was entered into the FMC.

No other combination of data entries was found which would achieve the same result.

81.5% N1 ~ should danger bells not have rung on stand, given 179 passengers (189 max) & the sector length (Corfu). A reasonably heavy ac, on a not exactly long runway for a -800
Had it been a short hop to say the Glasgow, & virtually empty, then a much lower N1 would have been generated by the FMC.
No point talking below about EPR values. The NG doesn't deal in that currency.

Simply a lack of airmanship & empathy. Clearly a training issue as well??

BOB ERIC is getting close to the truth......

And the penny is finally dropping with EMIRATES training dept as well according to pprune postings. No doubt pressure from above ?

In my company we use derate and assumed thrust method. The other day we took of with 22k and mid 30 degrees
on the assumed temp. Result was 81% N1 and flaps 1. It is possible and you have to follow your SOP's very closely not to screw up. There have been lots of accidents and incidents entering data from the EFB into the FMC.
But I never heard someone entering the TOC temperature on the N1 limit page as OAT.
Training standards and SMS system should be reviewed in this company.

I would also gently question the certification of automation that doesn't even try to validate key inputs.

We've had thermocouples since about 1820.

Back to basics
 

If only they had built the aircraft with some sort of forward facing glass panel so the pilots could look forwards and take note of what was going on ..

I seem to be flying a lot of odd weight freight these days, and the "70% of V1 by 50% of the runway" rule works well for us. We aborted one last month that seemed a bit slow ... nailed it faster and harder the second time around and got the numbers up.

If you are just "flying the numbers" that the FMS says are right, then you are not a pilot, you are a passenger.

Not mentioned was any cross-check of the N1 in the "documentation" (93.3) verses the N1 the FMS calculated (81.5). I assume we will read about the company procedures in the final report.

I don't fly the 737 but wouldn't the EFB also produce an N1, which would be crosschecked with the FMS?

By coincidence an accurate description of what happened to me one night take off. . Actual event at night. A Pacific atoll runway 5600 ft long with the ocean at end of runway. B737-200. Planned 2.18 EPR with associated 101% N1. I was dead heading in the jump seat.

F/O was PM and captain set thrust levers to 2.18 EPR. Cockpit lighting dim due night. The difference between 101%N1 and 90% N1 on the needles was about 3mm and easily missed especially as both pilots were concentrating on the take off run and no visible horizon. All other engine instruments were parallel and in the needle positions where expected.

It was not possible to gauge exact rate of acceleration at night by seat of pants feeling. With about five runway lights to go, it became evident we were not going to get airborne by the end of the runway. The captain immediately took control, simultaneously fire-walled both engines and hauled back on the control column at what was then V1 minus 15 knots.

It was then an immediate kick in the back increase in thrust was felt. The FDR later showed we maintained 20 feet over the water for half a mile and before climbing even though the body angle was 15 degrees up. The high body angle caused the engines to blow ground debris back over the runway. . On fire-walling the engines the EPR shot up to 2.25 an unheard of figure in the JT8D-17 engines.

On setting 1.98 EPR eventually after flap retract, the rate of climb was significantly less than scheduled. Comparison between 1.98 EPR and expected N1 showed the N1 was down by 10%. The decision was made by the captain to return to land. Investigation revealed that both PT2 probes were blocked thus giving an erroneous EPR reading to the crew. In fact at 2.18 indicated EPR we actually got about 90%N1 instead of the planned 101% N1.

The point being made that it is difficult to accurately predict the rate of acceleration in airspeed between 101% N1 and 90%N1 until it may be too late with the end of the runway coming up fast - especially with lack of cues at night.

N1 as a measure of thrust is more reliable than EPR as the Centaurus incident shows.

The amount of thrust is not directly proportional to % rpm though. I seem to recall that on the CF6-50 (approx 52000lbs thrust) the difference between 100% N1 and 95% N1 was about 10000lbs of thrust, ie. 20% of power. This on an A300.

