News:

SimuTranslator
Make Simutrans speak your language.

Year 1750 - passengers won't travel - route always too slow

Started by japadapa, July 11, 2013, 02:25:04 PM

Previous topic - Next topic

0 Members and 3 Guests are viewing this topic.

japadapa

Hi!

I was planning on starting a superlong testing game and figured I'd start all the way from 1750 :) What could I be doing wrong, as I can't get passengers to like my routes no matter what I do? I'm using pair of horses and stage coach. I've tried to do transport inside a big city, between cities, tried all sorts of distances between stations, but just can't seem to get anyone to travel. Or actually passengers do come at first, but then they start to complain route being too slow. I can't really get any faster with the horses :)

Everything works just fine in later times, like 1990 I can easily get passengers to my routes.

jamespetts

Thank you for your feedback. Passenger transport in the very early days is difficult, as not many passengers are prepared to travel the very many hours that a sagecoach journey would take: this is how it was in real life. You should only be able to sustain one or two coaches travelling between any given set/pair of larger towns on a single route.

It might be helpful if you could upload your saved game for me to see whether the calibration is workable, however, if you literally cannot get any passengers to travel at all.
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.

Jando

I started an new early game with the new release as well, in 1800.

I do get passengers for longe range travel, however I've yet not seen a passenger travelling in the same town. The funny thing is that I have a map where 2 towns (a large and a smaller one) are close together. Passengers take my coach to go between those two although the distance is smaller than other routes in the larger town that see no passenger at all.

Funny with the long range travel. It's per ship. The round-trip journey takes about 3 months, 2 months of those are spend loading and unloading the travelling 10 passengers at the various ports, doesn't matter whether a passenger is actually boarding the ship or not. :)

Michael Hauber

I always start all passenger lines with at least three vehicles.  If a line has only one vehicle the waiting time is guaranteed to be at least as long as the travel time, which means that a horse drawn carriage can do at best half of 15k/h which is 7.5k/h.  When you consider walking time at each end, loading time, and the fact passengers can walk in straight lines that do not follow grids or go around obstacles then a single vehicle horse line is going to have a quite tough time competing with the walking speed of 5k/h.

Jando

Here's a screenshot of my current passenger network (it's small, but is's mine :) http://i.imgur.com/7QwdnfH.jpg

Only passenger routes shown. What I see is that passengers board my horse carriages for non-local traffic, but never for local traffic. For example, people travelling from the northern towns (Cleston and Diding, shown on screenshot) to the southern towns take a city route carriage to go to Cleston Dock to bord my ships, but they will never use a city route carriage for local traffic.

FYI: all local routes are run by 2 single-horse drawn Hackney carriages.

jamespetts

I should point out that, in real life, nobody used horse drawn carriages for passenger transport within towns in 1800. It was the 1820s before this was first done, and only in very large towns - much, much larger than you have on your map (with populations in the hundreds of thousands). For the reasons given by Michael, short distance journeys with horse and carriage are not generally viable.
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.

Jando

Oh, I don't mind that they use my coaches for their long-range itinaries only, the little loss I make with the carriages I can cover with the profit from the long-range ships.

japadapa

Here is a download link to an example savegame in 1750 showing my problem with passengers: https://docs.google.com/file/d/0B71DKgiEp3CebUNXeG50NVFfMVE/edit?usp=sharing

Here is another savegame in 1850 where everything works just fine even tho the situation is the same: https://docs.google.com/file/d/0B71DKgiEp3CedTZHQVp0ZnJ5Z3M/edit?usp=sharing

jamespetts

Thank you for the feedback - this is quite useful. The relevant information is:

1750 scenario
Distance: 7.5km
Travelling time: 45:42h
Waiting time: 48:00h
Total journey time: 1:33:52h (excluding walking)

1850 scenario
Distance: 6.25km
Travelling time: 23:48h
Waiting time: 38:18h
Total journey time: 1:07:00h (excluding walking)

The maximum local distance journey time tolerance is 150 minutes: 2:30:00h for journeys up to 16km; but this is deliberately skewed to the bottom end of the range, so, for local journeys, passengers will very rarely travel for such a long time.

In future versions, I plan to abolish the distinction between local, midrange and long distance journeys and discriminate based only on journey time. Your testing is a useful illustration of why this is important: see also this Wikipedia article.
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.

dannyman

So, the question is, in these early days, do I need to provide in-town carriage service for passengers, or will they walk across town to board the ferry to the next town?


-d

jamespetts

With small enough towns, they should walk accross town if the destination of the ferry is far enough away.
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.

dannyman

Sweeeeeeeeett!!!

Our little pax are growing up to behave like healthy simulated people!

