News:

Do you need help?
Simutrans Wiki Manual can help you to play and extend Simutrans. In 9 languages.

[Mega Project] pak128.Britain Balancing

Started by The Hood, January 25, 2016, 08:40:27 PM

Previous topic - Next topic

0 Members and 2 Guests are viewing this topic.

The Hood

As I've hinted in a few places elsewhere I think it's time to make a serious attempt at balancing the pakset properly. I'm under no illusions this will be a massive project but no doubt worth it in terms of playability in the end.


I thought it worth mentioning my vague strategy as it's likely I'll need help (advice and testing!). A lot of it I intend to follow on from my previous attempt in 2013 (discussion here: http://forum.simutrans.com/index.php?topic=11849.75


Basically I intend to fork development of pak128.Britain now and I'll be making a series of releases for play-testing. I'd appreciate people taking the time to try out the physics (acceleration and power) and economics of the new versions, either on existing games or on new games, and to comment to me for refinement. First step will be to rebalance the trains (as the most numerous and most complicated mode so hardest and most important to get right). I will start with the conclusions from my old work in 2013 but will add in ideas about station cost which have now been added to the code. This will mean a significant change in how physics and economics work in game, so old saves may not play to well - I wouldn't recommend switching mid-game for any reason other than to see how the new versions work!


Any thoughts or ideas are also very welcome (except where they involve a big headache and lots of work to solve! :p)


jamespetts

Goodness - I wish you the best for this marathon. I am still some way off (in coding terms) being able to do this for Experimental. It will make a huge difference to how worthwhile that it will be to play the Standard version, however.

I should note that I am part-way through producing quite a number of rail data (including many new locomotives, carriages and electric multiple units) in addition to re-scaling large numbers of things already in the pakset, the new vehicles of which will need balancing when they arrive.

I expect that the balance for Standard will end up being really quite different to that for Experimental in the end owing to the limitations in what can be simulated in Standard; it may be worthwhile in some cases adding .dat file comments (marked with the # symbol) showing what has been changed from a researched value, where applicable.
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.

Drewthegreat87

The Hood,


What all are you planning on using to balance out the trains? When 1.17 was released, I went though the trains and updated (Mostly LBSCR and a few Midland locos) to bring up their power/gear within ranges which could haul trains(LBSCR locos for example might not have had a gear listed, or it was low and couldn't sufficiently haul more than, say, one or two carriages). I didn't update them to prototypical specs, but changed them to fit in with contemporary units for their era. My playing bench mark is to utilize locomotives that can haul trains at their maximum speed (i.e. the green bar is green when I fill out the carriages/goods wagons etc). I've only recently gone past that so much as speed isn't impacted too much.

Road hauling vehicles in later game need some help.

As far as economic balancing, what are you specifically looking at? Pricing? Factory output/needs per era? I think some haulage forms are a bit overpowered (ships for example) I am willing to help play test when I have the time family/work/real life come first naturally, but I usually have a bit of time during the week to do whatever with.

The Hood

@jamespetts - I think experimental is doing a great job of cornering the "realism" market. You've noted in many places how standard doesn't accurately simulate the real world so for balancing the standard pakset I intend to adopt a "semi-realistic" model, i.e. reasonable historical top speeds, kW powers (for diesel and electric where known), but after that to imagine what a "sensible" historical load for each vehicle would be and fiddle the other numbers to get it about right in proportion with other vehicles. The idea is to get a sort of realistic "feel" to the whole game without breaking the limits of what Standard can achieve. Clearly that will mean quite different data from experimental, but standard is supposed to be more entry-level and there will always be experimental for those who wish to go for the extra realism.

@DrewTheGreat - thanks for the offer, it will be very helpful. Power and gear will be completely overhauled. The plan is to have gears more or less the same (actually electric, diesel, early steam, late steam (superheated) will all have different gear factors I think but within each group the same gear). I have an excel spreadsheet that should ensure a "sensible" load is possible with each loco. Many of the vehicles I made over the last few years have placeholder dat values or just copied straight from experimental, hence the need for rebalancing. Economics-wise it will be everything. First job is to ensure that full loads are profitable for sensible convoys and infrastructure, considering prices, maintenance costs and track costs. After that I'll look at the supply/demand side.

wlindley

Perhaps the various station add-on buildings could be sized and priced along a somewhat logarithmic scale.