As all big fan engines work the same way and most of the thrust is developed at the top end of the rpm range, I doubt they had anything more than about 50% of available thrust with an N1 of 81%.

Very lucky on that day!

Lower than expected N1 values as a result of erroneous EPR indication we're a contributory cause of the Air Florida B732 crash many years ago.
It is easier to cross check EPR with expected N1 at the commencement of the roll than judge acceleration rate.

Lower than expected N1 values as a result of erroneous EPR indication we're a contributory cause of the Air Florida B732 crash many years ago.
It is easier to cross check EPR with expected N1 at the commencement of the roll than judge acceleration rate.

indeed, and that became an SOP in my B732 outfit. Both numbers were written on the bug card.

Back to the topic incident. As Bob E says there is a lack of knowledge of the a/c today and application of gross error checks etc. It is a training issue, no doubt. Today's concept is learn how to use the EFB, learn how to program the FMC, learn the SOP's, learn to follow the FD & let the automatics do everything for you. "It's safer the way." Duh.
There is no feeling for what should be correct and if it ain't then something is wrong, so check it.

Increasing thrust during takeoff roll due to lacklustre acceleration is a good idea & reactive: checking the thrust setting after the calculation seems low is a good idea & is proactive. Which is better? But the latter can only happen if you've paid attention previously and not been a trained monkey.

Although not directly related to the OAT error which caused this incident, is anyone else reading this thread and the AAIB report wondering why it took the pushback crew to notice the nose gear tyres were so badly worn they both needed replacing before the airplane actually departed ? Surely an item that should have been spotted on the crews pre-flight inspection of the airplane. Did they miss it, or if they notice it, did they still deem the airplane airworthy ? Who would the pushback crew inform after they noticed this, apart from the crew ? And who would make the actual call to have them changed ? The crew or operator ? To me that points to lack of attention to detail and preparation. Maybe the FMC programming suffered the same way. Just a thought.

BOBERIC; First post and excellent. Many others have repeatedly made similar comments. Selection is vital. Let anyone in with a fat wallet and error numero uno rears it's ugly head. Minimal initial and advanced training compounds the problem. Airline selection almost non-existent. Airline, minimal button pushing training ( even to CBT and "at your own pace" modules) and we have a woeful mix. SOP's that then discourage manual flying (because the new breed couldn't cope with the unexpected) and you are now heading for incident after incident culminating in a prang .


But the low cost operators and bean counters who insist on all this from the risk-free comfort of air-conditioned armchairs still have their way. And like the CEO of one messed up major carrier just stated; " Once trained, the pilot's job is easy" ! Couple of hull losses and a few near misses resulting in zero confidence by the customers might shake up that argument, a bit. But, I doubt it.

"a. The expected top-of-climb outside air temperature (OAT) was entered into the OAT field on the n1 limit page instead of the OAT at the airport (a figure of - 52°C as opposed to +16°C); and
b. The correct assumed temperature of 48°C was entered into the FMC."

When I write software (mostly medical) I take great care to "sanitise" operator input, i.e., make sure that it is likely or even possible. For example, in calculating fluid resuscitation there are cross-checks between the "date-of-birth", "age" and "weight" fields.

Entering a 48kg for a 1-year-old will generate an explanatory error - "Do you really have a 1-year-old who weighs 48kg. Input refused."

The OAT at Belfast may have been -52degC during the last Ice Age but it is way out of the normal in 2017.

Making sure that all inputs are sane is a chore and can make up as much as 50% of the code, but everyone can make mistrakes and one must check of them.

a. The expected top-of-climb outside air temperature (OAT) was entered into the OAT field on the n1 limit page instead of the OAT at the airport (a figure of - 52°C as opposed to +16°C); and
b. The correct assumed temperature of 48°C was entered into the FMC."

There also seems to be a lack of X-checking by both plots of data entered into their electronic toys. How does it work in most airlines? Do both pilots make takeoff performance calculations independently on their own i-pads as a X-check, or is it just one who does it? And even then, is the data entered read out by the other or is it just a one man band calculation? If the latter then there really is a chance of an unmonitored screw up.