-danny

Jando

What I have seen on a number of early maps now is that passengers will use my coaches for local travel in the first month after opening the route - but then never again. Maybe my service is so bad that they don't come again. :)

Here's a typical screenshot: http://i.imgur.com/aBeld1c.jpg

During the first month after opening the route 115 passengers travelled on the 2 stage coaches between Upton Monument Dock and Berrydale Dairy Stop. In the following months 7 passengers used the route. These passengers arrived with a ship from another town (long range travel).

Edit:

Also a question/observation. At a certain point in my games I frequently get a load of AI horse carts on the streets, but the car ownership value in the city information is always 1. No idea whether it has any connection to the passenger numbers or whethere it's just "fluff" to make the towns more lively. Interesting is that it doesn't happen gradually, it is zero AI coaches and then - bam- suddenly there are a dozen.  8)

jamespetts

Actually, you are not far off about the service being bad and the passengers not coming again: when a new route is inaugurated, there is no way for the game to judge the waiting time, so it is assumed to be the lowest possible level (about a minute). Once routes have been running for a while, the waiting time can be fixed properly, and, that together with the journey time can make the journey too slow for all or most passengers.

As for the AI traffic - this is generated after the end of the first month, which is why you will not have seen them initially. It is based on the number of passengers in the town in which it is generated who use private cars (including horse drawn carriages in the early days). The figure 1 that you are seeing represents the percentage of people in the town with access to private cars: here, it is 1%, as, in the 18th century, only the very richest (one might say approximately 1%, although precise figures are not, to my knowledge, available) had access to their own carriages.
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.

Jando


asaphxiix

the problem is that, while the game is very quick to judge the actual waiting time (which at first is usually quite irregular, with convoys usually released at some location, taking time to fill up the schedule), this waiting time will be fixed for months after the schedule is regular and waiting time has been fixed to a few minutes. During this time, the line will not be used. I suppose the reason for this is the average between the large amount of passengers trying to go the line with 1 minute waiting time (waiting time unknown), but in fact registering very high times at first, and the very small amounts that continue to use the line with registered waiting times exceeding 45 minutes, but in fact waiting very little.

This is especially trouble for new players starting with slow vehicles, who might not be able to understand the situation by checking the waiting times in the stops.

A workaround is to start the line going before stops are built - and build them when all convoys are in place and schedule is regular.

dannyman

How can we tell what passengers of different eras consider as satisfactory service quality?  When a line is losing money for lack of riders, it is a hard sell to INCREASE capacity, but if we have an understanding that we can expect an increase in ridership when we hit a 30 minutes wait time, then it is worth a try to spend precious money upgrading a losing line.

If the answer gets a bit complicated, perhaps this information could even live in the Stop Details dialog.
Passengers:
Next Town (12km) 40 minutes travelling, 40 minutes waiting (Desired: 40, 20 minutes)

In a case like that, I would see that if I wanted to make the line healthy, I might want to double capacity.

jamespetts

Asaph - yes, these dynamics do get somewhat complicated. There is not really any other way of recording the waiting times, however, at least of which I can think. Introducing a good service immediately seems to be a sensible workaround.

Danny - there is no single desired time for passengers. Each packet of passengers has a journey time tolerance, which varies between a minimum and maximum for each of three journey distance ranges, with both the minima and maxima and ranges can be set in simuconf.tab. I am currently in the process of modifying this system to abolish distance ranges and replacing it with a system that more closely simulates Marchetti's Constant, which means that there will be fewer anomalies in this respect (long journeys and short journeys will be treated in the same way for journey time tolerance).
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.

dannyman

Hrmm, why is there a minimum tolerance level?  Is that just to even out the distribution?

My think is that this tolerance would change over time.  In the 18th century, you would expect that a trip to the next town over might take all day.  After all, that is a "medium distance" trip.  Today, this would be a local trip, with a much lower tolerance.

-danny

jamespetts

The minimum tolerance is there so that there are no passengers generated whose tolerance is so short that they have no realistic chance of travelling at all. As to tolerance changing over time, the evidence is quite the contrary: see the link above to the Wikipedia entry on Marchtti's Constant.
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.

dannyman

I read the Marchetti.  It seems highly relevant to the maximum settlement area of towns and their spacing from each other.

But ... Marchetti seems to imply that tolerance is independent of distance.

In our modern world, when we go in for distance travel, we assume that several hours or days will be spent in transit.  If I want to go 2,000 miles to Chicago, I accept that it will take most of my day.  If I lived in the 18th century, if I wanted to go 40 miles to San Francisco, I accept that it would take most of my day.

Similarly, for local travel, if I were 18th century, I would try to keep my pedestrian commute under a mile, whereas in modern California, a 12-mile commute by car takes the same time.

