Sleepover analysis

[Continental Sky]

There was an interesting debate on Discord yesterday whether sleepover is beneficial or not; I am skeptical towards that idea, but passion it was defended with made me to check it closer. I really cannot type that much on Discord, hence a topic here, feel free to move it if inappropriate.

I got the impression that there was a confusion when to use sleepover. IMHO, it should not be used with 7-day-7-a/cs schedule - that schedule is usually done with red-eye flights, at least I do it that way, so it surely cannot benefit if you replace a flying plane with plane sleeping somewhere.

It may be used only as alternative to "regular" scheduling, from 0500 to 2355. My calculations show that one indeed can cram more distance during one day, it is logical and expected, as TAT is included in that idle time over night, but it comes at expense of one route per week, for A-check. Then it depends on specific excluded route whether sleep over earns or loses money. There was a screenshot on Discord of schedule without A-check, but it's something I'm not willing to deploy...

Given, my calculation is incomplete, as I considered only Y pax and standard cargo, for simplicity, but I don't think other classes would change the picture significantly. It also does not take into consideration flight direction - I took first routes I found to fit in schedule exactly (actually, with 5 minutes offset, one TAT per day is 5 minutes longer than my default, I didn't have patience to look for routes that fit exactly); depending on direction, there are probably routes where longer distance can be covered within same flight time, but it pushes my analysis too far.

Anyhow, here is the story; I wanted to check A300 combi from my fleet, it is close to full day. I prefer to avoid delays, so my TAT is usually 5 minutes more than 1% delay probability, which is 2 hours for this bird.

Let's say it carries 100 pax and 7 t of CS per flight.

Here is my schedule:



Now, if I move longer route, DEN-PHL, to sleepover, I indeed can fit longer route into a daily schedule, DEN-YYC which is 780 nm, instead of DEN-DAL, which is 566 nm; however, I have to drop one route per week, obviously the shorter one, so that increase in distance is not sufficient to cover the missing route, I'd lose nearly 5% of weekly income.

However, if I keep shorter route, DEN-DAL and set it to sleepover, then the picture changes - sleepover scenario earns 3% more than my original one!



Now let's see what happens with 3 routes on smaller airplane, A320-200, with 75 min TAT (1% delay probability is at 70 min), 130 pax and 2 t of CS.

Sleepover scenario yields bigger distance covered, too, but here it does not matter which flight, longest or shortest, sleeps over, in both cases additional distance is the same, and not sufficient to cover missing route, sleepover scenario loses money.



So I'd say financial picture is not black and white, it depends on a/c, distance, flight direction, ticket prices, TAT, etc. When sleepover is required because of slot lock in home base, it is certainly beneficial; but for financial benefit, it should be first examined.

Here is Excel file, if anyone else wants to dabble with that further, or look for possible error in my calculations... 

https://1drv.ms/x/s!ApixNzy4AI0gn45b-ZYiGZuZ_RHWmA?e=WBwnhb

[groundbum2]

that is some hard core Excel, I salute you and kneel before you! I keep some data in Azure/AWS Sql/Server cloud databases but your front end is far far prettier than my grubby efforts..

Simon

[tungstennedge]

I got the impression that there was a confusion when to use sleepover. IMHO, it should not be used with 7-day-7-a/cs schedule - that schedule is usually done with red-eye flights, at least I do it that way, so it surely cannot benefit if you replace a flying plane with plane sleeping somewhere.

This is absolutely not true. First of all, I love your detailed and thorough analysis of the senario you proposed, but the initial assumption that sleepover are not worth 7 day scheduling is incorrect. Lets take a 737/a320 with a 40 minute minimum TAT as an example.

Suppose I fly 7 planes, single day scheduled from 5:00 to 23:20 every day. I can also fly between 5:35-23:55. The point is, every day, my plane spends a time of 5:40 minutes on where it cannot fly no matter what, for a total of 340 minutes on the ground everyday, and 2380 minutes per week.

