The International Simutrans Forum

 

Author Topic: Improve mail-behavior in 11.xx  (Read 4938 times)

0 Members and 1 Guest are viewing this topic.

Offline AvG

  • *
  • Posts: 308
Improve mail-behavior in 11.xx
« on: April 08, 2014, 12:12:17 PM »
James,
The major release takes a long time, wich is very understandable.


The latest EXP11.23 is however unuseable if you are balancing vehicles. Pax is not perfect, but useable; mail in the bundle-sytem, due to the low numbers, gives heavy losses.


Is it realy not possible to add somewhere a multiplier of 25 (bags>bundle) and make another minor-release?


As long as we have a useable 11.xx we can wait for 12.xx and you can take the necessary time without being pressed.


AvG

Offline Sarlock

  • Devotee
  • *
  • Posts: 1340
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #1 on: April 08, 2014, 02:29:48 PM »
I am finding mail to be perfectly profitable in the online game.  As speed is less important for mail, low running costs become the primary driver of choosing an appropriate vehicle.

An easy tweak is to just raise the revenue for mail a slight amount by editing goods.dat and compiling a revised pakfile for your own use.

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #2 on: April 08, 2014, 06:56:27 PM »
Sarlock,
I am balancing. That means in this case that if I raise the transport-fee a lot that it only looks to be balanced. I can do that in 2 minutes, but will end with wrong information.


No, my question about a temp-fix for mail still stands.


AvG

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #3 on: April 08, 2014, 10:06:14 PM »
What do you mean by "wrong information" here? What exactly is the issue with mail; does its generation not conform to realistic patterns? If the only issue is profitability, the price per unit can be increased; if the problem is the quantity being too low, both the price and the weight can be increased to represent a single unit standing for a larger number of postal packets.

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #4 on: April 09, 2014, 08:37:13 AM »
I know. All this "solutions" crossed my mind.


But I still have no answer on my question.
AvG

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #5 on: April 09, 2014, 09:53:52 AM »
If the question is,

Quote
Is it realy not possible to add somewhere a multiplier of 25 (bags>bundle) and make another minor-release?

then the answer must be no, since adding any parameter (and I am assuming that you are referring to a parameter that can be set in simuconf.tab rather than multiplying mail production in all cases by 25 times, which would unbalance existing paksets) requires stepping the saved game version, which cannot be done on a minor release. But I am still not clear on what you mean by "wrong information"; can you elaborate?

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #6 on: April 09, 2014, 03:45:52 PM »
By wrong information I mean: I have to make conclusions based on information, of wich I know it is not right.
When you introduced the mail-bundle you said it would be possible to make a nice profit on mail-transport.
I recompiled all my DRC-vehicles based on the mail-bundle.
The average numbers for mail at stops should now be a lot higher. (a bag of 25 kg >> bundle of 1 kg)
Somewhere in the mail-generation is an error hidden, wich is leading to constant losses in mail-transport.
Most of my mail-capacity vehicles drive around with zero load or max 10%.
Now I wonder what would be the result, if you, in the existing function, add a multiplier of 25.
I understand that it will lead to a situation of zero mail when zero is generated; but IF mail is generated the outcome is multiplied [size=78%]by 25 and you have something to transport.[/size]
This result is not perfect but better than we have now.


At the moment my DRC-scenario (high prices/low maintenance) is now ~20 years running.
I have a monthly profit of ~400K after investing now 50 millions in total. So ~10%.
But now I have to make conclusions.
My early horse/steam vehicles all have some mail-capacity. It is however quite impossible to judge what a proper mailfunction
would do.
I am using a pax-fare of 0,04HCr. My latest info on pax-fare is in 1941 0,03 HFl, wich is < 0,02 HCr.
I planned my next scenario on a pax-fare of 0,03 HCr. Might be critical.
I hope with this info the situation is clear now.




Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #7 on: April 09, 2014, 06:42:27 PM »
May I ask - by how much have you recorded that the mail generation in 11.x falls below a realistic level of mail generation based on historical statistics?

