News:

Simutrans Sites
Know our official sites. Find tools and resources for Simutrans.

Implementing the industry boost (etc.)

Started by jamespetts, May 06, 2013, 08:56:13 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

jamespetts

I am about to start to implement the industry boost (etc.) features in this pakset ahead of the next release. As with the general approach to the Experimental version of this pakset, I am trying to ascertain the basic parameters from real historical figures. Some things are more amenable to this than others: the number of people working at a factory, for example, is in principle easy to ascertain (more on that below). However, the extent to which a good postal service makes a difference to an industry's overall productivity (by allowing more efficient communications between managers, owners, accountants, lawyers, bankers, etc., as well, perhaps, as the supply of specialised parts or drawings, etc. in later industries) is very hard to measure, and will in effect have to be guessed. I intend to document that process here partly to have a record of what I have done so far, and partly to invite discussion of the parameters of the exercise to enable it better to be refined if anyone has any better ideas.

Firstly, the number of workers at a given industry: from what I understand the "passenger_demand=" parameter specifies the number of passengers demanded by the factory per game month. A game month in Pak128.Britain-Ex is 6:24h (or 6.4 hours, or 1/3.75th of a day). Since we assume that each worker will make one trip to and from the factory per day, to get the passenger demand figure, we need to divide the number of workers that any factory is likely to have by 3.75. Mail needs to operate on a similar principle: 1/3.75ths of the number of bundles of mail that a facility is likely to receive in a day.

As an aside, some reworking of the code relating to the distribution of passengers will be necessary in due course to reflect what is discussed here. In particular, consideration will have to be given to how the passenger demand per factory interacts with the planned distribution of passenger journeys by destination type (work and other). I need to have some idea of what is going to be implemented in order to know how to calibrate this now, so as to avoid having to re-do it. Provisionally, I am of the view that the passenger demand figure should be calibrated on the basis that it will be used - for all types of industry - to represent people who work there, and make work journeys to that factory. One of the effects of the passenger demand is to limit the number of passengers whose destinations are at the factory: in Standard, passengers bound for that factory which are beyond that factory's demand are distributed to other destinations in the city. I think that this might well not be implemented in the current Experimental, but in future versions it should trigger a failed route attempt, with the packet of passengers in question having to find another journey to work. Something similar will have to be implemented for commercial and industrial city buildings.

Two complicating factors arise here. Firstly, we have the "workers live in" or "customers live in" parameter

As to the production boost in so far as all the passengers (workers) are supplied, it seems sensible to assume minimal production without any workers, and the originally implemented production levels with a full complement of workers. This is inconsistent with the system planned, where passengers/mail will find their destinations using a weighted random system based on up to a fixed number of retries, and might simply have to be abolished entirely. Secondly, and more problematically, whilst it is easy enough having the production of, say, a mine or a factory depend on the number of workers, the "production" of a shop is actually the consumption - a figure which more naturally depends on the number of customers; but the shop still needs workers to sell things. Any ideas on how to square that particular circle would be welcome. (We could just leave it based on workers, and assume that customers will get there somehow, which is the best of which I can think at present, but any more sophisticated ideas that do not overcomplicate things, cause anomalies when compared with commercial city buildings, or otherwise cause more problems than they solve would be welcome).

Electricity is more difficult in that figures on the power consumption of industries by number of megawatts is hard to come by - if anyone could supply any information (including historical information as far back as the beginning of the 20th century) or point me in a direction where I might find some, I should be exceedingly grateful.

Likewise the extent to which electricity increases production: this is something which, I imagine, is even harder a thing on which to find historical data, and guesswork is needed. There is some progress that can be made by common sense: a vegetable farm is going to be less assisted by electricity than a printworks, but the nuances are not easy. Any thoughts would be welcome.

Finally, the periodic production "boosts" for being in an active state is not something that precisely corresponds to reality, so is difficult in principle to calibrate. Any ideas about how best to manage this would be most welcome.




Edit 1: There seems to be some considerable variability in coal mine employment rates: anything from 800 or so in larger later mines (see, for example, here on the Maltby mine) to fewer than 150 in the smaller Welsh mines (see, for example, here on the inaptly named "Copper pit"). 200-300 for the early 20th century seems typical for Welsh mines (as documented on this site), but larger collieries such as the Seven Sisters pit employed over 500.

Edit 2: Incidentally, the adjusted figures are: 800 / 3.75 = 213; 500 / 3.75 = 133; 300 / 3.75 = 80; 200 / 3.75 = 53; and 150 / 3.75 = 40.

