The International Simutrans Forum

 

Author Topic: 11.9006 - difficulty starting a passenger network  (Read 2402 times)

0 Members and 1 Guest are viewing this topic.

Offline Michael Hauber

  • *
  • Posts: 36
  • Languages: EN
11.9006 - difficulty starting a passenger network
« on: May 27, 2013, 02:23:18 AM »
Downloaded, and using the Britain 128 0.84 exp pack.

I am finding it very difficult to get a profitable passenger network started.  I am using timeline, starting in 1900, and generating a 768x768 map with 32 cities with median pop 1200.

Horsedrawn carriages seem to make a loss unless transporting to close by cities, even if reasonably full.  At this density level it is very hard to get enough passengers to make a rail service profitable, my best effort so far covers my running costs, but only about 1/3rd of my maintenance costs, which look to be extremely difficult or perhaps impossible to meet without increasing the population or reducing the size of my starting map.  I was able to start networks with a little difficulty using similar parameters on the previous version.

edit:  I now have made a profitable network.  Started with a cheap train line, was making about 80% of my costs.  A few years of population growth got me to almost exactly break even.  Then when I added some horse drawn carriage lines to help feed my train line I started to make an ok profit (currently 18% margin).  Not so much from additional passengers on the train though, but mostly the revenue on local journeys.  Currently the train line connects four cities that have grown to about 15k each, and the local horse drawn carriage revenue in each of two of these cities is larger than the entire train line revenue (at just over 5k/month).  I've made a horse drawn carriage line in another city with about 6k residents and this is currently giving me about 400 in profit per month.

So it appears that the best profit is to be made in local transport, and not long distance as I'm used to. (New to the forum but have been playing simutrans on and off since about 0.82).   Perhaps if I upgrade the train line to the best rail and fastest loco, and increase from one to 3 or 4 trains I might get a big increase in passenger numbers and this would become my biggest profit.  But if not it could send me bankrupt (I'll have to save and then try I guess).  I've noticed a big frequency effect with horse drawn carriages.  Run a single carriage and you get very few passengers.  Increase to four carriages on the line and you almost always find the four carriages more crowded than the single carriage.  This isn't too surprising to me as most journeys are within the 16 tile station coverage radius to allow walking, and it is not easy to get waiting + journey time faster than walking with a single carriage.

I've also noticed some odd behaviour with passengers.  If there is a large pile (say 300) waiting for the local horse drawn line the lot can suddenly switch destinations.  Perhaps this is because route finding tells them they can use the other line with no wait and a longer walk to get to their preferred destination.  But they all switch at once and their wait is just as long on the new destination.  I've also spotted such piles on bus stops near the main train line, which should not be there normally as the station is not a transfer point for any line so should only collect passengers up to its capacity (12) and then stop collecting.  Can the passengers walk en-masse from the train station to a nearby bus stop to try and find a less crowded route to their destination?  I've also noticed that often when watching the statistics for such stations that the happy count will tick over, without additional passengers being added, instead of the unhappy count ticking over.  Do passengers that would walk to this stop change to an alternate stop due to crowding but still register as happy passenger for the original stop?

Finally if I click through the various display options for the passenger waiting list at a stop I seem to only get the 'via' transfer point no matter which option I choose, and cannot seem to find out what the final destination of a passenger is.

Edit 2:  A further oddity in this game: I created a tram line which is a long narrow loop that bends around.  This route has 4 trams and they are all three quarters full of passengers with the same destination, and which appear to never get off.  There are 11 stops on the loop, and no transfer points between with any other line, and the loop is isolated so walking should not be a factor.
« Last Edit: May 27, 2013, 11:10:44 AM by Michael Hauber »

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18753
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: 11.9006 - difficulty starting a passenger network
« Reply #1 on: May 27, 2013, 11:14:25 AM »
Michael,

thank you for your feedback. Firstly, the passenger generation quantities were intended to be realistic: in other words, a town with 1,200 people is intended to produce as many passengers to transport as a real town (village?) of 1,200 people would do so. This is new for this version (previously, the numbers were unrealistically high), so you might find that there are fewer passenger transport opportunities with smaller towns than once there were. To remedy this, generate your map with more realistically sized towns for the era.