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #8 on: April 10, 2014, 03:00:40 PM »
I have only one piece of info in the real world.


2006 Post.NL handles daily 15 millions pieces of mail, of wich 400.000 are parcels. In that year Holland has ~15 million inh.


So for starters we can assume 1 piece of main per inhabitant.


I tried to compare that during 5 gamedays.


I watched in the biggest city a tram/busstop to see howmuch mail was generated in these days.
The stop is located at the cityhall wich is surrounded by high offices and tenements.
IMHO certainly not an average area, but a place where lots of mail would be generated.


In those 5 game-days I saw mail arrive (vanish instantly), I saw mail arrive and depart (through post), but no mail generated in the stop-environment.
Coincedence??, or is mail only generated by low-level buildings??


Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #9 on: April 10, 2014, 08:03:11 PM »
When you refer to "5 gamedays", are you using the hours/minutes scale or the months/years scale? The mail is intended to be calibrated by the hours/minutes scale. The way to check the proper calibration of mail generation is to use the "mail" graph in the city window and compare it to the population. The individual buildings' mail levels can be shown by clicking on them. Note that even a bundle of mail is taken to be 10 individual postal packets; note also that mail can "walk" to its destination (i.e., be delivered by hand) if this is quicker than using the player service, which might affect how many items show up at stops. This hand delivered mail is not recorded on the "sent" graph, although this anomaly will be fixed in the next major release.

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #10 on: April 12, 2014, 09:33:45 AM »
OK, I think I understand.
The timer I use is the one at the left bottom.


When generating mail are all building treated the same way, or are there differences between res, ind, com, etc ??


Should a building with a mail-level 50 produce 5 times more than a level 10 one??
AvG


Edit: A townhall seems to have no pax- and mail-level. That right or is it only not shown?
« Last Edit: April 12, 2014, 10:35:58 AM by AvG »

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #11 on: April 12, 2014, 10:06:47 AM »
The timer at the bottom left will be different depending on simuconf.tab settings, and only in one particular mode (the default in Pak128.Britain-Ex) will show the hours/minutes/seconds of simulation time.

In Standard and the 11.x versions of Experimental and earlier, commercial and industrial buildings generate twice as much mail as residential buildings, but, otherwise, the mail is dictated by the mail level in the way that you suggest.

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #12 on: April 12, 2014, 10:44:09 AM »
OK.
So I click on an office-building and I see a mail-level of 50.


Is the 50 inluding the the multiplier of 2 or is it doubled after picking up the mail-level of 50


BTW: Allow_routing_on_foot is UNchecked in my scenario.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #13 on: April 12, 2014, 10:54:58 AM »
The "allow_routing_on_foot" only affects whether passengers can walk between stops: it does not affect whether mail can be hand delivered. If I recall correctly, in 11.x, which is in this respect the same as Standard, the mail level that you see already takes into account the multiplication for industrial and commercial buildings.

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #14 on: April 18, 2014, 04:38:19 PM »

As you can see in the attachment: I am not realy happy with the mail-simulation as is.
It is not reflecting reality and is handled like all other goods.


What is the reality:
When your company has the mail consession in a country:
- You handle huge numbers of mail. At present ~ 1 mail-piece/day/inhabitant (2013).
- You have a fixed fee per mail-piece (inside the country borders)
- You deliver normal mail within 24 hours.


I wonder if it is possible to reflect reality in Simutrans in a better way.
I am thinking of:
- Mail-generation once EVERY day. f.i. at mid-day. Simutrans "knows" mid-day at all BmP-   


settings. Average amount 1 mail-piece per day/inh. Can be adjusted via mail-level per   


building.
  Office-towers and the like should produce a lot more than they do as is.
  1 mail/day/person is a valid value in 2013. I think that 200 years ago there was a lot less
  mail per person. This value can be made year-dependend.
- A fixed fee could mean:
  Pick up mail as is with only the destination filed. On arrival the fee is received.
  Allthough the fee is fixed it has to be year-dependend.
- Capacity of vehicles/buildings is given in mp (mailpiece).
  The average weight of the mail can be set on f.i. 20 gram. (Combo of postcards-magazins)
  The amount of packages is neglectible.