For example, in 1850, taking a smaller base railway tile capacity [perhaps 16 or 8 versus 32 now] as 1 unit:

BuildingCapacityBuild CostMonthly Cost
Goods siding
Livestock Pen
Goods Shed
Goods Warehouse10×
Bulk Goods Loading Tower20×
Fluid Storage Tanks50×10×10×
Bulk Goods Bunker100×20×20×

and likewise for the various passenger stations.  In the 1850s-1860s, there should be some incentive to use brick platforms over wooden ones, perhaps much higher building costs and lower monthly:



BuildingCapacityBuild CostMonthly Cost
Wooden Platform
Brick Platform10×
Wooden Platform with Overroof
Country Station Building20×
Brick Platform with Building50×
Wooden Station Building10×20×
Brick Platform with Overroof10×50×
Stone Station Building20×100×

Note too that the fluid storage and country station are introduced later but fill in between some of the earlier larger tiles.

This would go a long way in upgrading station construction from mere eye-candy to require considered planning.


Iluvalar

Before going too far, may I suggest you give a try to my pak128-ilu (http://forum.simutrans.com/index.php?topic=14891.0). Where pak128, pak128.britain and pak64 coexist without timeline and all vehicles being balanced ? It might show you what you can achieve, and what it will cost in term of realism and gameplay. This is balanced a completely different way then other paks.

There is some high level game design problems I needed to tackle before doing the actual balancing. Mostly, once everything is balanced, what will make the player opt for a slower road ? How to make different speed roads really coexist intentionally by the player ? And not just enforce it by the last better vehicles given in the timeline.

Also a good one : What will make a 3t bulk truck and a 5t bulk truck really stand out for each other and make the choice engaging for the player ?

Another thing you need to pay attention before hand : The very important balance between vehicle earning, operational cost and maintenance cost. Their ratio  dictate the gameplay and you need to figure what you really want. I made a big mistake first on that, I'm happy I was using a script that could redo the balance from scratch I would have been mad if I was doing it manualy. Here's the  trap : You will probably try to make the game hard the wrong way around. I.E. you will try to make the vehicle earning low compared to the costs. Maybe 1:3:3 (profit:vehicle:road). Thinking this will make the player more prone to bankrupt and therefore the game more challenging. But nope, this will lead you to the exact opposite; the lack of room for creativity in the lines will cause the optimal solution to be in the great majority of time a straight road between 2 points. And you will have to make the balance so that's a viable (if not optimal) option for most of the lines. The players would draw straight lines everywhere which is just the opposite of difficult.

I'm very enthusiastic about balance. I'd be happy to answer any question you have :).

jamespetts

Here is some music to keep you going during the long, lonely hours of rebalancing purgatory:

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.

The Hood

Quote from: wlindley on January 26, 2016, 02:27:04 PM
Perhaps the various station add-on buildings could be sized and priced along a somewhat logarithmic scale.

For example, in 1850, taking a smaller base railway tile capacity [perhaps 16 or 8 versus 32 now] as 1 unit:

BuildingCapacityBuild CostMonthly Cost
Goods siding
Livestock Pen
Goods Shed
Goods Warehouse10×
Bulk Goods Loading Tower20×
Fluid Storage Tanks50×10×10×
Bulk Goods Bunker100×20×20×

and likewise for the various passenger stations.  In the 1850s-1860s, there should be some incentive to use brick platforms over wooden ones, perhaps much higher building costs and lower monthly:



BuildingCapacityBuild CostMonthly Cost
Wooden Platform
Brick Platform10×
Wooden Platform with Overroof
Country Station Building20×
Brick Platform with Building50×
Wooden Station Building10×20×
Brick Platform with Overroof10×50×
Stone Station Building20×100×

Note too that the fluid storage and country station are introduced later but fill in between some of the earlier larger tiles.

This would go a long way in upgrading station construction from mere eye-candy to require considered planning.



What do we think about this? Certainly earlier in the game the comparative cost of stations should be less (a because they were less permanent structures at first and labour costs would have been cheaper, but also from a gameplay perspective as revenues are lower with lower capacities). I was toying with the idea of reducing the capacities and costs of all the sidings and platforms (therefore requiring more extension) - the flip side is it becomes cheaper to build sprawling stations that have huge coverage. Any thoughts?