However, the passenger generation was also recalibrated in such a way that the "passenger_factor" setting is normalised to a different value. By default in Pak.128.Britain-Ex 0.8.4, it is 9 - in the next release, it will be 15. You can change it to 15 in the advanced settings menu ("i" key > Economy tab > "passenger_factor" setting).

Secondly, I am somewhat surprised that shorter distance journeys are generating more revenue than longer distance journeys. Could it be perhaps that the average speed of your services on the longer distance routes is below the speed bonus speed (which matters more the longer the journey)? It might help if you could upload a saved game so that I can see what the position is.

Thirdly, the re-routing is not a new thing: it has been around for some time. The routes for passengers/goods will periodically be recalculated. Passengers/goods waiting at a station when their route is recalculated will take the new route instead of the old route. If the new route from their current station involves as a first step walking to a nearby station, they will immediately walk to that station and go from there. This can often have effect when the waiting time of a particular stop increases greatly when the service is insufficiently frequent to meet demand.

Finally, I am afraid that I cannot reproduce the bug that you report with the list of passengers waiting at a stop. Can you elaborate on exactly how this appears?

Offline Michael Hauber

  • *
  • Posts: 36
  • Languages: EN
Re: 11.9006 - difficulty starting a passenger network
« Reply #2 on: May 27, 2013, 10:37:57 PM »
Quote
generate your map with more realistically sized towns for the era.

Even today in Australia if you picked a 200km square at random, you would probably get less population then the 50k I'm getting at a median of 1200.  Some such regions have railway lines through them which today are not usually considered viable, but cannot be closed due to political pressure, but presumably where viable some time in the past.  An example would be Roma with a current population of 7k.  In general railway lines in Australia were built for freight not passenger though, and I would think the rail line in Roma was built to transport agricultural produce to Brisbane, several hundred kilometres to the east.

It is a fantasy and part of the appeal of simutrans that I'm sure appeals to many more than just me to start with a basically empty region and turn it into a thriving metropolis.  And there is one real life precedent that fits within the simutrans time-frame - Brisbane was first settled in 1824, and today a 200km square around Brisbane has something like 3 million people.  The town did have to wait until 1879 when the population was 30k for its first railway though.

Quote
Secondly, I am somewhat surprised that shorter distance journeys are generating more revenue than longer distance journeys. Could it be perhaps that the average speed of your services on the longer distance routes is below the speed bonus speed (which matters more the longer the journey)? It might help if you could upload a saved game so that I can see what the position is.
The speed bonus may play a part, but it is the number of passengers on the short distance routes that is generating all the additional revenue.  The train is still the highest revenue route overall, but if I add up the 3 or so routes in each town services by this train I get more revenue in two of the cities.  The train has an average trip time of just over 2 hours, and probably carries about 50 passengers on a typical trip, so that is 25 passengers per hour generated.  In contrast some of my busiest horse carriage routes are on a frequency of about 5 minutes, and are almost totally full.  At an average of say 25 per carriage that would be 300 passengers per hour generated.  And each city has about 3 or 4 such routes.  Considering that a cities I've lived in with roughly 100k population have run bus routes at one hour or worse frequencies, and almost certainly less than 25 passengers in the typically 3/4 empty bus this does seem a bit excessive for a city with population 20k, even considering potentially higher public transport patronage in 1900.

Quote
Finally, I am afraid that I cannot reproduce the bug that you report with the list of passengers waiting at a stop. Can you elaborate on exactly how this appears?

My mistake.  I just couldn't believe that so many passengers could all be heading to final destinations on the same route as the stop with some more searching I find the occasional passenger packet that does have a final destination that requires a transfer and displays as I'd expect.

Lastly, you haven't commented on the apparent bug where passengers on my tram line do not seem to be getting off for one of the stops.

I've uploaded my save game at http://simutrans-germany.com/files/upload/stuck_passengers.sve


edit:  I have an idea about the passengers not getting off the tram - they were bound for a farm, and the tram stop was outside the factory transfer distance.  I built a new stop right next to the farm and a horse carriage to shuttle passengers from the tram line to the farm and all the passengers got off the tram at this station to transfer to the farm.  I would guess that passengers can be generated to an industry according to the passenger connection distance, but can only actually get to the industry using the factory connection distance.
« Last Edit: May 28, 2013, 12:53:33 AM by Michael Hauber »

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18753
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: 11.9006 - difficulty starting a passenger network
« Reply #3 on: May 28, 2013, 11:48:12 PM »
Hmm - I do not know a great deal about urban density and passenger traffic patterns in Australia and how different that they are to passenger patterns in the UK. Certainly, your towns are still really very small by early 20th century standards: no bigger than about 23,000 or so people by the look of things. To put that in perspective, that is smaller than the size of the small town of Tonbridge (pop. 30,340) to-day. That town has a single railway station and about 12 'bus routes, all of which pass through the town to/from neighbouring towns and villages, and many of which are small variations of each other.