Now suppose I sleepover flights for 6 days of an entire seven day schedule, and on 7 day I land at 23:20 so I can start my weekly A check at exactly 00:00 Oclock so I can take off at 5 the next morning. Every sleepover flight spends 305 minutes on the ground 23:55> 5:00, and on the seven day with the a check, spend 340 minutes on the ground. Now the average is 310 minutes per day, and 2170 minutes per week.

As we can see, seven daying sleepovers is better than single day scheduling. It is not however, better than seven daying non sleepover, 5:00>23:55 scheduling. However, usually sleepovers are done in conjunction with seven day schedules to save 5:00 slots at your main base as these deplete first in markets without red-eyes available. In addition, I personally believe that a (2380-2170)/10080 = 0.02083333333 = 2% efficiency gain is not worth the trouble of finding exactly the correct time routes to make schedules minute perfect.

Picking better routes is more worth it in the first place. Also keep in mind, this 2% efficiency gain only effects two costs, staff and leasing+deprecations costs. For my airline this represents 27% of my expenses, meaning the actual effect on my company is about 0.5% to my actually expenses, or 0.35% of my revenue at a `~30% margin.

0.35% is basically a rounding error, but nevertheless the efficiency gain is definitely there. Also note, that if using longer minimum TAT aircraft, these efficiencies can grow to significant amounts. In cases where players are using 747-400D's to save OOB aircarft, the savings are 110 minutes per sleepover. Over 6 days this would result in increase of usable time by 6% which is definitely noticeable.

[gazzz0x2z]

+1000 with Tungstennedge.

I'd add that sleepovers also offer some 7-7 opportunities. In a base like ALC where nearly everything worth flying an A320 is 800NM away. Which causes 2 problems :

(1) You can't use red eyes
(2) You lose a lot of plane use flying 2 routes a day, but 3 routes a day does nto fit

So the idea is to fly two routes and a half each day (and use leftovers to fill the 7th day of the week), with one sleepover every 2 days. And suddenly, ALC becomes a far better base. And don't launch me on LCY with this nasty 06-22 curfew, where sleepovers are the only way not to lose money.

I like sleepovers overall because they add this kind of flexibility, sometimes, you can't fill 0500-2355, but by adding sleepovers, you can far better fill the schedule. Which means that ultimately your gain is far better than the ideal case counted by Tungstennedge. It's not always as clear as in ALC or LCY, but it's definitively a tool in the toolbox that helps improving  planes use. A tool amongst others.

[MikeS]

Posts like these make me question my commitment to the game..... just when I thought I spent too much time on AWS 

[Meddix]

2 other arguments that are not specifically mentioned for using sleepoversin a 7 days schedule even for small and medium planes

- Lack of slots between 05.00 and let's say 07.00
- High demand routes where you like to schedule a plane leaving let's say every hour of the day

[Amelie090904]

Now this is some extensive research, but please don't forget that these tests happened under ideal conditions (which rarely exist in an actual game). You will rarely find routes that just fit perfectly into your schedule.

The reason to do sleepovers, in my opinion, are as follows (from most important to least important):
1) Adding frequencies at different times: In China (where you have about zero redeye opportunities), most routes have a demand of 1000 or more. If you wish to stick to 60min intervals (or even 30min), you just need to find different departure times. The evening hours are perfectly suitable. You could depart at 2200, land at 2355, and fly back in the morning. This means I have way more opportunities when it comes to adding frequencies compared to only flying in the morning or throughout the day.
2) Avoiding slot locked morning hours: In airports where you cannot have redeyes, morning slots are the first to disappear. In China, you will often notice that there simply are no slots from about 0500 to 0900. To avoid departures during these times, sleepovers are crucial. You land 2355 at a foreign airport, fly back 0500, land back at your base at 0800, and fly at around 0900 to your next destination. This is how you keep expanding when you got no morning slots.
3) Saving time: By incorporating the turnaround time into the "sleepover time", you save ~60min (depending on airplane type) which means you got more time during the day to fit in more flights. This can be crucial if you can fit in 2 daily flights or 3 daily flights. This is also important for your OOB as you want to maximize your plane utilization.
4) Saving money on slots: This is mostly important for the early days of a game world when you have not too much money. By paying morning slots at the foreign airport (which should be cheaper than the slots at your HQ), you spend less money on slots and can expand quicker overall.