Vladki

I would just suggest putting higher maintenance on wooden platforms than brick ones.
Capacity should be the same, dependent only on having overroof or not.

Why a bulk goods bunker has so high build and maintenance costs? I thought it is just an open place to pile up coal (etc.), perhaps with a wall to separate it from other bulk goods.

Anyway, I'd gladly do some testing. Though in my previous game with pak128.Britain, I felt that the most unbalanced things are airplanes, busses and mail trucks. Those could not make profit even if running full both ways.

The Hood

Problem with that is that wooden platforms are earlier than brick. More maintenance per month hits profitability really hard with old slow convoys with smaller capacities. The incentive for brick wood be higher capacity instead.

Spenk009

Wooden platforms should be cheaper to construct, but if a tile is in use for more than 3-5 yrs it'll be much cheaper to have invested in stone. Station buildings should be the way to go in terms of adding capacity.

May I suggest you fork a portable, compiled and readily-playable version of Simutrans-Ex? Every incrementation would get a new branch and there'll be fewer version discrepancies between savegames. People download the most recent branch and can move savegames as they wish.

jamespetts

Quote from: Spenk009 on February 01, 2016, 04:16:22 PM
Wooden platforms should be cheaper to construct, but if a tile is in use for more than 3-5 yrs it'll be much cheaper to have invested in stone. Station buildings should be the way to go in terms of adding capacity.

That is the way that I have done it in Experimental.
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.

The Hood

Latest thoughts - it seems most very early railway stations didn't even have platforms so maybe we just use the existing goods station graphics at that time and have the passenger facility and capacity via an extension. The siding may even not be goods enabled (requiring extra facilities for storage/goods enablement). Wooden platforms and brick platforms would come in later with higher capacities and maintenance. Stations costs cease to be a significant burden once trains can go at 50km/h or haul longer loads e.g. 1840 onwards.

Secondly I've taken another look at the power and gearing situation. I reckon we can do a better job of balancing by significantly reducing overall gear*power, especially for steam engines. They would only hit their theoretical maximum speeds on long straights/downhills with moderate loads and have quite low acceleration compared to now. The question I need to answer is this: given top speeds didn't increase much with the introduction of diesel power (and on most routes aren't much better even today) does anyone have any knowledge of comparative accelerations of steam vs diesel vs electric from the 1950s to now? Obviously you can get much faster overall and much better capacity if you have better acceleration...

Spenk009

I like the platform idea, it adds realism and additional use for the sidings. You need to ensure that it isn't more effective to build this way throughout the game (althought transfer times may cancel that out). Also, is this intuitive for new players?

Maybe it's worth having a look at train sims, comparing how different convoys do on similar sections. These usually feature a selection of vehicles from many ages. But that raises the question of whether such a sim used is accurate for our purposes and (especially with steam) how hard the engines are driven.

Junna

As I recall Petts has done quite a lot of detailed work into that as far as I recall. Top speeds will be limited by other factors as well though; modern speed limits are far more restrictive than they were (because monitoring and so on is better), things that in the past were more fluid and there was a somewhat large tolerance for speeding which would be unacceptable today. A somewhat large percentage of travel time is spent on long and slow-speed approaches to stations and associated pointwork.

It seems to me that in general the "power at drawbar" for steam engines is always ridiculously overestimated (given those old American steam locomotives that were rated at absurd levels near 4000KW, only to be replaced by sub-3000KW diesel-sets with greater loads...) Like the A4 being given at 2,200KW yet in practice it seems to have been nowhere near this.

Spenk009

I have no idea how the drawbar measurement is set up, whether all systems are at maximum or usual operating level. Testing a loco at full steam and power is different to normal operating conditions. Electric and ICE are run at their "best" performance which is always available and used under normal operating conditions.

jamespetts

I should note that the physics in Simutrans-Experimental is different and, signifncatly, includes tractive effort whereas Standard does not, so one cannot just use the same values.
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.

The Hood

Quote from: jamespetts on February 03, 2016, 09:30:36 PM
I should note that the physics in Simutrans-Experimental is different and, signifncatly, includes tractive effort whereas Standard does not, so one cannot just use the same values.
Quite. From your research into this though did you get any idea about the variation in accelerations from older to more modern traction systems?

jamespetts

Quote from: The Hood on February 03, 2016, 10:09:00 PM
Quite. From your research into this though did you get any idea about the variation in accelerations from older to more modern traction systems?

