News:

Use the "Forum Search"
It may help you to find anything in the forum ;).

'Bus dwell times

Started by jamespetts, February 11, 2017, 03:12:00 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

jamespetts

There is a very interesting report here giving per passenger dwell times for different sorts of 'buses. The interesting data are at p. 62: boarding time per passenger for "pre-payment" (which I assume means that there is no need to interact with the driver when boarding, so this could also represent a system where there is a conductor) is said to be 2.5 seconds per passenger for a single door arrangement, and 3.3 seconds per passenger for alighting at the front door, or 2.1 seconds for alighting at the rear door.

When passengers have to pay with exact change, the boarding time increases to 4.0 seconds.

To make this usable for a single range of loading times, I suggest the following:

Minimum loading time: 5 seconds (all types of 'bus)

Maximum loading time:

Rear platform with conductor: 2.5 seconds per passenger capacity (e.g., a 'bus with a capacity of 56 would have a maximum loading time of 140)
Front door pay driver: 4.0 seconds per passenger capacity (e.g., a 'bus with a capacity of 56 would have a maximum loading time of 224)
Twin door pay driver: 3.0 seconds per passenger capacity (e.g. a 'bus with a capacity of 56 would have a maximum loading time of 196)
Single door touch card/prepay: 2.5 seconds per passenger capacity (e.g., a 'bus with a capacity of 56 would have a maximum loading time of 140)
Twin door touch card/prepay: 2.0 seconds per passenger capacity (e.g. a 'bus with a capacity of 56 would have a maximum loading time of 112).

Whether to use the touch card/prepay times is a difficult question, as this tends to vary between operators rather than between vehicles (e.g. London with its Oyster card system), and requires investment in infrastrucure not simulated in the game and quite independent from the vehicles themselves.

Long-distance coaches should probably have a significantly higher minimum loading time (e.g. 45 seconds) and a 5-10 second per passenger maximum loading time on the basis of the time that it takes to load luggage.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

I think you could generalise that city buses use some sort of prepaid tickets (electronic or paper) and tickets for inter city buses are sold by the driver.

Sent from my ONEPLUS A3003 using Tapatalk


jamespetts

#2
Except that that is not how it has often worked - in the 1960s and 1970s in the UK at least, conductor 'buses were replaced all over the country with 'buses in which passengers had to pay the driver. In London, this was only changed in the 2000s with the introduction of the Oyster card. Outside London, most places still have people paying 'bus fares in cash and being issued with paper tickets by the driver, I think. The longer dwell times that this process caused was problematic at the time and is a significant feature of road passenger transport.

Edit: On reflection, some of the times above seem to be too high. For example, this source gives a loading time of 2.0 seconds per passenger for a single door 'bus and 1.2 seconds for a two door 'bus with 0.7 seconds for a four door 'bus.

I suggest the following revised figures for maximum loading times:

One door, pay driver: 4.0 seconds per passenger
Two doors, pay driver: 3.0 seconds per passenger

Rear platform with conductor: 2.0 seconds per passenger
Rear platform with conductor (8ft wide): 1.8 seconds per passenger
One door with conductor: 2.5 seconds per passenger
Two doors with conductor: 1.8 seconds per passenger

One door smartcard: 2.0 seconds per passenger
Two doors smartcard: 1.2 seconds per passenger
Three doors smartcard: 1.0 seconds per passenger

Coach: 7.5 seconds per passenger

Minimum loading times

Rear platform: 3 seconds
Doors (any): 5 seconds
Coach: 30 seconds

The "smartcard" options are necessary for 'buses such as the Mercades Citaro, two versions of which are represented in Pak128.Britain.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

killwater

James, these times are so far form the truth... There is no way I can board a bus within 4sec in pay the driver mode. Even when I use a weekly pass that has been bought before it should be at least 5-10 sec per passenger(depends if we assume that the driver checks the ticket or just says ok).  Once people start buying the tickets it takes min 20 sec per person. The minimum loading time should be at least 20 sec - opening doors, checking if everyone is seated or grabbed a rail and closing the door). Boarding the bus in pay the driver mode should have the same time for one and two door configuration as ticket checking is done by the driver and you have to board through the front door. Smart cards are no faster than weekly paper ticket. I would also rise the time for coach 10 sec and min 1 minute.

