United Overweight Takeoff on Computer Mistake Prompts Changes

[SeenItAll]

According to this report, incorrect information provided by dispatchers to the flight crew with caused this UAL 737-900 to take off at a weight about 20,000 pounds (9,071 kilograms) heavier than than pilots believed.

United sent pilots a weight estimate that assumed the coach section of the Boeing Co. (BA) 737-900 was empty when it was full. ... While the pilots, who didn’t catch the mistake, had difficulty getting the jetliner airborne, the plane wasn’t damaged and the flight was completed without incident. ... Boeing’s 737-900 models can take off weighing as much as 187,700 pounds (85,141 kilograms), according to Boeing’s website.

Given that the pilots obviously knew how much fuel they were carrying, and that they had a full load of passengers, it seems concerning that they didn't notice such an anomalous figure for total weight. Note that this instance seems different from the Emirates MEL flight where an incorrect weight was inadvertently keystroked into the FMS. Here the provided weight sheet was wrong.

See full story at: United Overweight Takeoff on Computer Mistake Prompts Changes - Bloomberg (http://www.bloomberg.com/news/2012-08-11/united-overweight-takeoff-on-computer-glitches-prompts-changes.html?cmpid=yhoo)

[Robert Campbell]

Perhaps the crew didn't get enough rest.

[ironbutt57]

Bit of a "situational awareness" issue there.....:eek:

[framer]

Is there an NTSB report we can read? Any link to that would be appreciated.

[mach2.6]

Seems like we may be back to discussing the possibility of implementing a line check speed, as when Emirates took a tour of the departure end in Oz a while back. Every military fighter I ever flew had a line check speed. The number was calculated for each flight before anybody ever headed toward his aircraft. Infallible system. Sir Isaac Newton: Force=Mass times Acceleration. After brake release, you looked at your ASI as you passed the 1000-foot mark (2000 feet for some aircraft) on takeoff roll. If you didn't achieve that number, you aborted. Figured out what was wrong. Lived to fly another day. Didn't run off the end of the runway dragging a bunch of navaid equipment with you. If you were off by 3-4 knots, that was ok. Off by 5 or more knots, abort. If the Force (thrust) was inadequate or the Mass was too great, not enough Acceleration was produced, making you "check speed" too slow. Simple, effective. No laptop or dispatch needed.

[boguing]

For the whole of my working life, which has mostly been Maths and Physics (same thing really) the cost of electrical strain gauges has been trivial.

So either fit them to the gear and know how much you weigh, or at the least install some measuring pads at each runway hold point.

[Prober]

Come back, STAN, all is forgiven!

[Basil]

Maths and Physics (same thing really)

http://img.photobucket.com/albums/v233/Capn_Basil/purity.jpg

Our 747 freighters had a weight readout but, whilst the B744 was OK, the old B747 system was considered too expensive to maintain and was disconnected.

the possibility of implementing a line check speed
I understand the V Force used that system. I wonder if it would be workable with the civil airline complication of variable thrust? Whilst I guess the speed could be tabulated or otherwise calculated would errors result in unnecessary aborts?

[boguing]

Who or what is STAN?

[Joetom]

Very intersting subject.

I have been very interested in performance for many years, T/O and LDNG performance is so interesting.

We have seen so many examples of getting it wrong on T/O and LDNG, I guess many many get un-reported to a greater or lesser extent.

I used to think an easy "rule of thumb" chart could be used to help/assist in helping/informing/assisting to avoid falling into traps, I guess it can be of help, but history records it has limited use/effect.

I list a a few examples of classic events that may of been helped/avoided if a system was in place to remove/reduce/reduce-effect them, see below.

Air Florida, EK at MEL, SQ at AKL etc etc.

As the previous and so many un-mentioned events show that operating pilots can't see the big picture of performance/power/thrust/EPR/N1/availabe runway lenght/distance avail etc etc.

So many systems avail to help assist this prob, but cure could be less safe overall.

Reminds me of also runways not in use and reduced runways problems.

[The Ancient Geek]

the possibility of implementing a line check speed


There is a simpler way which is easily automated.
What really matters is acceleration which is easily measured and runway length which is a simple database item. We do not need to rely on weight or thrust calculations which are error prone. If the acceleration is not enough to achieve a safe accelerate/V1/stop distance within the available runway length just sound an ABORT ABORT alarm.

