News:

SimuTranslator
Make Simutrans speak your language.

City growth

Started by eekizz, November 05, 2015, 06:20:45 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

eekizz

Hi,

I have always been interested to grow my cities and now I am making some kind of experiment using different methods to see, how the city grows fastest.

I am testing which is the best:

1. Airplane, boat, train, bus or maglev
2. Sparse or thick city
3. Hub or not
4. Different terrains: mountain, river, plain or sea
5. Extra roads or no extra roads
6. One bus line or multiple bus lines
7. Which is best in the city: bus, tram, metro, ferry or monorail

All cities have ~5000 inhabitants and are connected with fast train (except the first test)

I am using 120.0.1 nightly rr7571 so it is mostly for my own interest but the results might be applied for other paksets too.

And here is the link to pictures of the experiment:
http://imgur.com/gallery/hpuU2/new

DrSuperGood

I think you are looking for complexity which does not exist. Growth is really simple (for now at least).

Currently (120.0.1) cities will reach maximum growth rate once 100% of passengers and mail is transported (not walked) and all consumer industries that serve the city operate 100% of the time. Cities have 3 growth rates for the hardcoded thresholds of 0 to 999 people, 1,000 to 9,999 and 10,000+. The growth rates come in the form of customizable divisors where smaller means faster growth. Periodic growth is computed off the current month transport performance, so metrics accumulated during the start of the month are worth far more to monthly growth than metrics accumulated at the end.

This is being changed in the upcoming release 120.1 (or recent nightly) in the following ways. Growth is now based on transported and walked metrics. The periodic growth is now based on the metrics which occurred over each period or the metrics from the previous month if no deliveries had to be made in the period so that monthly growth is no longer dependent on when in the month the metrics are accumulated.

Transport type does not matter since it does not factor into growth at all. Utilization of producers and factories does not matter as only input utilization of consumers is measured. Even delivering passengers and mail does not matter as long as you pickup all passengers and mail.

There is still an outstanding issue where multi product consumers still count towards city transport metrics (and so growth) even though one or more of their input products cannot be delivered due to a lack of the appropriate supply chain. This may be fixed eventually if a better industry generation model is introduced.

eekizz

Thank you very much for the answer!

Just checking if there is some dark mystic I have not noticed before. And also this shows me how cities build roads and buildings in different situations, not only growing the population.
And I am really interested how it's going with HUB or not HUB, because in my normal game, I have a feeling that if station is in the city, it will grow faster even though citizens are only passing by. That is why i am trying to avoid HUB's in the wilderness and building them just edge of the city to grow them more.

Combuijs

As for the hub / no-hub situation. My experience (with the current release that is, not the upcoming, and with default settings) is that if there is a hub in the city that is overflowing growth will be hampered. That is a logic to that, passengers and mail generated near the hub won't be able to travel, so less is transported, so growth is less. A hub outside the city does not have this problem.
Bob Marley: No woman, no cry

Programmer: No user, no bugs



jamespetts

An overhaul of the city growth system is planned for Experimental fairly soon (it is the planned major project after next), basing growth on the rates at which local buildings have successfully had their passengers transported (or received enough visitors or goods, depending on the building type) so that growth is not at a fixed rate per city.
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.

Ters

Quote from: jamespetts on November 05, 2015, 12:13:59 PM
An overhaul of the city growth system is planned for Experimental fairly soon (it is the planned major project after next), basing growth on the rates at which local buildings have successfully had their passengers transported (or received enough visitors or goods, depending on the building type)

Sound like what Simutrans already does.

Octavius

Quote from: Combuijs on November 05, 2015, 10:30:02 AM
As for the hub / no-hub situation. My experience (with the current release that is, not the upcoming, and with default settings) is that if there is a hub in the city that is overflowing growth will be hampered. That is a logic to that, passengers and mail generated near the hub won't be able to travel, so less is transported, so growth is less. A hub outside the city does not have this problem.
On the other hand, placing the hub in the city removes a transfer and local train ride for the passengers going to (or from) that city, reducing travel time and increasing the number of passengers, which is good for city growth. At least in experimental.