- 24-hours-limit is probably to difficult. Any limit here will also mean necessity of time-
  storage.


  I can't judge the influence on CPU-time.
  Maybe that the increase in numbers is compensated by the absence of computing distance.


BTW: I saw in your repositry a new variable appearing n.l. to distance(x)
   I did not see this before so I think it is not in effect.
   It is also not in line with reality.
   The size of a Simutrans-map can look high in tiles, but in km it is still quite small.
   So small that a fixed mail-fee can be used for all mail.


AvG

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #15 on: April 18, 2014, 05:21:12 PM »
There are a number of separate issues here. The first is the total amount of mail generation. Remember that one bundle is intended to represent a considerable number of individual items of mail (the previous measure was a bag, which was intended to represent ten times as much mail). Remember also that a "month" is actually 6:24h (or 6.4h, or 1/3.75th of a day) in terms of the number of mail/goods/passengers generated, assuming that you used the default settings. A "year" is thus equivalent to 3.2 days, and in 11/12ths of that "year", the town with a population of 17,500 people generated 9,000 bundles, the equivalent of 3,068 per day (as measured in the shorter time scale used for generation of mail/passengers/goods). This seems to be entirely correct.

The second issue is the distribution of the generation. You remark in the comments in the spreadsheet that large commercial buildings seem to be generating an inadequate amount of mail compared to other buildings (if I understand you correctly). Any change to the relative level of mail generation between different buildings goes beyond bug fixing (and I am currently only fixing significant bugs and performance issues in 11.x) and impacts on the structural design of the game as a whole. However, the new passenger generation code for the next major version will be much more customisable, allowing the mail capacity of a building to be set separately to its passenger capacities. The passenger generation code also affects mail, and should make the relative quantities generated in different places make more sense.

The third issue is the pricing structure for mail. Again, any feature change would have to await a new major version. However, one must be clear as to conceptually what is actually being simulated. Is Simutrans actually simulating the whole postal system (which would include the sorting of mail being fully simulated) or just the transport subcontractors of the postal system (the Post Office in the UK, for example, would pay railway companies to carry its mail for it)? If the former, then one would need to simulate the various different types of mail (letters, parcels, postcards, etc.) and the differing pricing structures for each, and would need to have a totally different way of pricing mail before and after 1840, when the postage stamp was invented. If we assume the latter, and that some other entity, not directly simulated but merely assumed/imagined acted as the Post Office, then the current pricing structure makes sense, since a subcontractor, such as a railway company, would charge the Post Office more the further that it carried the mail.

As to what you describe as a new variable relating to distance, I am not sure to what you are referring - do you mean the range here? This is a new feature for the next major version relating to the maximum distance that vehicles may travel between stops and has no direct relationship with mail/passenger/goods generation or pricing. If you mean something else, I am not sure what you mean, I am afraid; can you provide a link to the relevant commit so that I can comment further?

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #16 on: April 19, 2014, 10:01:03 AM »
James,
My bad-1: I forgot to mention my BpM-setting. It is 23
I am setting up now a next scenario (2100*2400) and will run it at BpM=24.
Why a higher BpM-setting: IMHO Simutrans is more and more giving you the feeling of being busy in reality. The higher that
BpM-setting the better that feeling. A side-effect is however that you notice the eventual shortcomings better.
Your answer w.r.t. total amount is based on BpM22. In BpM23 I should have the double amount.


My bad-2: My intention was NOT to have my reflections implemented in any minor-release. Just possible ideas for the future.
I realise I was out of the thread-title.


I was not referring to your new parameter range. I mean the variable to_distance in your goods-data of mail.
I still think that this new variable will be bad in the mail-environment, but realy perfect for pax.
I checked reality and saw that fee/km can be four times higher for short (bus) rides compared to long (+100km) train yourneys.