Pay the driver mode is the reason that in UK I am faster on bike than by bus. Twice faster. In fact my current bus commute takes more or less the same time as walking...

jamespetts

The figures that you give are substantially at variance with the figures from a number of official sources; are you aware of any controlled empirical data suggesting a dwell time of 20 seconds per passenger? That would work out as over 25 minutes to load a large double decker 'bus from empty to full (not overcrowded), which seems a long way off. At four seconds per passenger, a 77 passenger 'bus will still take over 5 minutes to load from empty to full.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

killwater

Well no official data only >10 years of commuting by public transport. Please note that very rarely everyone is buying tickets so people with a weekly or monthly pass will still take only 5-10s to board. It takes me about 20s to get on the bus, buy a weekly ticket and take a seat. Saying Good morning alone takes 2s out of 4s. Also almost never there is a full exchange of passengers on a stop. On Monday mornings it is not unusual to load for 2-3mins as many people buy weekly passes ( buses sometimes start bunching up). Very large loading times are exactly the reason why in many countries you cannot buy the ticket from the driver during the day neither he is checking them (or in some cases you pay an extra fee). Make a little thought experiment and imagine a line of 77 people to the cash register.

Please be careful with using official data - especially standards .  In the buses it is required to have 0.15m^2 of floor area per standing passenger which works out to 30x50cm. Good luck with getting a full capacity (as stated per manufacturer) in real life. This type of information should be treated more as a benchmark to compare different vehicles than a true, real life measure.

jamespetts

The trouble is that studies show that humans are terrible at estimating relatively short times accurately, so empirical data are really needed for this. Also, what is needed is an average boarding time per person (based on empirical data), taking into account those who have season tickets, etc..

I still think that 25 minutes to load a whole 'bus from empty is rather absurd, and it must happen quite regularly that an empty 'bus is filled entirely at one stop: think of 'buses coming off the stand to meet a crowd of people at a busy stop in the direction of peak flow during rush hour.

'Buses do indeed take a long time to load (and note that this calibration exercise increases the maximum loading time on many 'buses from what they were before quite substantially) when passengers have to pay the driver, but, for a 'bus, 2-3 minutes at a 'bus stop is a long time. In off peak conditions, 10 seconds is fairly normal for a quiet stop. If a 'bus stops 3 times per kilometre, then even if it were to teleport between stops, it would average only 6.67km/hour at 3 minutes per stop, which can be beaten by a bicycle quite easily.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

I think killwater is right. Buses that expect to be filled on the first stop and sell tickets to almost everyone have to arrive several minutes in advance. When I go to visit my parents I'll try to check it.

Here in Brno, the city bus drivers sell tickets, but for an extra fee. And even one buyer makes a significant delay.

Sent from my ONEPLUS A3003 using Tapatalk


killwater

Quote from: jamespetts on February 12, 2017, 01:26:08 AM
The trouble is that studies show that humans are terrible at estimating relatively short times accurately, so empirical data are really needed for this.
Good point. It is shame that this thread did not come up 2 weeks ago as I just moved and no longer go on my 4km long commute by bus. Otherwise I could just measure the times for a week. By the way I moved because it took me on average 40min to get to work and I used a direct line...

Quote
Also, what is needed is an average boarding time per person (based on empirical data), taking into account those who have season tickets, etc..

I could agree with 4s if there is no person on wheelchair/with a pram/with luggage, no one buying a ticket, and driver not checking the tickets properly. It would be a low estimate even then.

Quote
'Buses do indeed take a long time to load (and note that this calibration exercise increases the maximum loading time on many 'buses from what they were before quite substantially) when passengers have to pay the driver, but, for a 'bus, 2-3 minutes at a 'bus stop is a long time. In off peak conditions, 10 seconds is fairly normal for a quiet stop. If a 'bus stops 3 times per kilometre, then even if it were to teleport between stops, it would average only 6.67km/hour at 3 minutes per stop, which can be beaten by a bicycle quite easily.

To get 3 min per stop you need 9 people buying tickets. That only happens on Monday mornings/during heavy rain and not at every stop (only the popular ones). On most stops there are 2-3 people boarding top. In off peak conditions the bus do not even stop at half of the stops. At least where I live.

In the previous city I dumped the bus because I was TWICE faster on my bicycle... During off peak it was more or less the same time. And I am definitely not a fit person (and I did not rush to work to avoid sweating).

As for the coaches I noticed that often on Victoria Coach station there is an extra person to load luggage to speed the things up. On the airport transfer coach I use you are supposed to be 15min before it leaves to accommodate loading. If you are late they call you 2-3min before departure to ask if you will make it on time and if you confirm they wait about 5 min. And the tickets are not bought from the driver (either shop or www).

jamespetts

Thank you both for your replies. It is indeed a shame that you are not able to time your 'bus commutes any longer - that would have been most helpful. If anyone is able to find any better empirical data than I have so far (bearing in mind that human estimates are likely to be a long way off), I should be most grateful.