The expectation for trip length may be a fairy human constant, but the distance covered in those trips scales with our technological expectations.  Nobody in the 18th century would ever expect to cover 2,000 miles in a day, and almost nobody in the modern industrialized world would tolerate a 40 mile trip that took all day.

jamespetts

Indeed. That is where alternative destinations come in. There are already alternative destinations for passengers: in the version on which I am working at present, there will be the option for even more alternative destinations (it is currently limited to 16). Passengers will have a fixed number of alternative destinations (that is, fixed per packet on the basis of a random number between a minimum and maximum). If they cannot get to their first choice destination within their tolerance, they try to get to their second choice destination until they run out of alternatives according to the maximum number of alternatives assigned to that packet. If they have not found a good route before running out of alternative destinations, they will not travel; if they have, they will.
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.


Jando

Quote from: jamespetts on July 24, 2013, 07:21:30 PM
...

As for the AI traffic - this is generated after the end of the first month, which is why you will not have seen them initially. It is based on the number of passengers in the town in which it is generated who use private cars (including horse drawn carriages in the early days). The figure 1 that you are seeing represents the percentage of people in the town with access to private cars: here, it is 1%, as, in the 18th century, only the very richest (one might say approximately 1%, although precise figures are not, to my knowledge, available) had access to their own carriages.

Perhaps a newbie question but anyway: is it possible for me to switch off the AI traffic somehow or at least to significantly reduce the amount of AI traffic? Perfect would be when faster transport could pass slower coaches of course.

I notice that lots of traffic gets stuck behind AI horse-drawn carriages, the slower of them travelling at 6 km/h. Those things cause the whole network to crawl at the same speed. Here's a screenshot from a year 1800 traffic jam: http://i.imgur.com/xvVh0Im.jpg, the leftmost coach is running one of my routes, little running involved, it's only stop and go. :)

jamespetts

Yes - you can change the number of the private car graphics that appear by using the citycar_level setting. This is set to 11 by default: if you change it to 0, there will be no private cars at all (do this by pressing the "i" key for the advanced settings tab: you can change the default for new games by editing the "citycar_level = 11" line in Pak128.Britain-Ex's simuconf.tab file).
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.

Jando

Thanks for the quick answer, James. Much appreciated!

Jando

Some observations about passenger travel during the early years (1800/1801 in my case):


  • When I open a new route (Stage coach) the game immediately calculates a travelling time (as seen in the details screen of a stop), however this travelling time is roughly double the time of the actual time needed. Assumption: calculation perhaps based on round trip instead of single journey?


  • After a few weeks this travelling time gets adjusted, in my experience to a more reasonable time. But now the actual travelling time will be a bit longer longer than the time in the details screen of the stop. Assumption: calculation of this new time doesn't take loading times (long for a stage coach!) into account?


  • Waiting times. This actually puzzles me a lot. How are waiting times calculated? What I see is that the waiting time (equally from the details screen of the stop) has little to do with the "real" waiting time of a passenger. I made a test with a route within a city. Waiting time was given as about 30 minutes and I threw 20 or so Hackney carriages on that route. The effect was that at that station there was practically no waiting time anymore (whenever someone came to the stop there would be a coach waiting just for him :) - still the waiting time in the details screen didn't change at all - still 30 minutes although a queue of coaches was ready at that stop.


  • Generally about waiting times. To me it seems it's a concept well suited to a particular type of passenger transport, i.e. regional or inner-city short-range transport, but less suited to long-range transport. After all in real life I don't just randomly go to the station to catch a long-range train, I only do that for a small hop on the tube.


And sorry again for my ramblings.


jamespetts

Thank you for your feedback. The initial calculation of a journey time is based on your convoy's average speed. This will often be an average speed for a short journey inside a town from your stables to your coaching inn, so might well be a lower average speed than can be achieved in service, which is why you are seeing low numbers.

Travelling time does indeed exclude loading time: loading time is taken into account as part of the waiting time instead. This is because the time that passengers spend waiting inside a convoy until it leaves is as much a part of waiting time as the time that they spend at the coaching inn waiting for it to arrive. This might explain what you see with waiting times. The waiting times are calculated periodically, so changes to waiting might well not register instantly.

I do see the point about the interaction between waiting times and longer journeys. However, this is not an easy problem to solve for two reasons. First of all, it is not easy to know where to draw the line between when a waiting time system does and does not make sense (and any such line would result in very troubling anomalies on any journey that was near that line). Secondly, there is no realistic alternative. There is no way of the game knowing in advance when a convoy will be coming, so we cannot fully simulate the effect of timetables. It is thus hard to know what, if anything, could be done about this.
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.