Your answer about the pricing structure gives the impression that I propose a very complicated situation.
However, all I want is large realistic numbers based on small weights, forget parcels and if possible daily mail-generation.
That way you have to consider more often to use a mail-boy on a bike or with a small hand-cart than the use of an universal
mail-van. Delivering a postcard somewhere 5km away with a mail-van will lead to a loss if it happens more often.
It can even lead to a necessity of erecting a postoffice somewhere. From that office you run post-bikers for the neighbourhood
and transport post-office-station via a van.
But again, I do not know what the effect would be on CPU-time.


At the moment I am working on new formulas for running costs of vehicles. When ready I will check again what they do with respect to transport of mail.
AvG

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #17 on: April 19, 2014, 11:26:40 AM »
To avoid further confusion - how many hours are in your game months with your current settings?

As to the "to_distance" variable, this is not new, but was introduced some time ago, I think before even 11.0, although I forget precisely when now. The use of this is optional: it is perfectly possible to have a uniform price per unit of distance simply by not using this parameter.

As to pricing structures, I realise that you were not specifically proposing a complicated solution yourself, but what you did propose - a fixed price for mail transport irrespective of distance - does raise the question of what precisely is being simulated, whether the Post Office itself or merely its subcontractors, that gives rise to the somewhat complicated issues to which I referred. It seems to me that it would be inconsistent to have a fixed price mail transport, inferring that it is indeed the Post Office being simulated, without also simulating other Post Office level mail activity, such as deliveries to individual addresses rather than to nearby post boxes and mail sorting, and any pricing system would have to take account of the fact that not all mail packets are small letters. If one is not simulating the Post Office itself, one can omit these things, but then fixed pricing no longer makes sense.

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #18 on: April 19, 2014, 02:38:59 PM »
I have 12.48 hours per month.
AvG

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #19 on: April 19, 2014, 10:16:03 PM »
According to the adjusted figures, with a month of 12.48h (or 12:29h or 0.52 of day), your town of 17,500 people generated 9,000 bundles in 11 game months (being 5.72 days), which is equivalent to 1,573 bundles per day, which, even if a bundle represented as few as 10 individual items of mail (if I recall correctly, I think that they were intended to represent about 50), would yield 15,734 items per day or 0.899 items per day per person, so this still does not seem significantly inaccurate.

Offline greenling

  • Lounger
  • *
  • Posts: 1728
  • Simutransarchology it my hobby!
  • Languages: DE,EN
Re: Improve mail-behavior in 11.xx
« Reply #20 on: April 20, 2014, 09:37:17 PM »
Hello Jamespetts
I think that the Mail in the 11.24 in the Year from 1750 until 1825 a little bit to high it.
The Traffic from the Mail beginning to grow up then it gives the Steamrailway.

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #21 on: April 20, 2014, 09:58:55 PM »
Can you elaborate? What pakset are you using, and how are you calibrating what the mail level ought to be?

Offline AvG

  • *
  • Posts: 308
Re: Improve mail-behavior in 11.xx
« Reply #22 on: April 29, 2014, 05:32:29 PM »
James,
In your entry 18 of this thread you mention the possibility of using or not using the to_distance
parameter?
Where or how do I make that choice? Settings?? Simuconf.tab?? passenger.dat??
When it is set can I see that somewhere?


Another simple question. You use for Simutrans money a special symbol looking like a C with a
vertical line through it.
How can I make this symbol? Is it a key-combination?
AvG

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18763
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Improve mail-behavior in 11.xx
« Reply #23 on: April 29, 2014, 10:19:49 PM »
You can avoid using the to_distance parameter entirely simply by using the old format in the .dat file defining the goods types: just use value=[a number] instead of value[N]=[a number], and the fare stages system will not be used.

The symbol to which you refer is the American cent symbol, which is used in Simutrans to represent the currency there of "Simucents". It is a special character from the extended ASCII set, and how you enter it depends on the software that you are using - there should be resources available on the web that should tell you how to enter this character in your software.