I always try to have just two tracks in the station, served by several lines, with trains calling every five minutes in both directions on quarter or half hour services. Obviously, you need expansion buildings to handle that amount of lines on just two tracks to prevent overcrowding, but you need less land.

As for the city growth, I haven't thought of all the details here, but would that use the concept of an urban agglomeration? I mean, if a city is completely surrounded by other cities (or water on some sides), it cannot grow, but its surrounding cities would grow instead. I remember that providing electricity to overlapping cities was problematic too. The whole electricity system of Simutrans is rather primitive and unrealistic, and I have some ideas on improving that. I'd have to brush up my C++ coding though, so it will take a while before I could start on that.

Or would it merge cities together if they grow together? (One important difference: two parts of the same city, separated by 10 tiles of water, would be two urban agglomerations.)

DrSuperGood

Experimental's growth is very different from Standard, the two cannot be compared.

In Standard journey time and service fervency does not matter. Number of transfers also does not matter as long as the routing algorithm can still reach the furthest parts of your network.

In Experimental regular services, good journey times, comfort all do matter. Number of transfers does not matter as long as the furthest parts of your network can be reached by the routing algorithm however each transfer does incur a journey time penalty so will make a trip less desirable.

jamespetts

Quote from: Ters on November 05, 2015, 04:17:33 PM
Sound like what Simutrans already does.

In Standard (and, indeed, in Experimental currently) growth is based on the transport to a whole town. What I plan is for growth to be based on the transport in the immediate vicinity of each tile. This will make a very significant difference for larger towns, although will be marginal for the smallest of towns.
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.

Ters

In my experience, cities do not grow where there things are happening, but where there is room for growth, which usually is around the edges. Then those places become the hot spots, rendering the transport infrastructure obsolete.

jamespetts

Quote from: Ters on November 06, 2015, 06:33:07 AM
In my experience, cities do not grow where there things are happening, but where there is room for growth, which usually is around the edges. Then those places become the hot spots, rendering the transport infrastructure obsolete.

This a somewhat incomplete description of city growth: it is historically documented that transport has had a very significant influence in the growth of urban areas: see the Wikipedia article on the Metro-Land concept, for example.
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.

Ters

Quote from: jamespetts on November 06, 2015, 10:42:06 PM
This a somewhat incomplete description of city growth: it is historically documented that transport has had a very significant influence in the growth of urban areas: see the Wikipedia article on the Metro-Land concept, for example.

Yes, it's cars and public transportation that allows cities to grow on the edges, or even beyond them. Although sometimes the public transportation is in place first (which I think I've only heard of from England), the growth was triggered by the potential for transportation, not the actual historical transportation.

jamespetts

In Simutrans, the only way of measuring the potential for transportation is to use historical data (which might be data for very recent history).
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.

DrSuperGood

Growth is also limited by technology and economy. It just was not possible to make tall buildings back in the early 1800. Even if it did, it would not be economical to make them.

Actual city growth is a lot more fluid, with edges being built up and transport provided as needed. Edges are built up because people want to be near the city for work/livelihood and the land is cheap. As they build out wider new industries or shops start up around them and the land becomes more desirable. Smaller homes might be demolished for bigger ones or more valuable ones. New development then continues at the new city edges.

What has changed with city development with technology is that the distances involved have increased. Where as before a distance of 10-20km from home to work was long distance (poorly desirable, several hours of travel) as people walked, nowadays a distance of 100-200km is considered long distance as people drive or use public transport. As such cities are less distributed generally having a central commercial district (highest value land, where most of the trade happens) and then are surrounded by hundreds of kilometres of housing or other industries.

Historically growth has always been based on industrial transport (food supplies into a city, trade between cities) and private transport in the form of walking (to get to work). Today growth is mostly based on road availability for car (replacing people walking) and truck access (industrial transport). Public transport serves as an alternative to private transport.

