Topic: [Experimental release] Pak128.Britain-Ex 0.1

[jamespetts]

A preview and testing version of the Simutrans-Experimental version of Pak128.Britain is now available: download it here. This is a complete pakset and should be installed in its own directory. It is only compatible with Simutrans-Experimental - the normal Pak128.Britain should be used with Simutrans-Standard.

This is a preliminary version meant for testing and evaluation, although it should balance better with Simutrans-Experimental than the standard Pak128.Britain. Not all of the Simutrans-Experimental features are implemented yet (although I am planning to implement them all in due course). Here is a list of the features supported by this release of Pak128.Britain-Ex:

• Comfort
• Loading times
• Catering
• Upgrading (partial)
• Way constraints
• Reversing
• Tilting trains
• Overcrowding (only on a few vehicles so far - work in progress)
• Industry obsolescence

See here for a more detailed description of the work outstanding and so far completed, and also for information on how to get involved in contributing to the project and obtaining the sources.

Any feedback on this pakset variant (especially relating to the gameplay balance) would be very much welcome.

[jellyman]

Just tried downloading and playing.  Couldn't get any profitable routes happening.  Using default settings, turning timeline on starting in 1830.  Tried single platforms and horses, and was paying about 10 times as much maintenance as I was getting revenue.  Tried downloading the complete pakset and installing to fresh directory just in case something was weird with config files and same result.

[jamespetts]

Jellyman,

thank you very much for your feedback - that is most useful :-) Can you upload your saved game so that I can test it to see where the trouble is?

[The Hood]

Just a thought for jamespetts here - with the balancing in pak128.Britain standard there are a few problems in the early years - low speeds and low capacities mean low throughputs, which mean low revenues.  You can set low vehicle and track maintenance for early vehicles and track, but (strangely) you can't set costs for stations / depots - these are all set in simuconf.tab and are time independent.  I guess the same is true in experimental (unless you changed it).  You will need to allow for this...

For those testing, can you be profitable in later years (1900s)?

[jamespetts]

Ahh, the station costs are directly proportionate to the capacity, though, so earlier stations can be (and, I think, are in Pak128.Britain) lower capacity, which means that they cost less, and is also commensurate with their lower throughput. Perhaps the station maintenance cost needs to be reduced generally in Pak128.Britain?

[wlindley]

Experimental 6.3 loads all bus stations -- but under 6.5, starting a game in 1850 or 1930, only the Garage, Bus Terminal and Post Office are available. 

In the SVN for the pak128Britain-Experimental I see in stations/bus_stops.dat that those are the only ones that have an introductory year... the Basic Bus Stop and such omit the intro_year ... unlike the sources for  the "non-experimental" pak where the same stations are given an intro_year of 1740.  Presumably the fix would be to restore the intro_year lines.

[jellyman]

An attempt with starting time 1930:

http://simutrans-germany.com/files/upload/Britain%20exp.sve

[jamespetts]

Jellyman,

thank you very much for uploading that - that is most helpful. May I ask why you used the Claude Hamilton locomotives? Those are more for medium-distance passenger trains than heavy freight. The GWR Prarie or the LNWR A-class would both have suited your bulk freight traffic flow better: your trains would have been faster (because the locomotives, although having a lower top speed, are more powerful) and you would therefore have earned more money in any given space of time.

On the issue of profitability more generally, I have tweaked the balance in the next version of Simutrans-Experimental so that more different types of costs, including station maintenance costs, signal maintenance costs and station extension maintenance costs, are scaled with the distance_per_tile setting. This should reduce the infrastructure costs in most cases and therefore make it easier to make a profit.

[sdog]

yesterday i generated a medium sized map and saved it as base for further testing: start date 1890, 50 cities, 2k5 median pop, 9k tiles roadlength, industry grid 50, ind. crossconnect 30, map northengland 1680x1365

first try was a quick and dirty rail connection between two 15k pop cities. at a distance of maybe 500 to 1000 tiles. made major losses. found one minor bug: trains wont cross a grade change with a road with tram tracks. (won't include save as a similar will follow)

today i tried transporting wool. half of the track the trains could transport textiles on the way back. but they hardly got their running costs back. tried to grow to reduce the impact of maintenance.

before tweaking it i had a crash. an autosave has been saved directly before it (less than 1 minute). when restoring the savegame time was set to 1 jan 1910, instead of 1893. see savegame wool.
(where can i find logfiles providing more details about crashes?)

next setup started with the try to make profit with city transport. i set up tram networks in both cities. made a small loss. (with tweaking i could have come to zero profit, i expect)
tried again to connect cities, now with small towns inbetween. having both a slow train connecting all stations and a fast express train increased the number of passengers available greatly. the fast train in particular seems to draw passengers.  the setup is making a huge loss by the way.

i found a nasty state, where trains are partially filled with passengers going to the station they departed from. in one case a train leaving blackpool east station was almost exclusively full with passengers traveling via blackpool east station. sending it back to that sation forced about half of the people out of it.  
you can find this also in the savegame, with the train departing from manchester.

not having cheap bus stops available is also a difficulty when connection less dense areas.

there are some irregularities with some vehicles maint. costs. the blackpool tram costs only 0.06, while the other available electric tram in 1890 costs about 0.8. tram horses cost 0.06, while horses for carriages cost 0.66


PS: no files since to large, 2 min of battery left, ill try to upload again later


edit: james, i sent you the files in pm.
i've been in a hurry, and quite annoyed, when writing above (twice). please excuse that it is rather confuse and rough.

