Weight and Balance: How Much Do You Weigh, Miss?
A number of airlines suspended operations during the worst of the pandemic that has disrupted the world over the past year. One of these, a British charter airline owned by a German-Anglo travel and tourism company, were in the process of upgrading their systems, including their reservation software, when the pandemic hit.
The integrated check-in system allowed passengers to check in twenty-four hours before departure and used their information for various paperwork, including generating the load sheet for the flight. The load sheet documents the weight and balance for the flight, listing the weight of the aircraft, the crew, fuel, passengers, and any baggage and cargo. This information is used to calculate the aircraft’s take-off performance, for example what thrust will be required for take-off and the V-speeds.
Passenger weights are done as an estimate, based on an average weight. Most airlines differentiate between adults and children. In the UK, the Civil Aviation Authority requires separate weights for men and women.
Because a correct load sheet is so critically important, there are a number of checks and balances in place to ensure that the load sheet is both correctly calculated and that it reflects reality.
Training meetings for the upgraded software took place in February 2020 and the software worked as expected. Passengers select their own titles when checking in and in the meetings, the titles available were checked against standard usage as set by the International Air Transport Association.
There was no further discussion documented in the training session about what effect the titles had, other than cosmetic.
Flights were resumed on the 10th of July. On the first flight after the upgrade had been put into service, the flight dispatcher noticed something wrong: an adult female passenger had been marked by the check-in system as a child and the load sheet added her weight as 35 kg, instead of the estimated weight for an adult female of 69 kg (77 pounds instead of 152). It turned out that as a part of the upgrade, an error had been introduced into the software, in which any passenger using the title Miss was marked as a child.
They found that two more adult women on the flight had used the title Miss and both were added to the load sheet as children. The load sheet was fixed and the problem escalated.
The report does not mention where the system software was written, other than to clarify that in that country, the equivalent title to Miss is used for a child and Ms for an adult female. Certainly in modern German, I would only use Fräulein for a child or teenager, with Frau for any adult woman, without regard to her marital status.
This was the system that the upgraded software used, with the result that all adults using the title Miss were logged as a child and added to the load sheet at 35kg, almost half the weight of an adult woman.
A fast response was needed, as the software would need changing and testing and rolling out.
The operator set up a team to look at all upcoming bookings and changing all adult females with the title of Miss to Ms, which avoided the issue. As an ongoing fix, two teams were assigned the task of checking bookings every afternoon and evening for the next day’s flights to ensure that adult women were not logged as children by the system. Where possible, flights were checked again on the day of departure.
Check-in staff were also asked to watch for female passengers at check-in and boarding and verify that they showed in the system as adults.
The fast fix, from a software point of view, was to automate what the two teams were doing: change the title of all adult females from Miss to Ms, which was implemented on the 17th of July. However, this fix could only modify passenger information before check-in. Any passengers already checked in were not amended, for example those who checked in online up to 24 hours before the flight.
The teams doing the manual checks did not work over the weekend after the software fix was implemented.
The software was modified again on the Monday 20th of July. There’s no clear information as to whether that stopped the program from working or somehow reset it. That day, online check-in was opened early for flights on the 21st. Female passengers checking in as Miss did not have their status or weight corrected.
On the 21st of July, the day after these modifications, three flights departed the UK with incorrect load sheets. The first of these, G-TAWG, was a Boeing 737-800 on a 5am scheduled flight from Birmingham International to Palma de Mallorca airport in Spain.
That morning, the flight crew reviewed the documentation for the flight. The flight plan gave an expected take-off weight of 66,889 kilograms while the load sheet stated that the actual take-off weight was 64,889 kg, a discrepancy of 1,606 kg or 3,540 pounds.
The flight crew noticed that the load sheet showed 65 children on board as opposed to the 29 which were expected. This seemed high but plausible and explained the difference in weight. As travel restrictions were changing rapidly in July, the captain said he’d noticed loads changing more often than usual as passengers cancelled and scheduled flights at short notice.
He said that it was not uncommon for the expected Zero Fuel Weight (the weight of the aircraft including crew, passengers and cargo) and the actual Zero Fuel Weight to be different.