All in all, it's not so much about how economical it is, but how much you can optimize your schedules to add frequencies, save on OOB  and to avoid slot issues...

[swiftus27]

All facts that were explained but unheard

[groundbum2]

now if only we could do "A" maint down route....!

I also suggest a new achievement - the person who does the most hard core data analysis in a gameworld.

Simon

[Meddix]

That is a complete summary Andre !

[Continental Sky]

Thank you for the feedback, guys. Interesting read from Andre, which I would say is more or less on the same wavelength as my two cents - primary reason for sleepovers is not direct financial benefit from saving that one TAT, but rather indirect benefits from avoiding the slot locks, and also adding routes at different times, which I didn't think of - I have never had such an issues, never flown in China. 

Suppose I fly 7 planes, single day scheduled from 5:00 to 23:20 every day. I can also fly between 5:35-23:55. The point is, every day, my plane spends a time of 5:40 minutes on where it cannot fly no matter what, for a total of 340 minutes on the ground everyday, and 2380 minutes per week.

I'm not sure I understand this part.  I thought famous 7 days / 7 planes schedule referred to schedule like this:



In this case, I don't see how can you fit sleepover here, and why would you do it at all - there's maybe total of 20 minutes per week wasted here, it's difficult to pack routes more tightly than this. Without reducing TAT beyond acceptable, of course.

But if you have 7 planes that fly from 0500 to 2320 or 2355, whatever, it is just classic single day schedule, isn't it? Regardless if there are 7 or 17 or 27 planes - if they complete they daily rotations at 2355, they can start over next morning those same rotations. Why do you need 7 planes to combine routes between them, when each plane can continue her own schedule every day?

Posts like these make me question my commitment to the game..... just when I thought I spent too much time on AWS 

Oh, tell me about it...  Let me tell you my story: I had had pretty stable and longstanding airlines in The Age of Flight and The Speed World, I was really satisfied with them, around 150-200 planes, the ones I liked and from medium airports, not from big ones, I played as I enjoyed it, not purely to climb the ladder, and I learnt a lot new things about planes along the way.

Then Modern Times started, I devoted a lot of time to it, built a solid regional airline in Athens, I had had ~65% of slots and transported pax there with ~50 planes, B734 and Saab 2000, all B734s were in a tight 7 day schedule like the one above, 15-20 routes for each set of 7 planes. Then I realized that I indeed spent too much time on those 3 GWs, I tried to decrease the time, but didn't manage, it was like an addiction, dammit... and then one day when I was tired and angry, I decided to close all three airlines, and did it in a blink of an eye... Sure enough, after a week or two, I regretted it, and now I'm building my airlines from the scratch again... 

As for the excel in my post above, come on, it took 15 minutes to find routes that fit the schedule and 15 minutes to insert it into Excel, nothing special... I see now that I wrote EUR instead of $ at the summary rows, lol...

In The Speed World, I successfully operated seven A388s, I was the only one to fly A388, and I realized why, the profit was really tight, I had to follow and hedge fuel carefully,  and had to monitor routes - as soon as competition brought supply close to 100%, I had to look for new route, as increase of fuel prices and decrease of LF easily sent their profits to red. But I enjoyed having them in the fleet and even made a livery for them, and that's the point of the game!

