Airlines' 'padding' of schedules

[Melburnian1]

In the past some have criticised me when I've referred to how airlines, including QF, VA, JQ, TT, ZL and all the major international carriers routinely add extra time to schedules.

Typically for a flight to (or from) Asia this seems to be about half an hour extra.

Domestically in Australia, it varies according to the route (and sector length) but there's little doubt that it occurs. This falsely boosts what are often these days poor punctuality figures for arriving flights. Have a look at what can be terrible monthly BITRE figures by route for domestic airline timekeeping: on many routes, more than a fifth of some (or most) airlines' flights are 15 minutes or more late arriving at their destination gate despite the generous timetables that are padded.

Padding means we arrive earlier at an airport than we would need to if the schedules were more accurate. It's a 'hidden' cost in wasted time.

Sure, conditions weatherwise are not identical by season (or day) but surely after operating routes for many years, airlines have sufficient information to publish far more accurate timetables.

Compare this to many international high speed rail routes where on some rail systems, timetables are accurate to the minute and require train drivers to operate the train to 'best advantage', as fast as the speed limit for a particular section of track allows. Only some rail operators insert 'recovery time' (padding).

Interestingly, the article points out that timetable padding creates even more inefficiencies for airlines and arguably costs them money:

Why airlines make flights longer on purpose

You can bet this is a subject that companies like OAG will never discuss when they boast that airlines and airports can deliver punctual flights.

 

[trooper]

And you don't think there are some significant differences between rail and air operations??

 

[Melburnian1]

And you don't think there are some significant differences between rail and air operations??

There are, but the bigger point is that the way the airlines insert additional time into schedules is, as the article points out, just a way to make their timekeeping look better than is actually the case.

 

[jb747]

Somehow, I wouldn’t be surprised if the article used your previous comments on this subject as it’s sole reference.

We’ve discussed this previously. Your beloved trains do not have to contend with random routing changes to get to the start of a runway, nor do they have to worry about controllers deciding that a totally new route would be a good idea....simply as a way of wasting time. How would your train schedules handle variations of between 5-30 minutes just getting to the starting gate?

Crew planning is based on flight times. Some crews are paid based on the planned (not actual) times. Crew numbers also vary with flight times. A change of even as little as a minute to the plan can change the structure. Whilst I forget the exact number, the figures chosen for those schedules is (around) the 85th percentile.

Now, I do know of many examples of airlines shortening their schedules compared to what is actually achievable. This was widespread in the days of Ansett and AA. As both airlines perceived some commercial advantage in having a fast flight time, both would print whatever time the other used. So, an Ansett 727 (which was pretty fast) would be listed with the same time as an AA 737 (which was pretty slow). This came to a head when the QF 767 started flying those routes, and we found that the published times were much faster than we could actually fly (and the 767 was pretty fast). So, this was an outright lie to the public...but it was then compounded by the stupidity of basing the turnaround and next departure time on these truncated timings. The upshot was that the aircraft would be late on the first flight, and then proceed to get later all day. As the pilots flying the 767s were not on a post 89 contract, their day ended at 11 hours....and such timings were actually a breach of contract and law.

Have you made an extensive comparison of the achieved gate to gate times vs the schedules?

 

[Melburnian1]

Somehow, I wouldn’t be surprised if the article used your previous comments on this subject as it’s sole reference....

Have you made an extensive comparison of the achieved gate to gate times vs the schedules?

Very funny first comment, but if you read the article you'll see it uses scholarly comments from people who are very qualified on the subject. Importantly, they provide hard evidence. One doubts Australia features at all in the thoughts of those in the UK or Europe who author or contribute such articles. They wouldn't pay any attention to small blogs such as this and wouldn't know (or particularly care) what anyone on an Australian blog thought or wrote. To them, we might be a nice place to visit but they regard 'The City' and 'Europe' (plus New York) as the centre of the universe. We're a pimple as a size comparison.

One doubts very much that flight timings are based on '85th percentiles.' There are so many examples of flights picking up time on a gate-to-gate basis that it must be less than that.