Another oddity was that the cargo was calculated as 35 bags at a standard measurement of 16kg each and 150 bags whose actual mass averaged 14.5 kg per bag. This was unusual but the Operations Manual was clear that actual mass could be used.
There were thirty-eight adult women on the flight who had been incorrectly estimated at weighing 35 kg each. The actual discrepency was 1,244 kilograms less than the actual weight.
The wind was light and variable. The flight crew selected a five knot tail wind to calculate take-off performance using the load sheet data, which showed their weight as 1,606 kg less than it should have been. This had the following effect:
- The take-off thrust was calculated as 88.3% and should have been 88.9%
- The departure airspeeds (V1, VR and V2) were one knot less than they should have been.
Entering the tailwind helped: when the same calculation was done using the actual take-off weight and a calm wind, as was the weather at take-off, the thrust required was 88.2% N.
The aircraft took off normally and the crew did not notice anything unusual as they departed Birmingham.
The AAIB released the Serious Incident Report this week:
Conclusion
A flaw in the IT system used by the operator to produce the load sheet, meant that an incorrect takeoff weight was passed to the flight crew. As a result, the aircraft departed with
a takeoff weight 1,244 kg more than stated on the load sheet. An upgrade of the system
producing load sheets was carried out to prevent reoccurrence.
When this was discovered, the pilot and the operator filed an Aircraft Accident Report Form. They reinstated the process of manually checking all flights daily to ensure that the title Miss was amended to Ms when used by adult female passengers with a secondary check by Operations staff. They also gave a reminder briefing to Ground Handling agents to watch for adult female passengers showing as Miss or as a child at check in. A formal procedure for a Customer Care Executive, which appears to be a classy term for customer service representative, to check all bookings was put into place on the 24th.
The airline have made a quiet statement on the issue:
The health and safety of our customers and crew is always our primary concern. Following this isolated incident, we corrected a fault identified in our IT system. As stated in the report, the safe operation of the flight was not compromised.
All in all, rather a non-incident which could have been handled better. I found it interesting, how a minor cultural issue can have such an unexpected result. Next week, I’ll post about another case where the take-off weight was calculated incorrectly, this one with fatal consequences, although, of course, that wasn’t the only thing that was wrong.
Speaking as a software person, I would regard it as wildly beyond my remit to assume that anyone with a particular name style fell into a particular weight category without even telling the operators. Someone working on safety-critical software should have such a feeling even more strongly.
Also search for “falsehoods programmers believe about names”.
As another software person, I feel this was a minor requirement buried deep in the specs that never got reviewed. Personally, I would have raised it as a “wait, what? are you serious?” issue, but 75% of the coders I know would accept it in the blink of an eye and throw another case statement on the fire.
I’ve also seen where the developers do raise such issues and get slapped down by management as “it’s in the [holy and not to be questioned] customer spec!” – after a few of those, they go “f*ck it, who cares if the plane crashes everybody dies?” and retire to the pub.
Fortunately, I work for a company that takes concerns raised by the rank-and-file seriously and escalates them, but that’s rare. In my experience, maybe 10% of software development firms do that.
I’ve gotten “We don’t want to upset the customer!” where later the issue bubbled up anyway and the customer was “why the hell wasn’t this brought up sooner?” – in that case, I was fed up enough that I did throw my management chain under the bus.
Almost the same PR boilerplate as the usual bleeting of “the security and privacy of our customers is always our primary concern. Following this isolated incident, we corrected a fault identified in our IT system.” after a serious security or data breach.
It’s the “thoughts and prayers” of the IT industry.
I live in Germany, and if the example of Fräulein is close to what actually happened, it is in current times nearly inconceivable that an adult woman would choose the title “Fräulein”. I say nearly only because I am not a native German speaker and other German countries sometimes have wildly differing customs around titles (looking at you, Austria). But I can absolutely understand this one.
Furthermore, as a software developer, it is probable that the devs were not involved in the translation of (for example, but it’s a good example) Fräulein to Miss. That sort of thing is usually sent out to a translation agency. Then, even if a dev were aware that Miss can be a title for an adult – they may never even have seen it.
Quality Assurance, on the other hand, damn well ought to have.