[DanDan]

my two cents on this:

- when doing route planning, i try to stick to 7 day schedules in any case. even when there arent any overnights. especially route image depletes if just one plane is flying a specific route and needs a c/d. when less than 7 planes are remaining, i try to let them shuffle the routes.
- redeyes are better than sleepovers and preferable, even if sometimes at just 50-60% demand/supply; any hour in the air counts (but maybe i take that a bit to the extremes)
- sleepovers are great when you are based at an airport with a long curfew, especially when flying to airports that you can arrive at late and leave early and when at slotlocked places
- it certainly makes sense to try for sleepovers instead of having the plane on the ground for another hour or two during a day.

so really sleepovers only make sense when you have a lot of shorthaul demand where no redeyes can be made instead and/or in highdensity route situations, where you want to make 18 flights a day.

p.s.: i do supply/demand in spreadsheets, i do 7 day scheduling in spreadsheets, i do fleet planning in spreadsheets, but i never filled my database with any price information! kudos to you!

[AngryOpossum]

The original argument being made was that you get a “magic hour” of time with a sleepover that you don’t get by keeping your plane at base overnight. This should be obviously false; there will always be 24 hours in a day. If you plane lands at 2355 and takes off at 500, that is 305 minutes on the ground whether it’s at your base or any other airport. And the turnaround time is included in that 305 minutes, regardless of the airport. This shouldn’t need an example, but since it seemed to be difficult to understand for some, here you go:

Let’s say you need to schedule two round trips from your base, A. One is A-B with a 1 hour one-way travel time and A-C with a 3 hour one-way travel time, using a one hour turnaround time in both scenarios. Taking off at 2055, you can schedule this two ways:

Sleepover:
A-C 2055-2355
C-A 0500-0800
A-B 0900-1000
B-A 1100-1200

Stay at base:
A-B 2055-2155
B-A 2255-2355
A-C 0500-0800
C-A 0900-1200

In both cases, the plane is flying the same number of routes and the same distance in the same amount of time (2055-1200*). There is no “magic hour”.

And this is all dependent the routes being flown. If route A-B were 2 hours travel time in the example above, you would have to do a sleepover, as you couldn’t complete A-B-A from 2055-2355. So of course the sleepover is preferable. Alternatively, if A-C were 4 hours travel time, you couldn’t complete A-C from 2055-2355 so the stay at base scenario is preferable. While I nearly always do 7-day scheduling, I still schedule both sleepovers and staying at my base depending on what’s appropriate. But I don’t pretend I get a “magic hour” when my plane is on the ground from 2355-0500 at another airport instead of my base.

Now, there are other arguments being made in favor of a sleepover such as slot usage. Sure, if airport A is slotlocked at 0500, the sleepover scenario is preferable. But the converse is also true: if airport C is slotlocked at 0500, the stay at base scenario is preferable. If you have a curfew of 2300 at A, the stay at base scenario is impossible. But if there’s a 2300 curfew at C, the sleepover scenario is impossible. I understand these arguments, and they may be a valid reason for a sleepover. But these arguments work both ways and may at times argue for keeping your plane at your base.

Finally, I like Continental Sky’s analysis, but if there is competition on a route it underestimates the detrimental effects of a sleepover. Let’s say the demand on his routes is 130 pax/day, and two players are flying an A320-300 (130Y) on it with all else equal. Now, the person with the sleepovers (let’s call him player A) is flying 65 pax/flight every single route he flies. The person without sleepovers (player B) is running each flight 7 days a week, so player B gets 65 pax/flight six days a week and 130 pax/flight one day per week on the route player A isn’t flying – an average of 74.2 pax/flight for that particular flight. There will be a similar bump in cargo load factors. Now, both sleepover scenarios lose 7.4% per week to the non-sleepover (see attached excel edits). This may overstate the case a bit since the route would likely be not flown on Saturday, when demand is lower. But regardless, I love it when I go to a new base and see someone flying a route 6 days a week. All of his flights will have competition, whereas 14% of mine won’t.

EDIT: The first scenario, while getting more passenger revenue, also burns less fuel (10% shorter mileage flown).

[Amelie090904]

Let's shorten it.