[awblain]

There's a bit of a complication to that accelerometer-based approach in a potentially gusty wind; but, it should be able to spot when either the mass is significantly higher than assumed, or the thrust is less.

You'd need to wait until power has stabilized too, but with both GPS and inertial navigation giving acceleration, there should be no doubt about whether Mr Newton recommends it'll be time to go flying before hitting the lighting boom.

[golfyankeesierra]

T/O and LDNG performance is so interesting.:eek::eek:
You must be a ground-course instructor! ;)

[mutt]

the flight was completed without incident. On 2 engines... what would have happened if they had an engine failure on takeoff.....

Mutt

[limelight]

Guys, this is not rocket science.

A half decent programmer can write a calculation program that asks pertinent questions to error check the fuel/TO calcs.

In the case of EK, a simple series of questions, like departure location, destination, aircraft type (for the AB guys who swap a lot), POB, windspeed would come back with an answer that can be compared with the crew calcs.

A simple result of 'very near' or 'miles out' would be returned.

Now, back to programming....

[Bigmouth]

Or you could ask yourself: Does it make sense?


(Complacency will kill us all)

[Fris B. Fairing]

As a non-pilot I don't pretend that I've ever used it but as a former load controller I seized the opportunity to acquire a piece of technology that failed to replace me! I'm told the acronym means Sum Total And Nosewheel.

http://www.adastron.com/aviation/vault/stan.jpg

I'll leave it to an expert to explain how it worked.

[DOVES]

Please feel free to give a look at:
http://www.pprune.org/rumours-news/148355-mk-airlines-b747-crash-halifax-32.html#post2147179
So we keep making the same mistakes again and again.
This time everybody went away on their legs.
Next time?
Fly Safe
DOVE

[Los Endos]

the old B747 system was considered too expensive to maintain and was disconnected.

The MK guys paid with their lives because they were hugely tired and evidently the price of their safety was too much......

[Heathrow Harry]

If we can send a car to Mars surely we can come up with sensors on the undercarriage that tell us how much weight they are carrying?

Maybe ask Dyson or Apple instead of Boeing and Airbus???

[DOVES]

Heathrow Harry:
http://www.pprune.org/rumours-news/492890-united-overweight-takeoff-computer-mistake-prompts-changes.html#post7358393

We had it on DC-8 in 1970's

[framer]

So either fit them to the gear and know how much you weigh, or at the least install some measuring pads at each runway hold point.
That would solve some problems but not all problems. The crew of the Emirates flight in Melbourne both understood that they were at 362 Tonnes, they just didn't know that they had done the data for 262 Tonnes.

[saltyfish]

Boguing,
Second that thought:ok:.
This seems to me like it would be a perfectly workable solution.
By locating load cell/strain gauge arrangements near or at the MLGs and NG, the weight and CG location could be measured directly, thus eliminating guesswork.
It looks like STAN did just that.
The whole setup could be periodically checked/calibrated against ground pads.
The challenge would then be to come-up with a setup that would remain accurate and calibrated in spite of the higher loads encountered during landing and wide temperature variations.
Could someone more knowledgeable on the subject shed some light:D?

[hetfield]

Don't know what UA procedures are. Working for a major EU carrier we received the flightplan (weight/fuel calc) about 30-45 min before departure. This gives a pretty good idea about T/O performance. After pax checkin completed we received the loadsheet.

Most of the time only minor changes.

Anyhow, a gross mismatch like not taking in to account a complete pax section would have been noticed by the pilots, ehhrrrm most probably.:ouch:

[MrMachfivepointfive]

If we can send a car to Mars surely we can come up with sensors on the undercarriage that tell us how much weight they are carrying?

The problem is the wind load on the aircraft sitting on the tarmac. In anything but still air, the weight measured at the oleos will be off.

[Joetom]

Or you could ask yourself: Does it make sense?


(Complacency will kill us all)

Yes, This could save some, but some some will slip through.

It just reminds me of that triangle thingy, keep using up all the lower blocks and then just the top remaining one to be picked up and used, horrible.

It just seems a shame with so much data available (weight/trim/C of G/pwr/flaps/location/avail runway/accel/weather/temp/pressure/position of controls in the flight deck/state of a/c systems etc etc) that we can't have a nice safe attention getter when it's reqd.

Investigation: AO-2012-020 - Pre-flight planning event - Boeing 737-476, VH-TJL, Melbourne Airport, 22 November 2011 (http://www.atsb.gov.au/publications/investigation_reports/2012/aair/ao-2012-020.aspx)

Thinking of this one above and the EK at MEL, just great examples of how crew can get out of the real loop.