Jando

Quote from: jamespetts on August 05, 2013, 11:09:44 PM

... This might explain what you see with waiting times. The waiting times are calculated periodically, so changes to waiting might well not register instantly.  ...

Many thanks for your reply, James. Much appreciated indeed. I agree that it's hard to draw a line between what could be considered regional/inner-city and long-range travel. Makes sense.

A short comment however to the quoted part about periodical calculation of waiting times. During my experiment with the 20 Hackney carriages on that inner-city route I fast-forwarded the game for several months, no change of waiting time happened during these months - and not a single passenger travelled on that route either. The factual waiting time would have been several seconds (there was always a coach ready to go at the stop), still the detail screen stated 30 minutes waiting time.

I will try to reproduce it.

jamespetts

Ahh, that is the issue: if no passengers are travelling on the route, no waiting time can be recorded, for what is recorded as a waiting time is the actual time that passengers wait (to take account of the fact that passengers will have to wait longer than the spacing between convoys - which may be uneven and impossible to predict anyway - if some convoys are full). Old waiting times are eventually made stale, however, so the waiting times will revert to their default of about one minute (showing as "waiting time unknown").
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.

Jando

Quote from: jamespetts on August 06, 2013, 09:53:40 AM
Ahh, that is the issue: if no passengers are travelling on the route, no waiting time can be recorded, for what is recorded as a waiting time is the actual time that passengers wait ...

Thanks for your answer, James. Is this recorded waiting time somehow hidden from the player? Cause I see waiting times in the detail screen completely static for long times, in some times for more than a month - even on routes where a few hundred passengers are travelling. (I'm not sure whether the detail screen refreshes, I close it and open it new.) Can't believe all these passengers wait for the same exact time, up to the same second. :)

Quote from: jamespetts on August 06, 2013, 09:53:40 AM
...  Old waiting times are eventually made stale, however, so the waiting times will revert to their default of about one minute (showing as "waiting time unknown").

Oh, this didn't happen when I watched it, and I spent at least 6 months or so fast-forwarding the game.

Thanks again!

jamespetts

What you see in the details window (which does not refresh automatically) is the average waiting time. This is recalculated every time that the passengers' routes are recalculated.

Waiting times certainly ought to be made stale - if this is not happening, then this might be a bug. Can you upload a saved game in which this issue can reliably be reproduced?
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.

Jando

Thanks James, I'll try to reproduce it. Sadly I don't have a save of that game anymore.

Blame my addiction for new maps. :) Okay, with new maps I also try to avoid the errors I made on the map before, hehe.

Jando

Quote from: jamespetts on August 07, 2013, 06:34:10 PM
What you see in the details window (which does not refresh automatically) is the average waiting time. This is recalculated every time that the passengers' routes are recalculated.

Waiting times certainly ought to be made stale - if this is not happening, then this might be a bug. Can you upload a saved game in which this issue can reliably be reproduced?

Hello James, once again thank you!

I have now reproduced this behaviour with "wrong" waiting times on another map freshly made with 11.5. Here's a screenshot for illustration: http://i.imgur.com/2yQRPtc.jpg. Waiting times are made stale, it seems after at least 2 months of no traveller, but it has little practical effect.

Here's my interpretation: At the start of a route waiting times are assumed to be zero. This causes a spike in the demand for the route, the first spike in the yellow transported graph on the screenshot. Then waiting times are calculated during that spike - and that causes the demand for the route to drop to zero because the combined effect of slow travelling (in 1800) and waiting time are longer than what the passengers accept. For the next 3 months nobody uses the route, thus the waiting time never changes.

App. 4 months after the initial spike the waiting time is made stale. and what happened at the start of the route happens again. Waiting time is assumed to be zero, causing a spike in passenger demand, that's the second spike in the graph in the screenshot. Subsequently a "new" waiting time is calculated during that spike - but the value of that newly calculated waiting time is more or less the same as it has been before - small wonder, it's based on the same mechanism.

Of course this newly calculated waiting time also has the same effect, demand for the route drops to zero again. :)

[Edited to say that I have a save-game of that map in case you want it.]

jamespetts

This seems more like a design issue than an implementation bug. As such, it is best addressed in the next major, rather than minor, release. Hopefully, what I already have planned for the next major release will help to deal with this issue: the abolition of distance ranges. Passengers will no longer have preconceived ideas about how long that it should take to go any particular distance. Therefore, short to middle distance routes (I assume like yours) will be able to attract some passengers (not necessarily very many, but any is enough to update the waiting times) during the times when there are now no passengers at all. This should help to refresh the waiting times more regularly, and reach an equilibrium rather than the repeated spikes that you are seeing now.
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.