As to coaches, loading those is a whole different affair because of all the luggage: I have a 30 second minimum for those and 7.5 seconds per passenger.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Junna

Useage of paying for tickets issued by drivers was always the norm for both city and inter-city buses the last time I used road-based public transport. I remember before I stopped using buses whatsoever after being accosted by an oaf of a ticket inspector, that the time for issuing tickets was  generally quite fast, I'd say often around 5 seconds per passenger if there was no change needed, but if change was required it quickly picks up. These days one has to buy tickets with the mobile phone-- someone tried to rob the fare box and they reasoned the good idea was to abolish tickets altogether... good thing I don't ride the 'bus no more. The only machine for buying tickets from is in the city centre. Many pensioners weren't too happy.

Spenk009

I think 4 seconds is a very sensible figure. Commuters do their best to board buses without stopping and holding up anyone (waiting to enter and waiting in their seat). I used to enter a bus at its inaugural stop and it would fill more than half with passengers at said stop, to which the bus driver rarely left more than two minutes in advance of the actual departure to allow boarding (mid-sized bus, 30 people, mostly commuters).

A simple trick could be to introduce stops/tiles with a flag that enables the stop to reduce the loading time by a factor. If these stops have a high maintenance but offer lower loading times, enabling a faster movement and higher throughput of convoys at a stop. The vehicle checks the stop for the wait time reduction flag and reduces its loading time accordingly.

An extension to this is setting a flag in the vehicle description as to whether these stop additions have an effect on the individual vehicle/coach.

Slightly more complex would be having the same flag allow and disallow vehicles to transfer passengers. This could mean that a local rail service that uses trains with a conductor will cost more in monthly maintenance vehicle-wise, but will be cheaper to build and maintain stations on the line. Vice versa, a slightly lower loading time and cheaper trains would allow a more dense network to end up being more cost effective. The rail example falls a bit flat with usually the option being to purchase tickets while in transit, and this option being rather off-topic but I'm hoping to prove the concept versatile and an option for widening the sensible/economic options when selecting what vehicles to use.

Vladki

This is interesting discussion, so I tried to measure dwell times on my way to work and back. Here in Brno, the drivers do not sell or check tickets, people can enter and leave the bus/tram by any door. However there is quite some delay when the bus/tram fully stops and door opens, and also before leaving - driver has to ring the bells and whistles, before closing door... Anyway, even with one person getting off/on it takes about 12-13 secs. And it is not much more with 10 persons. The average on my way to work was 18-19 sec, on the way back 23-24 sec. But on the way back the driver was waiting a few times a bit longer (max 1 min 4 sec) as he was ahead of schedule. Longest stop, where more than half of pax (maybe 100) get off and about the same amount gets on was 40 sec (without waiting for schedule). Approx the same time it took to change drivers. Normal stops were with up to 10 pax getting on/off.

Please take the numbers as very approximate, precision of my measurement is quite low. The total length of journey was 10 km, 20 stops (including two requests stops, on which we did not stop today), 26 minutes total journey.

Fortunately (for me) I do not have to commute on intercity lines, where tickets are sold/checked by driver, but that also means, I cannot give you any measurement of those.


jamespetts

Thank you for the measurements. The complexity of this is that this is on a 'bus where the driver has to sound alarms before the door closes, which is a modern thing; on older 'buses with open platforms, of course, this would not be necessary.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

killwater

#14
I agree that you can easily exchange 100 passengers or more in a minute but you need a bus like this:


However I disagreed with times in pay the driver mode. Only one person can board the vehicle at the time while presenting ticket that is checked by the driver. If one has no ticket one has to buy one from the driver while the rest wait. Also majority of buses that I use have only one door so first people get off then the boarding starts. Thus the difference between your journey time of 26min/10km and mine of 40min/4km (this includes getting to the bus stop though so closer to 30min/4km).

Empirical data measured with a watch in the mobile phone. All buses where single front door buses in pay the driver mode off-peak time.


How to read:i - in, o - out, b - buying.
Example: 2i+1o+2b/30s means 2 people boarding, 1 getting off the bus and 2 people buying tickets in total time of 30 seconds. 

2i+1b/30s
1i/8s
2o/8s
1o/10s
1o/6s
1o/8s
1o/6s
1o/8s
2o/9s
1o/6s
1i/9s
1o/8s
2o/8s
2i/18s
3i/19s
3o/11s
1b/15s no change
1b/28s change
4o/14s
1o/8s
2o/8s
1o/8s
1o/7s
1o/9s
1o/8s
1o/7s
1o+1i+1b/20s no change
2o/8s
1o/7s
2o+1i/12s
2i/12s
1i/21s
2i/18s
1b/32s change
2o/8s

Not the best data but certainly something to work with.

Vladki

Quote from: killwater on February 13, 2017, 10:01:20 PM
I agree that you can easily exchange 100 passengers or more in a minute but you need a bus like this:
it was not on bus, but a tram: https://en.wikipedia.org/wiki/Tatra_KT8D5

jamespetts

Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.