Rather than quoting 30 year old examples as you have, the BBC article deals with present day, and is relevant to Australia and its domestic and international airlines just as it is to Europe.

You've also, if I recall, previously briefly commented that the eastbound times for the Oz to LAX flights had significant excess time built in to them, for the applicable seasonal conditions at the time of the discussion.

The airlines' practice of adding extra time to timetables is dishonest and (even though counter intuitive) costs them money in the end (as the article explains in detail) but is worst from a passenger perspective because it costs businessmen and women time, which to them is $$$.

While the BBC (like the ABC) here can be criticised for some things (political bias is one), when it researches articles on less contentious subjects like transport, it often reaches conclusions that are borne out by evidence, not assertions.

 

[oz_mark]

I'm not sure that basing timetables on historical records is a bad thing. Much is out of.the airlines control.

And if the timetables are so padded, how do you explain all the flights that arrive late?

As for trains, much of the punctuality is designed in, with dedicated lines and infrastructure. Then have a look at the punctuality in Melbourne with its jumbled mess of criss crossing lines. Our airports are more of the jumbled mess that dedicated airways, taxi ways etc...

 

[dajop]

I don't think it's unreasonable, due to the inherent variability in air travel. I am sure there are sound reasons for airlines to do this, due to the cost of extra time in schedule - after all there's capital and crewing costs to consider.

From my own experience, over last 11 years I've travelled frequently between SIN and MEL. Typically SIN-MEL is about 6:30-6:45 flying time, yet I've had it as fast as 6:10 and as slow as 7:00 on a "normal routing", (and once on an abnormal routing almost over DRW, 7:40). Vice versa MEL-SIN is typically 7:10-7:30. Had it as fast as 7:00 and as slow as 7:45. And it does seem to have consistency from week to week, but certainly not month to month, incredibly difficult to plan for this variation.

Most extreme example I have of variation is KUL/SIN, I've been on flights that have spent 29 mins in the air, whilst others have taken 55 mins ()without hold patterns). If you take off from KUL to the south and land in SIN from the north it can be quick. If taking off to the north and landing to the south (or vectored to the south to join the traffic landing from the north), it can take a lot longer.

 

[Melburnian1]

jb747 asked for examples.

Here's one: QF30 from HKG to MEL. In each case I've added the standard estimated five minutes from 'landing time' that the airline shows on its website to obtain an estimated 'at gate' arrival time, which is what matters to passengers.

Saturday 13 April ex HKG: Departed gate (i.e. off blocks) 2007 hours (three minutes early), arrive MEL next morning 0715, 20 early. Net gain 17 mins.
Sunday 14: depart 2044 (34 late), arrive 0826, 51 late. Net loss 17 minutes.
Monday 15: depart 2341 (211 minutes late), arrive 1042 (187 late). Net gain 24 minutes.
Tuesday 16: depart 2006 (four minutes early), arrive 0715, 20 early. Net gain 16 minutes.

So in a small sample, three of four flights gained time on a gate-to-gate basis.

Northbound: QF29:

Sat 13 April: depart 0944, four minutes late, arrive at gate 1655, 25 early. Net gain 29 minutes.
Sun 14: depart 1045, 65 late, arrive 1814, 54 late. Net gain 11 mins.
Mon 15: depart 1045, 65 late, arrive 1755, 35 late. Net gain 30 mins.
Tues 15: depart 0947, seven late, arrive 1741, 21 late. Net loss 14 mins.

So again, small sample, but three out of four flights gained gate-to-gate on the schedule.

This occurs across many airlines, not just QF.

It's pleasing that the BBC - in its scholarly way - has backed up what some of us have known for a long time.

 

[jb747]

So, consecutive days. Which means basically exactly the same weather patterns and routing. Now try it for every Monday over the period of a year.

I would expect that 85th percentile also means that most of the flights should be within the planned schedule.... It isn’t padding...it’s simply common sense in an extremely variable environment.

And yes, my example might be 30 years old. But it’s also first hand...

 

[Melburnian1]

So, consecutive days. Which means basically exactly the same weather patterns and routing. Now try it for every Monday over the period of a year.