Example route:
Flight length per leg: 1h
Turnaround time: 1h

With sleepover:
A-B: 2255-2355 (1h)
Turnaround time: 2355-0055 (irrelevant as it happens during the night when the aircraft "sleeps" until 0500)
B-A: 0500-0600 (1h)
Next possible departure: 0700 (1h)
Total use time during "usable" daytime: 3h

Without sleepover (random departure time during the day):
A-B: 0500-0600 (1h)
Turnaround time: 0600-0700 (1h)
B-A: 0700-0800 (1h)
Next possible deparutre: 0900 (1h)
Total use time during "usable" daytime: 4h

The advantage is obvious. With sleepovers, the turnaround time is included in the "sleep time" of the aircraft. You'd make use of the night time to turnaround the plane which means you have 60min more available during the day. It's not revolutionary much, but can be helpful to fit in tight schedules and/or longer routes.

EDIT: @AngryOpposom:

This also applies to your example:

Sleepover:
A-C 2055-2355
C-A 0500-0800
A-B 0900-1000
B-A 1100-1200

Stay at base:
A-B 2055-2155
B-A 2255-2355
A-C 0500-0800
C-A 0900-1200

Let's focus on the A-C route.

Sleepover:
A-C 2055-2355 (3h)
C-A 0500-0800 (3h + 1h turnaround time until next departure)
= 7h total usage during the day

Stay at base:
A-C 0500-0800 (3h + 1h turnaround time)
C-A 0900-1200 (3h + 1h turnaround time)
= 8h total usage during the day

[Igolyto]

The only thing that shows the analysis is that if you want to make a sleepover you need to do a 7-day schedule. The difference between the two methods in the analysis is that one of the flights (the sleepover flight) has one less rotation per week and therefore produces a smaller profit on a weekly basis. It makes sense as the 7th day is the B Check night that is lost.

The 7 days schedule helps even if you do sleepovers as it increases aircraft utilization as you schedule flights one after the other. You can maybe have a few aircraft with good weekly utilization without sleepover but you will always have some odd ones leaving one hour or so as you lack flight with the right distance to make it work.

All in all, it offers more possibilities than the classic scheduling even if you risk tendonitis (or the AWS elbow  ) by doing so.

[AngryOpossum]

Let's shorten it.

Example route:
Flight length per leg: 1h
Turnaround time: 1h

With sleepover:
A-B: 2255-2355 (1h)
Turnaround time: 2355-0055 (irrelevant as it happens during the night when the aircraft "sleeps" until 0500)
B-A: 0500-0600 (1h)
Next possible departure: 0700 (1h)
Total use time during "usable" daytime: 3h

Without sleepover (random departure time during the day):
A-B: 0500-0600 (1h)
Turnaround time: 0600-0700 (1h)
B-A: 0700-0800 (1h)
Next possible deparutre: 0900 (1h)
Total use time during "usable" daytime: 4h

The advantage is obvious. With sleepovers, the turnaround time is included in the "sleep time" of the aircraft. You'd make use of the night time to turnaround the plane which means you have 60min more available during the day. It's not revolutionary much, but can be helpful to fit in tight schedules and/or longer routes.

EDIT: @AngryOpposom:

This also applies to your example:

Let's focus on the A-C route.

Sleepover:
A-C 2055-2355 (3h)
C-A 0500-0800 (3h + 1h turnaround time until next departure)
= 7h total usage during the day

Stay at base:
A-C 0500-0800 (3h + 1h turnaround time)
C-A 0900-1200 (3h + 1h turnaround time)
= 8h total usage during the day

You are cherry-picking. The plane, in reality, is flown for the whole day. Let's take your scenario to its conclusion and schedule the whole day with 1 hour routes and 1 hour turnarounds:

With sleepover:
In 0500-0600
Out 0700-0800
In 0900-1000
Out 1100-1200
In 1300-1400
Out 1500-1600
In 1700-1800
Out 1900-2000
In 2100-2200 (55 min turn)
Out 2255-2355 (305 min turn at outstation)


Without sleepover:
Out 0500-0600
In 0700-0800
Out 0900-1000
In 1100-1200
Out 1300-1400
In 1500-1600
Out 1700-1800
In 1900-2000
Out 2100-2200 (55 min turn)
In 2255-2355 (305 min turn at base)

Both planes are flying 10 routes and have the same exact same idle and flight time. There is literally no difference except where the plane is spending the night. You're just making up a difference in your head.