[knightly]

On the issue of profitability more generally, I have tweaked the balance in the next version of Simutrans-Experimental so that more different types of costs, including station maintenance costs, signal maintenance costs and station extension maintenance costs, are scaled with the distance_per_tile setting. This should reduce the infrastructure costs in most cases and therefore make it easier to make a profit.

James, I have already explained to you very clearly that you need to scale all fixed maintenance costs if you don't want to adjust the graphics scale, in my previous PMs to you a month ago. Probably you thought that I was just talking about the fixed maintenance costs of convoys, which I specifically mentioned just to refute your suggestion to increase profitablility by adding more convoys -- however, such specific mentioning doesn't mean that the fixed maintenance costs I was referring to were limited to fixed maintenance costs of convoys. If you had pondered carefully over my suggestion, you should realise that all fixed maintenance costs need scaling, and such problem would not have happened.

[jellyman]

 

Jellyman,

thank you very much for uploading that - that is most helpful. May I ask why you used the Claude Hamilton locomotives? Those are more for medium-distance passenger trains than heavy freight. The GWR Prarie or the LNWR A-class would both have suited your bulk freight traffic flow better: your trains would have been faster (because the locomotives, although having a lower top speed, are more powerful) and you would therefore have earned more money in any given space of time.

I think I made a mistake when choosing the loco, as I was considering hp to maintenance cost ratios, which would mean the Prairie, and I guess I accidently selected the one next to it.

[jamespetts]

Thank you all very much for your feedback: it is extremely helpful. I am currently working on version 6.6 of Simutrans-Experimental to try to address some of these problems and see whether the calibration can be improved. It will scale the maintenance costs of stations and signals, amongst other things - when that is released on Saturday (or possibly Sunday this week), I should be very grateful if people could try it and see whether making a profit is easier.

Meanwhile, I have tried SDog's saved games. The passenger saved game suffers greatly from the bug in Pak128.Britain regarding introduction dates of 'bus stops, which I have now fixed on my local machine, and will be in the next release. Also, I have noticed that the waiting times are very high for some of the stations, causing fewer passengers to travel than they otherwise might. This might affect profitability. As to the wool setup, the wagons used to carry the wool were out of date, resulting in a 18% increase in the cost of their maintenance, which might, again, impact on profitability. I do not know whether these factors alone, if eliminated, would enable the networks to run profitably, but it is something worth considering.

[sdog]

i think the wool wagons (and locomotives) were out of date due to the 20 year jump, wich i noticed after loading the autosave after the crash. sadly i don't have any other information about that bug, except the savegame.

Edit:

before tweaking it i had a crash. an autosave has been saved directly before it (less than 1 minute). when restoring the savegame time was set to 1 jan 1910, instead of 1893. see savegame wool.

could you look into the problem where trains left a station with many passengers traveling to that very station? in the savegame you should immediately see it if you check the passengers in the fast passenger train.

[The Hood]

Do the stop intro dates need fixing in the sourceforge SVN?

[jellyman]

When you talk about scaling, does this mean that revenue and running costs change according to the size of the map?  And that if I use a map of a certain size it should be well balanced?  If so what size and I'll give it a go at that size.  Or I'll wait until you get a fix out.

[jamespetts]

Sdog,

thank you for reminding me about that - I shall look into that when I have some more time.

The Hood,

yes.

Jellyman,

they are scaled with the distance_per_tile setting (look at the map and select "show scale" to see what the setting is). They do not scale with the actual size of the map.

[knightly]

It seems that the output storage capacity for wool at Sheep Farms is improperly set to 0, causing the quantity stored to alternate between a large amount and -1. Please see the attached images.

[jamespetts]

That's rather odd - I have just checked the sources, and the output capacities are all significantly above zero.

[The Hood]

There was a similar bug in standard pak128.Britain release 1.00, which has since been fixed in 1.03 - is this the same one resurfacing?  Double-check the sources used to make your release (I always make mistakes where I fix it in one place but then build using old sources!).


I checked intro dates in pak128.Britain SVN, and they are all present and correct - which also makes me think you may have built this pakset with slightly outdated sources.   Just a thought though...

[jamespetts]

Sdog,

I have had a go at seeing if I can reproduce your problems: no luck so far, I am afraid. Wool is indeed in January 1910 when I start it, but, perhaps predictably, there is not much that I can do to reproduce the earlier crash. Was there anything in particular that you were doing when it happened? Also, I cannot find any instances of anomalies with passenger loading in the passenger game.

One other profit-related issue that you might want to consider in the passenger game: you should set a higher number of cities. Simutrans-Experimental with PAk128.Britian is calibrated for a far higher density of cities than Simutrans-Standard and the Simutrans-Standard paksets. You had 50 in quite a large map - you'd need at least 100, if not 150 or 200 to work as intended. The passenger level is set lower to balance out having more, larger cities. (Your median of 2.5k is fine).

The Hood,

I did try updating from the SVN recently, but it made no changes.