But that's one of my points. Schedules ought be adjusted for seasonal weather conditions. Yes, that may have its own difficulties such as slot unavailability at the busier times and airports but surely it's not impossible.

The days of printed airline timetables that applied for the next six months are long past. There are some examples of schedules that vary by day of the week: MU on an Australian route is one I came across last week, so variations are possible even to a country that sees the military take over airspace quite often.

The facts - as set out by the BBC - are that schedule padding (whatever the time of year) is rampant in the airline industry,

I applaud people who have loyalty to their employers or former employers (provided they're not hiding corruption, not at issue here) but my interests are as a passenger, not airlines or airports.

And I'm not having a particular go at QF: this extra time inserted in schedules applies to its competitors domestic and international, plus other airlines that come nowhere near Oz using their own 'metal.'

 

[woodborer]

But that's one of my points. Schedules ought be adjusted for seasonal weather conditions. Yes, that may have its own difficulties such as slot unavailability at the busier times and airports but surely it's not impossible.

But isn't at least some of the schedule based on the agreed slots at the airport?
Not completely up with how slots work, but how much can you move an arrival time around based on a slot you have been allocated?

 

[Forg]

I'm a bit confused over where the perceived problem is with leaving extra time up their sleeves to avoid being late so often.

What's desirable about flights being late more often? Surely if the factored-in delays don't happen, you still get to your destination as quickly as if there were no padding; except you've been given a more accurate target-time for when you need to arrive at your departure point than if that padding wasn't there?

I'm not thinking of this in the airline's terms, just my own convenience.

 

[dajop]

But that's one of my points. Schedules ought be adjusted for seasonal weather conditions.

But how do you account for, to use QF29 as example, that it has 9:02 flying time (ignoing gate to gate time) on March 31, and two days later same aircraft, same flight has flying time of 8:41? You can have variable weather conditions within a month, not just across whole seasons.

Having said that, recent discussion of seasonal variations on Flyertalk in relation to the SFO-SIN non stop on SQ:

Flight 31 flight time suddenly an hour faster?? - FlyerTalk Forums

Singapore Airlines | KrisFlyer - Flight 31 flight time suddenly an hour faster?? - Hello experts, I am ticketed on SQ31 for Sunday, 31 March, taking off from SFO at 11:30am, landing in Changi airport at 7:10pm. I am connecting to flight 502 to BLR at 8:05pm. Tight connection, but I'll take it...

www.flyertalk.com

 

[Forg]

^ DST which Singapore doesn't partake in but LA does?

 

[Melburnian1]

But how do you account for, to use QF29 as example, that it has 9:02 flying time (ignoing gate to gate time) on March 31, and two days later same aircraft, same flight has flying time of 8:41? You can have variable weather conditions within a month, not just across whole seasons....

That conditions vary from day to day is true. I'm a Melburnian!

But even so, for the flight you quote, it's given a gate-to-gate timing of nine hours and 40 minutes (0940 to 1720, with HKG two hours behind MEL at present) so with the typical allowance for blocks off to takeoff of 20 - 25 minutes, and the standard five minutes from touchdown to gate of five minutes (although, yes, at some airports the latter can take longer), it has an arguably generous schedule even if the time in flight is 9: 02 as you state.

Airlines have sufficient data accumulated from years of flying to know typical flight times for a month, or seasons.

 

[jb747]

Airlines have sufficient data accumulated from years of flying to know typical flight times for a month, or seasons.

Which they do. And they do modify schedules, to a degree, over the year. But, any slot change is going to require you to give up what you have, and then attempt to gain a new one. How are you going to account for that. There would be a permanent bidding war....

Quite simply, what is your plan for making this work. It’s a modern era....perhaps we should just vary the schedule based on what happened yesterday. Or we could publish it an hour before departure, when the flight plan is generated. How would you make it work, allowing for slots, crewing, weather and all of the other variables.

 

[Melburnian1]

Which they do. And they do modify schedules, to a degree, over the year. But, any slot change is going to require you to give up what you have, and then attempt to gain a new one. How are you going to account for that. There would be a permanent bidding war....