Because acceleration is determined by tractive effort, and those figures are generally available historically (although hard to find for multiple units for some reason), I have not researched this specifically, but you can largely do the same thing by researching tractive efforts and extrapolating acceleration from that.
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.

The Hood

OK. Presumably a simple F=ma for the initial acceleration as tractive efforts quoted are usually maximum values which occur at lowest speeds. How have you dealt with TE curves though? I'm more into a "squares to max speed" sort of idea - overpowering the vehicles gives quicker rise to max speed.

jamespetts

I am not sure  how Bernd, who wrote the physics engine, dealt with curves: the idea is that one just specifies the starting tractive effort in the vehicles' .dat files. I know that steam locomotives are accounted for differently in the code on account of being constant force machines, whereas other types of traction are constant power machines, but I am afraid the details are a little beyond me. It does seem to work, more or less, when comparing the performance of the vehicles to real life performances recorded, however. In Experimental, altering the power has a very limited effect on acceleration.
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.

The Hood

I've done a bit more reading on this. There are differences in drawbar force at different speeds but these are quite complex (far too complex for standard balancing levers to model well). It appears diesels are better at getting heavy loads going but tend not to reach such high speeds as steam theoretically can (we can model this with higher power and lower max speed) and that they are better on gradients (don't think we can do this at all) but the main advantage is financial - they are simply much cheaper to maintain. In the overlap period it will probably be a choice of better performance (steam) v better profit (diesel).

Vladki

Quote from: The Hood on February 02, 2016, 07:08:44 PM
Latest thoughts - it seems most very early railway stations didn't even have platforms so maybe we just use the existing goods station graphics at that time and have the passenger facility and capacity via an extension. The siding may even not be goods enabled (requiring extra facilities for storage/goods enablement). Wooden platforms and brick platforms would come in later with higher capacities and maintenance. Stations costs cease to be a significant burden once trains can go at 50km/h or haul longer loads e.g. 1840 onwards.

Have a look at this type of platform:
It still is used on Czech railways, on small stations - low capacity, low construction and maintenance costs. I think Ves has already made something similar for pak128.Sweden. Here is even a blueprint of such platform (at the bottom): http://www.parostroj.net/modely/stanice/nastupiste.htm

jamespetts

I think that, with perhaps some very early exceptions, nearly all British railway stations, even the tiniest halts, had full height platforms: it is one of the distinguishing feature of British railways compared to those in many other parts of the world, where passengers often have to board at near track level.
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

Good luck and more power to you.

Couple of thuoghts:

1) Mighn't it be easier to start with the simpler stuff, like busses?  I played a game recently in the 1910s and the busses were tricky and then I think it is the little green Leyland Cub comes along at like 1/10 the operating costs and then suddenly we are rolling in cash.

2) Thank you thank you thank you thank you.

3) Personally, I am not enthusiastic about Experimental platforms have 0 passenger capacity. I have read the argument that British passengers are not allowed to wear hats, and so can not stand outdoors in the rain, but as someone who lives in a developing country (USA) I almost only ever know crowded open air un-elevated train platforms and little other infrastructure. Fortunately it is usually sunny in California ... but please please please let at least a few folks stand on platforms at say crowded urban stations where bulldozing extra tiles for the passengers to wait gets expensive.  (I always wait on tube platforms when I visit London Underground. ;)

The Hood

Thanks for the ideas.

1) Actually I'd rather not fix buses until I've got something that works for the trains - far more combinations of consists with locos + wagons/coaches etc. to think about. It'd only lead to wasted time once I got onto the harder stuff and found it didn't work after all. The buses should be easy to do once I have a system that works for rail.

2) Thanks

3) Currently I'm thinking goods stations need extra storage facilities. So basic sidings will do capacity=10 each with some of the towers and cranes going up to 50 but the extensions will each have capacities between 100 and 500. Similarly platforms will have small capacities of 20-100 but the extensions will be required for large interchanges and termini with capacities up to 500 per tile. That allows unprofitable early trains to get by with little extra infrastructure (which costs maintenance) but as train capacities and profits increase, more maintenance is required for extra station capacity. Just haven't got the cost numbers right yet...

jamespetts

I should note that the reasons that, in Experimental, I gave no capacity to the basic platform type were: (1) it is important to decouple the cost of accepting longer trains with the cost of increasing storage capacity; and (2) stations consisting entirely of bare platforms with no buildings were previously commonly being built before this was implemented, and this type of station is entirely unrealistic: in the UK, more or less all stations have some sort of facilities. It seems to me that both of these matters apply equally to Standard.
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.