As the AAIB report makes clear, a sanity-check on the entered OAT is a feature of later releases of the FMS software on the 737NG, hence the Safety Recommendation that the FAA mandate use of those subsequent revisions.

The incident and recommendation comes after a multitude of events.

In addition, it appears that the programming has now been made with efforts to attempts to trap even more errors made by the flight crew.

While many drivers claim the children of the magenta line mantra, with an over-reliance on the automation, it appears that much of the automation is has been put into place for a reason.

Might competing with Airbus and other manufacturers be one reason?

Would this error be possible to achieve on the bus ?

All aircraft are liable to data entry errors. While this particular combination may not happen on the bus, there are plenty of other ways of killing yourself.

Load sheet and performance data is something that will kill you if you get it wrong and requires the utmost vigilance in entry and cross checking. I believe the 787 cross-loads data from the EFB performance function to the FMC which removes 1 avenue of entry errors.

Until recently, we didn't even enter the OAT into the FMC on the 777 where I work. We have been doing this for a while.

Is this error something specific to the 737 FMC/aircraft or would it cause the same type of thrust error happen in the bigger Boeing's as well. I know that there are some FMC differences as well as significantly different engines.

TPWS?
 

Given the nearly centimeter level accuracy of CAT ADS-B GPS sources, it would not take much math to predict engine out height over threshold given cumulative acceleration and runway position.

If less than 50' a stop alarm could be issued while there still remains adequate stopping distance. A fancier system would message the auto throttles to advance.

This could all be done with GPS and airspeed inputs completely independent of FMS.

Threshold Proximity Warning System

Just like the RAAS system? Slavish adherence to it's long landing "guidance" mandating a go-around, as required by SOP's will almost certainly be a contributor to a 777 hull loss - Runway Impact During Attempted Go-Around despite there being ample runway available to stop.

This is not a simple plug-n-play fix. Humans make errors, software isn't context aware. We are a long way from solving our cognitive and software limitations.

As a side issue, I always cross check the FMC N1 against the EFB computed N1, most times it is incorrect when attached to an aerobridge on the NG (aspirated probes). Thus far, no one has been able to reason the source of this anomaly, in fact it is rarely questioned despite the consequences of an incorrect figure. After pushback, the computed and FMC figures resolve into agreement. Why? The answer is located on the upper DU.

How it was explained to me on B777.

EFB value is correct. FADEC's static value is current aircraft configuration (bleeds off). FADEC (% or EPR) changes to correct value once engines started and recalculates with bleeds on.

The FMC knows that the Standard temp is, and will calc per that, baring other input. As long as you are at 20 degrees C, and at sea level, you are good.

What about ISA DEV?

What do you do for FMC PERF INIT? and T/C OAT?

It's actually quite simple, the aerobridge deflects airconditioned air (either aerobridge or cabin) which then escapes around the seal to the aircraft directing air close to the aspirated TAT probe. The TAT reading is correct, the air is not at the ATIS temp. On a hot day TAT under-reads, and cold morning TAT over-reads. This alters the FMC computed figure, and disagreement with the EFB. After pushback, the TAT again measures the ATIS OAT and the two figures resolve.

The really surprising thing is the lack of acknowledgment that the anomaly exists, and the lack of further questioning as to why. As I said, the answer is is on the upper DU, with a simple cross check of TAT vs ATIS to reason the source of the discrepancy.

Acceptance of discrepancies without question can lead to these types of incident.

On the 727 and 737-200 we used to always crosscheck the N1 and had a pre-calculated minimum value to be reached which was written down on the Takeoff data card.

The call made was "Takeoff thrust set-N1 checks". Something to consider doing, especially on the shorter or field limited runways with JT-8D engines.