Infact, from memory the EK crew had operated at weights between 160-360Tones and flight times between 2-12 hours and a variety of runway lenghts avail, so I can understand when they did not question when the engines spooled up and all the pointers and indicators were at low values, again from memory, it may have been a full de-rate T/O !!!

[Basil]

The problem with a safety system is that you don't know how many accidents it has prevented. Without that proof it's easy to decide, like our 747 Classic weight and balance indicator, that it is not necessary and save on maintenance costs.
Having said that, would anyone argue for the removal of TCAS or GPWS?

What if Airbus fitted a sidestick position indicator which could be seen by both pilots - and a captain standing behind the seats? (or moved the poles to the front of each pilot)

[framer]

I think that adding alarms or weight strain gauges or installing weigh plates at holding points would have some benefit, but not as much as we would hope.
How about addressing the underlying factors that actually cause competent , experienced pilots to make errors they never thought they would make. Ie distraction during performance calculations, pressure to depart on time, added complexity to the preflight procedures with a new task jammed into the procedure every six months or so.
For example, if when the performance calculations were to be done, the flight deck door was closed and radios turned down, and the refueller / Engineer /gate staff / Purser / waited quietly in the front galley until it was opened.....do you think that would reduce the number of performance errors due distraction?
It would go a long way towards it. It would also impress upon the pilots the importance of the calculation so as to guard against complacency. The Captain of EK407 said he had drifted into complacency over the last year or so regarding distractions during this phase and on that flight there were a lot of distractions/ interactions occurring. There were other factors obviously like poorly designed SOP's.
Anyway, my two cents.

[boguing]

Slightly ashamed to admit that my cry for strain gauges on the aircraft or ground pad would not account for surface wind. Pretty sure that it would work when up and cosy with the terminal and in crosswinds elsewhere, but it is not an absolute solution.

[pontifex]

In the early days of the V Force we did have a system called ACP or accelleration check point. It was a specific, well marked point on the runway by which a pre calculated speed had to be attained. Otherwise, abort. Of course all take offs were from a very specific point with the brakes released just as engines were coming up to full power. Thus, this technique would not work with normal civil practise. When I became a civil pilot it always seemed slightly "gash" to me that such exact calculations were done but that the most fundamental element, the point of initiation of the roll, was so indeterminate in practise.

[Prober]

Fantastic to actually see one! It totalled the oleo weights and gave a c of g estimate. On the Merchantman (VC9 or converted Vickers Vanguard for those without a memory as long as mine), it was a very rough cross-check. There were no known limits which had to be observed and it was, in my view, mostly psychological. If it did not match, even roughly, your own T/O weight estimate and the load sheet weight, then you probably had a problem. The c of g was supposed (IIRC) to be roughly within 4% of calculated. That certainly saved my life on one occasion. But most importantly, what it do do, was make you have your own estimate of the T/O weight in mind.
Can you force someone out of complacency? Well, this went some of the way.

[DOVES]

Dear Sirs.
The CHECK "Time to 90 kts" could have been, and can be, of some help in so many occasions.
I found, among my many FOMs the SE 210 Performance Chart.
Entering: Temp., P.A., A/C Weight, Wind Comp., Rwy Slope, Eng. bleeds, Flaps position, Rwy contamination, it was possible to calculate the acceleration time (sec.) to 90 kts.

[error_401]

I used to fly on four variants at the same time PAX and CARGO. To have something to crosscheck loadsheets for plausibility because of the very different masses I had a sticker on my clipboard with:

DOM: Dry operating mass (based on a standard catering) of all variants
Wingspan:
Height:
Wheelbase:
Full turn radius:
(The span and other measurements are very helpful when doing a lot of charters.)

Up to today I've seen such very helpful stickers only in trucks.
Would greatly enhance awareness when DOM was e.g. found on the limitations placard instead of hidden in the performance section of the AFM or OM-B as we say over here.

I always crosscheck for plausibility using different figures from different sources:
DOM + FUEL + PAX (at 100 kg each including bag) = RAMP MASS

DOM (you should have that from your company or at least have an approximation)
FUEL (you should defintively know about that)
PAX (not what's on the loadsheet but what your purser tells you she/he has counted)
CARGO (the only source may be the loadsheet - but to some extent you even have to trust that)

Figure out your own requirements.