The Hood

Believe it or not I'm actually making reasonable progress with the balancing. There's just a huge amount of it to do!

I'm beginning to wonder if the selection of trains in Standard could do with a simplification, especially on the rolling stock front. This is for two reasons - one, depot clutter, and two, the fact that many vehicles, especially carriages, have no purpose in Standard (e.g. kitchen/dining/tpo). Jamespetts has created a huge variety of different images for experimental but I think carriages can be simplified down into express (low capacity high speed), normal, and suburban (high capacity low speed). I'm thinking to ditch the dining cars and TPOs and just have brakes and ordinary cars for passenger and mail. Also where there are several different lengths of car available at the same time in a similar style (e.g. a 8 wheel and 12 wheel version) just to keep one set. Anyone got any strong views? Do people want all the dining cars etc in Standard for eyecandy value or better to keep it simpler?

jamespetts

Perhaps those things that are not capable of having an economic function in Standard could be moved to an official addon? Splendid that you are making good progress with balancing, though.
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.

kierongreen

Getting rid of vehicles that serve no purpose seems reasonable.

Another question would be how many companies to support - when I set out I had envisaged at any point in time having inner suburban, outer suburban and express (which I guess would work out the same as suburban, normal and express) for each of the "Big 4" (or one of their constituent companies, or the successor BR regions or later one of the privatised companies that then took over each region).

Even that would mean up to 12 (or 24 once electric appears) passenger trains available at any one time which I think is more than enough. Up to not necessarily all of course, no use in hunting for particularly obscure examples just to include. In the BR era the multitude of DMU and EMUs I think should stay but strictly timelined so that they don't crowd the screen.

The Hood

What to do about companies is another tricky one. We currently have complete lines for LMS (and LNWR, Midland, LTSR) and Southern (and LBSCR, some LSWR). LNER (and GNR) is good for passenger locos but not freight and GWR is OK from grouping onwards but not before.

As part of the balancing process I have created a "timeline" of different types of vehicles for each company so I can manage the intro/retire years quite ruthlessly to avoid clutter.

That should give a manageable number but obviously there will be some duplication as different companies will still produce similar locos.

But it may still be too many for newbies. I have toyed with the idea of producing a "cut-down" set of just the most famous examples, with complete company timelines as an add-on, but then most people seem to quite like having lots of trains to play with anyway (I'm always surprised how much people demand any new locos to be released even when they duplicate existing ones in purpose)

kierongreen

Early GWR has the problem of gauge obviously... Yes, part of pak128.Britain's appeal is the number of vehicles, I think it's a case of if people choose to play it with timeline off on their head be it!

Great to see this anyway - once it's done I might even have to try a proper game :)

jamespetts

I do plan to get around to filling in some of these locomotive gaps and more* at some point (and am working on the LSWR now; just the Drummond era locomotives for teh time being), but coding will have to take priority for quite some time once I have finished the current project.

We must not forget that a lot of people like this pakset so that they can recreate things in a sandbox mode, for which full vehicle availability is a worthwhile thing.

* We could also do with more things from the Highland Railway, Great Eastern Railway, North London Railway, North Eastern Railway, Lynton and Barnstaple Railway, Lancashire & Yorkshire Railway, Mersey Railway and probably quite a few others.
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

You must remember that the trains various real life companies used in the past where engineered based on requirements and necessity. Although several of the companies never had certain types of train the main reason for that was that they never had need for such a train due to how they operated. If they had the need then they would have had the trains. In simutrans you have to provide the player with a choice of engines and they need to select one based on their requirements.

From a game play perspective it does not make any sense to provide 2 different engines that perform the same role because they were created and developed by two different historic companies in parallel to fulfill the same role. Equally well it would make sense that fake/imaginary engines are created for certain time periods for certain roles that were never needed in real life but might arise in Simutrans and were possible with technology at the time, or were performed by engines in other countries.

Vladki

I agree with putting dining cars, tpo and liveries as addons. I would keep similar but not identical vehicles from historical companies in the main pak.