Good public transport can encourage local growth because it increases an area's desirability (lowers commute time). Area desirability is not directly dependant on physical distance, but rather travel time to get to and from places people need to be. Good public transport reduces this travel time making it more desirable so more people will build nearby. However also a good road system for modern private transport does the same.

I am not sure if it is even worth modelling this in Simutrans. This is something from a city simulator rather than a transport simulator. You would need to use metrics like per-tile desirability. There would be needed for a private car simulation model with road congestion. You would also have to simulate economic limitations to prevent skyscrapers appearing around all well connected and served stations.

jamespetts

There is already a long thread somewhere about how to do town growth modelling for Experimental - most of these things will be (or already are) dealt with either directly or indirectly. Private car transport is already simulated, as is congestion, but to a very high level of abstraction owing to the computational intensity of doing anything else. The interrelation between industrial and passenger transport is intended to be simulated, and per tile desirability (and not just from transport) is necessary. Travel time determines success rates (these data are already created and stored), which in turn drive local growth. The availability of buildings in any given era will prevent Georgian skyscrapers. Density will require some careful thought but is not too difficult to attempt.

It is a mistake, I think, to think that, because Simutrans focusses primarily on transport, city growth is unimportant: in reality, the growth of towns and cities and the growth of their transport networks are so closely interdependent that one cannot satisfactorily simulate one without simulating the other. A city simulator, do not forget, simulates only one city (and simulates its police, fire brigade, health care, water, sewerage, education, parks and recreation and other such things that are not dealt with in Simutrans), and not the transportational relationship between cities. There is nothing in existence yet to my knowledge that simulates city growth realistically in multiple cities as well as the transport between them (and the relationship between the two).
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.

Ters

Around here, there's entertainment and some flats, plus dying shops, in the center, surrounded by pretty static residential areas. This in turn is surrounded by a growing commercial region, with diminishing industry. Beyond this again are suburbs, which are also growing. These have some commercial centers as well.

The city center is (relatively) well served by public transportation, but lack of parking and small city blocks, means that most supermarkets and large retailers establish themselves on the edges of the city. Smallers shops follow, because being close to the big shops gives access to more customers than being near a bus stop or train station. Nightclubs are the only places finding bus stops more attractive than huge parking lots. 24h fast-food shops tag along to cater to those feeling a bit peckish on their way home from a night out partying.

As for what indicates potential transport: The presence of a stop. With connections to where you normally want to go. (Which is perhaps more than can be simulated.) Crowded stops do however detract. Much traffic on a stop doesn't stimulate growth. It's an effect of growth in the past. Though well used stops, perhaps even crowded ones, would make the surrounding tiles attractive for fast-food and convenience stores. Perhaps more so if the stops are crowded.

jamespetts

I should note that car parking is also planned as a feature in a future version of Experimental.  The significance of a nearby stop is precisely that people can get to their intended destination quickly.
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.

Octavius

In the past over here, railways were build in straight lines from one city to the other. Villages could get a station, but generally quite far away from the housing. For example the railway Utrecht - Arnhem opened in 1845 with a station at Ede, located 2 km away from the village's midpoint , which was just 500 m in diameter. Quite a walk or (after 1880) bicycle ride.¹ Nowadays the midpoint of the village is only 1 km away from the station (which hasn't moved) and the village is 4 km in diameter. City growth was clearly influenced by the presence of the station (although soil type also had some influence in this case).

Nowadays when a city sets up a big expansion project, the first thing build is a new railway station on an existing line. For the first few years it's a station in the middle of nowhere, attracting hardly any travelers at all, but as houses are build it begins attracting passengers. If public transport is provided before houses, not only the houses will be more attractive (and therefore more expensive, which is the goal of the estate development companies), but if prevents the new people living there from developing a habit of using a car for everything, a habit hard to break if public transport is provided only later.