[DownIn3Green]

Yeah, well, as a Capt I know how. Much power is required to start my taxi based on what I think I weigh. If it's too much, time for a reweigh. Especially true in cargo ops. To miss by 20,000 lbs. is beyond the scope of my imagination. Maybe the Captain was one of those affirmative action hires from the 80's...Where's 411a when you need him, RIP...

[At ease]

No expensive measuring pads needed.

The heavier the aircraft, the closer to the ground it will be.

A laser light installed on the bottom of the fuselage at or near C. of G. should be able to give a reasonably accurate distance to the tarmac.

This type of laser, found quickly with the help of Mr Google, gives an accuracy to 2mm:

Laser Sensor Applications » Blog Archive » Drill Depth Measurement » Laser Sensor Applications (http://www.lasersensorapplications.com/2012/01/drill-depth-measurement/)

Obviously, an apparatus developed specifically for aircraft use would be even more accurate.

I imagine an a/c overloaded by 20,000lbs would compress it's main oleo's by more than 2mm.

It would not be hard to develop tables calculating the "ride height", if you will, for all weight conditions from empty to maximum allowable.

If the laser measurement distance on an overloaded a/c shows less than that allowed for an entered condition.......no pushback.

I thank the "Dambusters", and their downward triangulated spotlights for this idea.

[D.Lamination]

The A350 will have a takeoff acceleration monitoring system.

For me the analog basics say: All commercial airliners are designed for a similar cruise speed off similar runways and therefore have similar wing and high lift device designs.

I know from experience that a 350T aeroplane from any manufacturer going long haul will have a V2 between 170-180 kts. A computer that tells me V2 is 150-something is just wrong or at the very least worthy of serious investigation.

[sb_sfo]

You'd also need to factor in shock strut pressure if you use a laser.

[EEngr]

You'd also need to factor in shock strut pressureRight. And don't forget the static friction in the strut seals.

[DaveReidUK]

You'd also need to factor in shock strut pressure if you use a l@ser.


Tyre pressure too, come to that. I suspect there are too many variables involved for it to be a feasible solution.

[At ease]

I believe the tyre pressure situation has already been looked after:

Aerospace | Wireless Sensing Solutions | Aircraft Tire Pressure (http://www.phaseivengr.com/p4main/Solutions/WirelessSensingSolutionsInDepth/Aerospace.aspx)

If the T/P measurement is within spec, then the crew needs to look at other reasons why the laser "ride height" indication is too low today.

[At ease]

Quote:
You'd also need to factor in shock strut pressure

Right. And don't forget the static friction in the strut seals.Having been no closer to main oleo's on a "heavy" than the seats over the wing(as SLF), can any aircraft engineer give me a general idea of the maximum travel or "stroke" of a large aircraft oleo(only when on the ground , of course)?

Manufacturers of automotive struts/dampers such as Bilstein/KYB provide such information on request and/or on the www. for their products.

I imagine aerospace strut manufacturers(MessierDowty - Goodrich ??) would have such information available to those "with a need to know".

Unfortunately, I am not one of those with such a need.

Can any one help?

Data relating to a 737-900(the subject of this thread) would do nicely. :ok:

Ta.

[Dairyground]

Measuring mass directly or indirectly, whether by weight pads, lg strain, light or radio echo timing, or even Dam Busters-style triangualtion could be useful and would probabably pick up the recent United case, but would not detect the EK Melbourne case, where mass was known but the wrong value was entered into the TO calculations.

What is needed is something de-coupled as much as possible from information capture and data entry processes.

Various critical speeds, the V-speeds, are calculated on the basis of aircraft mass and other factors. Most of these other factors are reasonably well established, such as runway characteristics, including length. The only doubtful value is the mass entered into the calculation. Bad mass gives you bad V1 and V2.

However, the calculation can easily be adapted to give the time needed to accelerate from brakes off to any or all of thr v-speeds, or indeed any other speed.

In the EK melbourne thread I suggested that time to V1 could be used as a gross check on the mass and other inputs to the performance calculation. Nobody, as I recall, suggested that the concept would not work, but there were assertions that monitoring time to speed would be vert difficult to include in the process at a very busy time for the flight crew. I am not convinced that something along these lines would not be possible. Time to some speed lower than V1 might be better than V1 (cooler brakes if you decide to stop) and an automated "drop the anchors" rather than PNF having his eyes on a stopwatch could mitigate some of the workload concerns.