The call made was "Takeoff thrust set-N1 checks"

This was introduced on our B732 fleet after the Air Florida went swimming in Potomac. Same reason, blocked pitots, but via a different cause. In this AB event was in not a case of rubbish in rubbish out; i.e. wrong assumed temp entered in EFB. Cross checking the N1% for the Temp would have provided the correct corresponding answer, but still too low thrust. What was being discussed is too many modern pilots don't have a mental model with which to make gross error checks. I doubt these guys had ever taken off, even of a ferry flight, with N1% <90%, yet on this day it is reported they tried to do so. An error was made, I assume by one guy, and no bells rang for either of them. How to prevent it? Some say that the EFB should have an error alert program in it. I know some airlines where both pilots do the performance calculation independently and x-check. I'd like to see better awareness trained into pilots as well. Aviation hasn't changed that much with the advent of computers over pencil & rubber & tables. It is the same dumb humans who execute the tasks. The more complacent we become with "it's a computer so it must be correct" the more accidents of this nature we will have. After 30 years of FMC, pilots have developed a wise suspicion of VNAV profiles. They spot crazy things happening. If the FMC can be suspect and thus encourage gross error checks, why not EFB's? Too old school?
I'm amazed that some pilots are so keen eyed that they spot N1% increasing after takeoff when CLB power is selected. They often ask the reason, and can not always find the answer (it's been discussed for years on Tech Log) but at least they noticed. That's great, and we need more pilots who observe and ask questions about things that don't seem correct.
Isn't that part of our job?

Why a certified, calibrated accelerometer is not part of a modern jet aircraft instrument panel I'll never understand. It's a bottom line performance indicator.

I imagine most PF would consult it at about 80 knots on short runways take-offs. No?

Can you imagine the mental maths it would take in a split second. Plus how far down the runway are you?

No math required...acceleration at or above set bug value, you Go. Below bug value, you stop. Easy.

100% correct. Here in Canada, airlines are so short of pilots that basically anyone with a pulse can get a job. I've witnessed horrible check rides where the candidate still gets signed off. They just want to fill up the cockpits to keep the planes flying.

Yes it does, yes we do.

Bob gric
Excellent post home truths.
Pilots in the past had to have the ability & ability to have empathy & “gut” feeling of what the aircraft is doing. £100k does not give you that. Unfortunately too many are coming into the industry, because anyone can be a pilot can’t they. (Piece of cake old boy) well actually no.

Yes it does, yes we do.

But it still won't tell if it is wrong, if the wrong temp is inserted in both EFB & FMC. If there is some kind of x-check that the ATIS OAT is inserted in the boxes you may pick up an error. But no pilot should be content to attempt to takeoff with <90%. That's gut feeling at least; surely. nd it's 2 pilots who stopped thinking, not just one.
What has dismayed me is the dilution of basic airmanship in TR courses for cadets. The rapid expansion of young airlines has needed qualified bums on seats up front PDQ. Guys leaving after a few years experience faster than they can be replaced, and the expansion covered as well. Fair enough, you pass the LST, but the education then stops. SOP's designed for trained monkeys and then 3-4 years later you are sitting in LHS with not much to pass on to the newbie in RHS. Both of you are locked onto an SOP train line where you hope there will be no sharp bends, overheated this or that and definitely no 'wrong kind of leaves' on the line. Previously you would have waited and served an apprenticeship for 7 years in RHS, seen a lot and hopefully soaked up and learnt a great deal. You had the confidence that you could handle the a/c on an ideal day, lead the crew, manage the operation and cope with a few hiccups without having to think SOP first & airmanship/common sense second.

Tried the same thing twice recently on the 777. One was at a high derate/ATM setting(TO2 assumed 57) and the other was at a heavy weight. In both cases, there was little to no change in the target thrust setting. Perhaps the other Boeing aircraft have a protection feature for this kind of error.

that is not what we are discussing though Rat. What happened here is that the EFB temperature was entered correctly, the correct assumed temp was entered into the FMC , but the N1 was wrong. So in this case, simply checking the N1 from the EFB is the same as the N1 in the FMC does in fact trap the error.