There even are three cities, Almere, Lelystad and Dronten, that were completely planned along with their railway and stations even before the land on which they are build existed. Most stations were build at the same time as the railway and remained in a mothballed state until the expansion of the cities came close. One (Lelystad Zuid) has been present in mothballed state since 1988, as expansion of the city went slower than expected.

¹: Most people here think cycling up to 5 km from home to the railway station is acceptable and most people live within 5 km of a railway station. Consequently, stations have room to park anywhere between 100 and 22000 bicycles. It's hard to store that many cars at a station.

colonyan

This is the first time I learned that 3 city growth rate dependent on their population is hard coded.

Rather monotone city behavior was the single reason which alienated me from simutrans ...
After certain level of development, map feels like single large city. Every single one generate demand at same rate. Freight game did help a lot though. Especially when worker departure to industries can affect growth.




Ters

What I like the least about Simutrans' city growth is that it is pretty much tile-by-tile from the town hall. Stops and industries should also function as growth centers. Since Simutrans doesn't model where exactly people live and work, people won't settle down near their workplace, or near stops with easy access there, though.

Quote from: Octavius on November 08, 2015, 02:08:05 PM
[...] a station at Ede, located 2 km away from the village's midpoint , which was just 500 m in diameter. Quite a walk or (after 1880) bicycle ride.

2 km was nothing back then. In fact, over here, first-graders are expected to walk to school if they live closer than 2 km, so it's still pretty much nothing. (As they get older, the distance is doubled. They are eventually expected to ride bicycles, but that is usually only doable in the autumn and mid to late spring. It's technically also doable in the summer, but that's moot.)

colonyan

How about if growth is taken considered as a whole map instead of each city? Whole growth is considered entire map and cities are allocated the growth parallel to their contribution to the whole growth. Cities that contributed most gets larger pie for development. There will a disparities between cities and diversify the tone of growth among cities. The problem is how to.

Octavius

Quote from: Ters on November 08, 2015, 04:52:15 PM
2 km was nothing back then. In fact, over here, first-graders are expected to walk to school if they live closer than 2 km, so it's still pretty much nothing. (As they get older, the distance is doubled. They are eventually expected to ride bicycles, but that is usually only doable in the autumn and mid to late spring. It's technically also doable in the summer, but that's moot.)
2 km is 20 minutes walking or 6 minutes cycling. Not much, but more than nothing. You have to take into account the scale of the country. There always is a village within about 2 km. The fact that Ede has a station is just random. Had the railway been 1 km further south, another village (Bennekom) would have been closer, the station would have been named after that village and that village would have grown. So 2 km is about the maximum possible distance between a village and its station, if it has one.

Children are expected to go to school in whatever way they (or their parents) like. There is no organised school transport like dedicated buses, except for special schools. There are some bus lines running past schools and frequently used by the pupils, but also by other people. There is little snow in winter and most municipalities keep cycling paths clean (The Hague being a notorious exception), so that's no problem.

Ters

Despite the fact that 2 km where I am from was more likely the distance to your neighbor than to the next village (although I've read that we technically speaking never have had villages), I doubt anyone in the mid-19th century considered the 2 km walk to the station unreasonable considering that before that, they had to walk all the way to anywhere. It's kind of hard to think that pilgrims walked from northwestern Europe to Jerusalem back in the days (some still do I think). Or even the fact that Napoleon, well most of his soldiers anyway, marched from France to Moscow within living memory at the time Ede station was build. But the world has shrunk in a very non-linear fashion. Now people consider jet airplanes the only option going from Oslo to Bergen. My great grandfather, perhaps even my grandfather, had to walk 20 km to get to church in his youth (I'm not sure if he did it every Sunday), and 20 km back. (After 2 km, he could still be "at home", depending on where he started.) This was the 20th century, and that was half the distance even earlier generations had to cope with.

An Aussie can of course claim that 2 km is just the distance to the outhouse.