Right, I suspect a chain of people, each of which examining their own piece of the puzzle. The Register asked Tui where the software was written and they declined to answer, but it makes perfect sense to me in German. It’s frustrating that QA clearly looked at the titles but didn’t spot this aspect. They caught it on the first flight so for it to have then gone wrong again is what really kills me.
I am not sure there would even have been any single developer who would have seen the online check in form and the underlying assumptions and the translations and conversions and finally the data for the load sheet. It’s a lot of pieces of a very large puzzle which each person or even team working on a different corner.
The PR statement is terrible, especially as they’d discovered the issue and then still failed to deal with customer safety. It didn’t really matter in this case but I don’t feel bad at all that the airline is catching flak for it.
I don’t believe that any single person thought that a title should define the weight. However, it would not be odd that the customer wants to know details of the customer without asking and that simple things are thus used, like “Mr” means male and then in a different part o the system, male passengers are assigned an average weight. This seems simple enough on the surface — the miss example stands out to English speakers because the reason for Ms vs Miss isn’t adult vs child but in other languages, that’s exactly how miss is used.
When Donna and I flew from San Pedro, California to Santa Catalina island, they asked us our weight as we entered the seaplane and then told us where to sit.
Right, in small planes this is very common; sometimes there’s actually a scale at the gate but generally pilots just round up. But the margin for error is much less and the hassle and cost (in time and efficiency) of trying to weigh every passenger basically outweighs the risk.
You weren’t trying to convince Donna that we owned it, were you?
There is a reported incident involving SAA in a vaccine pick up flight from Johannesburg recently where it is claimed that the take off weight was understated by 90 tons, leading to an alpha floor event. A340. There is much hype around this event and it would be good to get an impartial view. It is currently under investigation by the SA CAA.
Another reader sent me details of this and I’m going to take a better look, but I was halfway hoping for a final CAA report first, just because that can add a lot of needed detail.
The incident was reported to the CAA late and there is all sorts of finger-pointing going on about special waivers from CAA for the flight due to crew currency issues and more. There is also an claimed failure to obey noise abatement issues on the flight back out of Brussels. I would be keen just to know the consequences of an apparent 90 ton underweight issue in an A340 from an impartial source. Luckily there were no passengers involved – but there is a lot of built up area around the field.
I can agree with Sylvia’s point that the whole issue was a bit “overblown”. In Sylvia words :”…a non-incident which could have been handled better”.
Yes, it was a mistake that could have had serious consequences. It was definitely a “reportable incident”, meaning that filing a report would have been a legal requirement.
The smaller the aircraft, the less room for error when it comes to discrepancies between assumed- and actual take-off weight.
The 737-800 had a seating capacity of nearly 190 passengers (correct me if I’m wrong).
It was also well within allowable weight limits.
What could have affected the flight (as mentioned in this blog):
– A wrong calculation of the take-off speeds, meaning that the aircraft might have rotated at a lower speed, and the take-off speeds (“Vee-speeds”) might have been set a little too low.
In general, an aircraft the size of a 737-800 will; accelerate quickly and it is unlikely that even an engine failure after V-1 would have had any major effect on the performance.
– Power setting: Modern airliners use a reduced power setting for take-off, often called “flexi-power”. In aircraft with engines of an older design, the power setting would be calculated according to the air temperature and pressure and presented on the gauges with a “bug”, indicating the EPR to be used. In later years the minimum engine power setting for a given day would be calculated and used. The actual calculated take-off weight would be used, but the power setting calculated to what would be needed on a day – same runway – when the atmospheric conditions were less favourable, e.g. air temperature. It is a bit more complicated than this, but essentially the actual power setting used would be less than the maximum available. All take-off criteria would still be met and the crew would have the option, if circumstances demanded, to bring the power up to the maximum. Flexi power gave some reduction in fuel consumption (the fuel consumption of a jet engine is very high at low level), it reduces engine wear, especially of the so-called “hot section” and it reduces noise and emissions on take-off.
The latest engines are not controlled directly by the power levers in the cockpit, but the demand is fed into a computer. Not too dissimilar to a car engine with electronically controlled fuel injection.
I don’t know if this is fitted in the 737-800, but the smart systems will automatically bring the power of the live engine up to maximum (and in some types of aircraft even feed in “rudder bias” in case of an engine failure after take-off.