[Arfur Dent]

DP Davies in 'Handling the big Jets' (essential reading for Cathay entrants many years ago') said 'The Captain of a 747 should know his aircraft's weight to the nearest 10 tons!'
Aircraft performance is pretty 'on side' with all the assumptions we make - that's why the 737 subject aircraft and crew survived. Gross error checks are one thing but we can't go down the cabin and weigh everyone as their final abuse before getting airborne - can we?

[DaveReidUK]

Nobody, as I recall, suggested that the concept would not work, but there were assertions that monitoring time to speed would be vert difficult to include in the process at a very busy time for the flight crew. I am not convinced that something along these lines would not be possible.

In the age of ADS-B, where pretty well every commercial aircraft on the runway broadcasts its velocity twice every second and acceleration can be readily derived, I'm surprised that the possibility of an externally-monitored safety-net isn't being considered.

[hetfield]

In the age of ADS-B, where pretty well every commercial aircraft on the runway broadcasts its velocity twice every second and acceleration can be readily derived, I'm surprised that the possibility of an externally-monitored safety-net isn't being considered. AFAIK update rate is one per second.

Don't think ADS-B is a solution for impropper performance calculations. ADS-B relies on GS (time delay!), performance on airspeed.

[DaveReidUK]

AFAIK update rate is one per second.

Nope, definitely twice per second. But even if you only take every other transmission, you would still have more than enough data points to produce a pretty decent acceleration curve.

Don't think ADS-B is a solution for improper performance calculations. ADS-B relies on GS (time delay!), performance on airspeed.

I think you're missing the point. Of course the wind on the nose affects the relationship between groundspeed and airspeed, and hence aerodynamic performance.

But that's not what we're talking about here. Instead, we're using Newton, rather than Bernoulli - i.e. initial takeoff roll acceleration as a proxy to deduce aircraft mass (for a given assumed thrust).

[The Ancient Geek]

Why bother deriving mass ?.
All you need to know is that the acceleration is adequate to achieve V1 and have room to stop within the available distance.

All you need is an accelerometer and a database of runway lenths.
It would be a good idea to take the runway lenth from precision GPS which would cater for cases of wrong runway, intersection and taxiway foulups.

One cheap and simple piece of technology neatly covering all of the major risks with no need for error prone pilot inputs.

If the magic box yells ABORT you obey and figure out why before trying again. The cause might be overweight but there are many other potential causes, all covered.

Always remember the KISS principle - Keep It Simple Stupid.

[ColinBJ]

Mass is still relevant, because if your mass is higher than you expect, not only will your acceleration be lower, but your safe takeoff speed will be higher - a double whammy.

[Vulcancruiser]

I have so much to say here but will keep it zipped for the good of the industry. 99.9% of the Atp's should have caught this before it was any factor at all with a simple check, which you can be sure UAL has now added to the manual. We can all just give a nod of the hat to Bill Boeing and happily go whistling down the street.

I still get a chuckle recalling a story told to me years ago by an A320 FO. He said he had worked his way up through the ranks the hard way. He was once a freight dog down in the Caribbean on a DC-3 outfit. His first day on the job the left seater informed him his job was to go back to the tail wheel shock and keep an eye on it. They would fork in freight until the rear shock compressed to the width of four fingers!........that was the weight and balance.......lol.........I miss the old days..........:D

[Reinhardt]

Some French aircraft have acceleration displayed during take-off roll - so just check value at 60 kts, mentally compare with the one you prepared before and which is written down on the take-off card... relatively easy.
Then now people can make mistake in the calculation of that desired acceleration, the same way they would do for the other values !
Believe it or not, a reason against using acceleration for performance checks, is the number of pilots uncomfortable with the picture, and how it works. Speed, weight and time is usually enough for so many colleagues... if you don't believe, check people around.
Same thing for deceleration when talking about landing : how many can write down the couple of physical formula by heart - and use them ?

[Mimpe]

One friend of mine flying an Airbus for Qantas mentioned the other day he tends to routinely add about 6 degrees C to the temperature on the data input just to add an extra layer of safety. Probably not SOP but it sounds healthy!

[RoyHudd]

Gross Weight is indicated on the A320/330/340/380. Permanently.

You have to be a nincompoop (Emirates Melbourne example) not to use this displayed weight as a cross-check for weight inserted into perf calculations and against the loadsheet. My company use it for TO trim calcs too.

This is important stuff.