[Continental Sky]

Let's shorten it.

Example route:
Flight length per leg: 1h
Turnaround time: 1h

With sleepover:
A-B: 2255-2355 (1h)
Turnaround time: 2355-0055 (irrelevant as it happens during the night when the aircraft "sleeps" until 0500)
B-A: 0500-0600 (1h)
Next possible departure: 0700 (1h)
Total use time during "usable" daytime: 3h

Without sleepover (random departure time during the day):
A-B: 0500-0600 (1h)
Turnaround time: 0600-0700 (1h)
B-A: 0700-0800 (1h)
Next possible deparutre: 0900 (1h)
Total use time during "usable" daytime: 4h

The advantage is obvious. With sleepovers, the turnaround time is included in the "sleep time" of the aircraft. You'd make use of the night time to turnaround the plane which means you have 60min more available during the day. It's not revolutionary much, but can be helpful to fit in tight schedules and/or longer routes.

What is "Total use time during "usable" daytime"?

Actually, I'm confused now, your post shows that sleepover does not give any advantage, look:

- With sleepover, you depart at 0700 and must land in 2155 (to allow departure at 2255); hence, total of 14:55 available for other flights during day.
- Without sleepover, you depart at 0900 and must land in 2355; again total of 14:55 available for other flights.

So in both cases, you have route A-B/B-A plus 14:55 for other routes; the same time available, isn't it. Except that with sleepover you must drop one flight for A-check...

[AngryOpossum]

Let's shorten it.

Example route:
Flight length per leg: 1h
Turnaround time: 1h

With sleepover:
A-B: 2255-2355 (1h)
Turnaround time: 2355-0055 (irrelevant as it happens during the night when the aircraft "sleeps" until 0500)
B-A: 0500-0600 (1h)
Next possible departure: 0700 (1h)
Total use time during "usable" daytime: 3h

Without sleepover (random departure time during the day):
A-B: 0500-0600 (1h)
Turnaround time: 0600-0700 (1h)
B-A: 0700-0800 (1h)
Next possible deparutre: 0900 (1h)
Total use time during "usable" daytime: 4h

Yeah, Continental Sky said it much more clearly in another way than I did.

With your sleepover schedule, your next departure is 07:00 and you must land at 21:55 (14:55 of usable daytime). With your without sleepover schedule, your next departure is 9:00 and you must land at 23:55 (14:55 of usable daytime). They're literally the exact same.

[AngryOpossum]

EDIT: @AngryOpposom:

This also applies to your example:

Let's focus on the A-C route.

Sleepover:
A-C 2055-2355 (3h)
C-A 0500-0800 (3h + 1h turnaround time until next departure)
= 7h total usage during the day

Stay at base:
A-C 0500-0800 (3h + 1h turnaround time)
C-A 0900-1200 (3h + 1h turnaround time)
= 8h total usage during the day

You're confusing yourself by comparing apples and oranges. You are comparing a 20:55 departure to a 05:00 departure without considering what happens the rest of the day. That's why my example has them departing at the same time.

The result is the same with this example. Next departure for sleepover is 09:00 and you must land by 19:55. Next departure for "stay at base" is 13:00 and you must land by 23:55. Both of those are 10h 55min of additional flying time during the day.

[swiftus27]

yet we still ignore the other glaring obvious positives and focus solely on the turn time question.     That's like only worrying about fallout after a nuclear war.