So yes: the “miss”-take was serious enough to report it and of course warranted an immediate software correction, but I don’t think that in this particular case there was any real danger. Fortunately!
In response to Neil Higgs: I do not know what incident he refers to, but the noise level created by modern aircraft engines has been decreased dramatically when compared with the older, low bypass (or even worse: straight jet) engines of the past. In addition, the modern engines are much more efficient and enable a much better climb rate.
In the BAC 1-11 we had to keep a sharp eye on the noise monitoring points as indicated on the SID charts. Making certain that we had enough speed (= kinetic energy), about 10 seconds before reaching it, we would make a sharp power reduction, float as noiselessly as we could past the NMP and before the airspeed bled off, brought it back to full climb power. Then accelerate and retract the flaps. Those of us who are old enough may remember the loud crackling noise of those now “vintage” jets with their trails of black smoke.
Flexi power was not in the manuals and even if, would never be used. One good reason was that we would need all power available to get to as high an altitude as possible over the NMPs.
Interestingly, at some airports (Schiphol was one), most and sometimes virtually all complaints came from a very small number of telephone numbers. One was caught out when he called the control tower again in the middle of the night. It turned out that he listened to ATC on a VHF receiver, but apparently missed a last-minute change of take-off runway. The flight that he complained about had therefore not been over, nor even near his house.
BTW: an underweight issue? Or was the aircraft overweight? 90 ton overweight is a very serious issue, even in an aircraft the size of an A340 and could easily result in a crash. In case of an engine failure after V1 I’d say the accident would have been all but unavoidable.
But underweight?
It brings in mind an old (sick) joke: In a light twin, what is the purpose of the second engine?
Answer: In case of an engine failure after take-off, the second one will bring you to the scene of the accident.
Hi Rudy,
As I understand it, the Vr and V1 speeds on the SAA were calculated on an aircraft weight that was 90 tons lower than it actually was, leading to lower than correct speeds being used.
The load sheet was underweight, the weight of the aircraft was fine.
Speculation as to what happened:
Some developer programmed the logic like this:
This works fine as long as “Herr” and “Frau” are the only titles you can choose in data entry. But then somebody else who worked on programming the data entry portion thought that it should be possible to enter “Fräulein” as title (possibly when the software was translated), but was not aware of the effect this would have on the back-end logic, which would assign each “Fräulein” a weight of only 35kg.
The issue with this is that this kind of software is not considered flight software, and not evaluated to the same standard that computer software running on aircraft systems is (MCAS issues notwithstanding). But even such software on the ground can turn out to be safety critical, as this incident shows!
P.S.: While researching this, I stumbled across an EASA weight survey published in 2009 which suggests that these “regulation” passenger weights are too low anyway, and as the population gets more obese, the discrepancy gets worse.
Mendel,
Yes I can see the problem now. Already before I retired from flying, there were debates about the “standard weight” assigned to males, females and children. It was a particular problem with some nationalities, Americans in general tend to be at least 10 kg. heavier than their “standard weight”. Japanese generally fit well, or may even be lighter.
Another headache is the luggage. Since airlines started to introduce charges for luggage, many airlines were confronted with passengers trying to haul everything with them as “hand luggage” in the cabin.
Climate is another factor: Passengers on a flight departing, for argument’s sake, from Oslo in mid-winter will be wearing heavy clothing. A flight to and from holiday destinations in summer will see them in T-shirt and shorts. Even the weather itself has an influence. If it rains, passengers will wear raincoats and their clothes may be heavier due to absorbed rainwater.
In itself that is not a large discrepancy, it just all can add up.
The problem that we discuss here is exacerbated by the new system of checking in. In the good old days we trundled with our luggage to the airport to join a queue at the check-in desk. The agents were generally well trained and experienced, if not they could call the supervisor. They would instantly realise whether or not a “miss” or “Fraulein” was indeed a 35 kg child, or an adult woman.
Nowadays many airlines (in the halcyon pre-covid days) require that the passenger complete their check-in at home and either print their boarding pass themselves, or download it on a mobile phone or tablet. The link that was the human in the chain was replaced by an algorithm.