To give an idea of the historical size of towns, the UK population in 1901 was 41,458,721 whereas it was 61,634,599 in 2009 - a difference of 0.67 (in other words, to get approximate historical town sizes from the turn of the 20th century, multiply modern town sizes by 0.67). Multiplying 30,340 by 0.67 gets 20,408 - close to the size of your largest towns in that map. In other words, you are playing a map about the size of Kent, Sussex and Dorset combined with a total population of about 17% of that of the historically adjusted population of Kent alone.

Also, your passenger factor is 10 (the default for 0.8.4, I believe); to get realistic passenger numbers for the new calibration system, it would need to be 15 for 125m/tile, but you are playing at 250m/tile (the default for 0.8.4), so you would need to double that to 30 (press "i" > economy tab > passenger_factor) to get the correct calibration.

As to the proportion of local, mid-range and long-distance journeys, these are based on present-day statistics for the UK (at least, 0.9.0 will be, although you are using 0.8.4), but these might not work very well if you have (as you do) a rather unusual layout of towns, or are trying to simulate a different place, such as Australia. You can, however, adjust these settings: press "i" > experimental tab > passengers sub-tab > local/mid range/long dist. settings. It is realistic, however, that by far the majority of passenger journeys are local journeys. I should note that I plan to change this system in future releases such that there is no formal division between local, mid-range and long-distance, but where passengers' destination choices are based on the attraction of the building in question (its level), combined with the journey time to that destination.

The individual profits of the trams and trains seem about right; the only issue is passenger numbers, which I suspect is accounted for by the passenger factor setting combined with your anhistorical town sizes/distributions. How this works will need to be considered carefully, however, and adjustments made to the routing system as necessary; but it is hard to tell how this will pan out with an older pakset version and anhistorical town sizes and distributions.

As to the stuck passengers - thank you for the saved game and furhter investigations: that is most helpful. I have not had time to look into this in detail at present, I am afraid, but I shall look into this when I get an opportunity.

Thank you very much for your feedback - it is most helpful.

Offline Michael Hauber

  • *
  • Posts: 36
  • Languages: EN
Re: 11.9006 - difficulty starting a passenger network
« Reply #4 on: May 29, 2013, 10:06:05 AM »
I've got a passenger network running nicely now.  Two profitable trains, but only just.  This is on a 256x256 map with 12 cities at 2k median pop, so still a very low pop base - the largest city was under 5k.

One issue I've noticed is overcrowding.  Its always been annoying to me that overcrowded stations stop passengers spawning, and that it seems harder to keep stations from overcrowding than trains.  In real life, at least in my home city it seems very much the opposite.  This problem is made worse by large station coverages, twice I had bus lines feeding my train stations become overcrowded, and by the time I'd noticed my trains had been running at a loss for several months as no one was turning up to the stations and the station catchment dominates the city.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18753
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: 11.9006 - difficulty starting a passenger network
« Reply #5 on: May 29, 2013, 10:48:56 AM »
For overcrowding, have you tried extension buildings?
« Last Edit: May 29, 2013, 06:49:19 PM by jamespetts »

Offline Michael Hauber

  • *
  • Posts: 36
  • Languages: EN
Re: 11.9006 - difficulty starting a passenger network
« Reply #6 on: May 29, 2013, 07:17:29 PM »
I use extension buildings where needed, and early game try to do without if possible to keep maintenance costs down.  Current finances:

Revenue 14,184
Running Costs 4871
Maintenance 7488
Op Profit 1805

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18753
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: 11.9006 - difficulty starting a passenger network
« Reply #7 on: May 29, 2013, 07:29:45 PM »
In the next release (0.9.0), the capacities of stations and extensions, as well as their cost, are recalibrated somewhat; it would be useful to know, when that version is released, what the position is.