Quite simply, what is your plan for making this work. It’s a modern era....perhaps we should just vary the schedule based on what happened yesterday. Or we could publish it an hour before departure, when the flight plan is generated. How would you make it work, allowing for slots, crewing, weather and all of the other variables.

Don't disagree that there are challenges.

But my whole point ever since commencing this discussion - well before the BBC article - was that there's such a thing as a 'median' - midpoint - flight time, but airlines consistently place padding in the schedules. The latter isn't the same amount by route (or perhaps by direction as well), but it's there. So in many cases the schedules for the time of year can become faster, as the BBC article implies.

I've even seen examples on Oz where the two major airlines may each be scheduling a flight such as VA679 and QF485 within 20 minutes of each other. Both appear to have a mix of A332s/B738s on this sector, yet in this case the QF gate-to-gate timing is 10 minutes faster than the VA. Granted, 0915 at pushback may be slightly busier than 0935 - but both are on weekdays generally a busy time, as lots of business flights arrive and it's also pretty busy at the MEL international terminal. In this case VA arguably has an even slacker timetable than QF. In the last week, it's generally easily taken less time gate-to-gate than the schedule, and sometimes by a significant amount. That's just one small example domestically.

'Median' flight times have more to do what happens over a period of time than 'yesterday.' The good thing about a 'median' is that it can be based on years of data if the airline so desires, or if it wants to take into account the modern era (as in more flights into or out of airport X than five years ago, but similar to two years ago), the operator can use the data from the last couple of years. It has the information available on its servers. It can split this into months, or what it knows are seasonal weather or traffic patterns.

Airlines have sufficient numbers of data analysts/executives/programmers/liaison staff with airports to know how to do it.

But they prefer at present to falsely pad schedules, and do this irrespective of the time of year. That's not optimal for travellers, and as the BBC points out, not the best for airlines (or airports) as it doesn't drive improved efficiency.

It's one of many reasons why I've never invested in airline shares, irrespective as to how much flavour-of-the-month an airline executive might be.

 

[dajop]

But they prefer at present to falsely pad schedules, and do this irrespective of the time of year. That's not optimal for travellers, and as the BBC points out, not the best for airlines (or airports) as it doesn't drive improved efficiency.

I don't buy this at all. Airlines try to minimise turn around time, to maximise their capital utilisation and minimise their crewing costs.

I am sure they are not padding schedules just for the sake of statistics. If even 25% of flights are running late - there are all sorts of implications, ranging from missed connections, luggage delays, to word of mouth damage to the brand. Not to mention a bunch of annoyed travellers.

I ask what is better - padded schedules with 95% of flights arriving on time or early, or schedules set to the median, which by definition means that up to 50% of flights could run late?

One area I am surprised airlines don't pad schedules - are flights that have durations such as 5:55 or 7:55. If they pad such flights an extra 5 minutes they might just find it easier to sell business class seats to corporate travellers, facing policies that restrict business to flights that are, for example 6 hours or longer (or 8 hours or longer).

 

[jb747]

Don't disagree that there are challenges.

I asked how you would fix your perceived problem.

I've even seen examples on Oz where the two major airlines may each be scheduling a flight such as VA679 and QF485 within 20 minutes of each other. Both appear to have a mix of A332s/B738s on this sector, yet in this case the QF gate-to-gate timing is 10 minutes faster than the VA. Granted, 0915 at pushback may be slightly busier than 0935 - but both are on weekdays generally a busy time, as lots of business flights arrive and it's also pretty busy at the MEL international terminal. In this case VA arguably has an even slacker timetable than QF. In the last week, it's generally easily taken less time gate-to-gate than the schedule, and sometimes by a significant amount. That's just one small example domestically.

A 332 is much faster than a 737....so if there's a mix, which aircraft timing would you use.

 

[offshore171]

I like the padding in the schedule. It allows some contingency and helps you when planning connecting flights.

The analogy to trains is a poor one. Aircraft have many variables to contend with such as headwinds, ATC, congestion, diverting around storms etc etc etc.