Edit 3: Turning to heavy industry, the Round Oak Steelworks, a post-war facility, is reported to have employed 3,000 or so at its height before the decline in the late 1970s. In 1842, the Blaenavon Ironworks is reported to have employed 2,000 people and produced 400 tons of iron per week. Adjusting these figures gets us a passenger figure of 2000 / 3.75 = 533; for the production, we have to calculate this on the basis of working hours: there are 6.4 hours per Simutrans month, which is a reasonable figure for the fully productive working hours of a factory (ignoring setting up and shutting down time), and five working days per week, so the figure of 400 tons needs to be divided by 5 to get us the per Simutrans month figure of 80 tons (or 81.28 tonnes).

Edit 4: Still on the theme of heavy industry, the Milwall Ironworks seems to have employed 4,000 - 5,000 people in the 1860s, translating to 1,066 - 1,333 for Simutrans purposes.

Edit 5: Brickworks seem to have been a far less labour intensive business: this brickworks was employing 44 (adjusted: 12) people in 1881, whereas this brickworks (note: link currently broken but available using Google's cache; also try the Wayback Machine) employed 184 people (adjusted: 49) in 1964 and produced 80,000 bricks per day. The Hemsworth Brickworks employed 22 people in around 1903, and produced 84,000 bricks a week. Those production figures pan out as follows: bricks in Pak128.Britain-Ex are measured in 820kg crates. A single brick weighs 2-3kg (source). Assuming the weight of a single brick to be 2.5kg, 328 bricks can fit into a single crate. The 84,000 bricks per week figure would give us 256 crates per week, equating to 51 crates per working day (or Simutrans "month") by dividing by 5. The 80,000 bricks per day would give 243 crates per Simutrans "month".

Incidentally, that page also has information about the Hemsworth collier, which, in 1952, employed some 1,250 people (adjusted: 333) and produced 400,000 tons of coal per year. That equates to 7,692 tons per week, or 1,538 tons per working day (or Simutrans "month").

That page further refers to a textile mill in Hemsworth, which opened in 1947 and employed 120 people (adjusted: 32), turning out 70 "pieces of cloth" per day. Textiles in Pak128.Britain-Ex are measured in 430kg "crates", and it is not immediately apparent how the one would map to the other, although one suspects that the quantities are fairly low. It seems on the face of it from the description of that factory on the website that this was an ancillary factory to a more important facility in Bradford, so the figures may be low.

Edit 6: Canneries are not easy things on which to find information, but this American cannery opened in 1918 employed about 150-200 people (adjusted: 40 - 53).

Edit 7: Car plants seem to be very heavy employers. The Longbridge plant was the largest in the world in the late 1960s, employing 250,000 or so workers (adjusted figure: 66,666). Its fortunes have waned in recent years, but it still employs 400 (adjusted figure: 107). About 6,000 (adjusted figure: 1,600) were still employed in 2005 when Rover went into liquidation, and 22,000 (adjusted figure: 5,866) were reported as having been employed in 1917. In 1908, there had been only 1,000 workers (adjusted figure: 267).

A modern cement works (the Hope Cement Works in the Peak District) is reported here as employing about 300 people (adjusted figure: 80). The Chinnor Cement Works seems to have been a smaller concern, opening in 1919 and employing at its unspecified peak only 120 workers (adjusted figure: 32) before closing in 1999.

A clay pit in Cornwall employed some 500 people (adjusted figure: 133) just before it closed in 2007.

Power stations seem to be fairly heavy employers: the Rugeley power station employed 850 people (adjusted figure: 227) in 1983, and 146 people (adjusted figure: 39) people in modern times. This is a 1,000MW coal fired power station.

Gasworks, it seems, employed up to about 300 people (adjusted figure: 80).

A furniture factory in the US (source) which opened in teh 1880s and closed in 1985 is said to have employed "up to" 200 people (adjusted figure: 53), although it is not specified when this high rate of employment was obtained.

The Fawley Oil Refinery employs 2,300 people (adjusted figure: 613), and produces 330,000 barrels of oil a day. A barrel of oil is equivalent to 0.159 cubic meters, the measure used by the pakset. 330,000 barrels of oil per day therefore equates to 52,470 cubic meters per day. Because an oil refinery is a constant process production, we must assume a full 24 hour day rather than equate each day with a Simutrans month: in other words, we need to divide the production by 3.75, too. This brings us to 13,992 cubic meters of oil per Simutrans game month for a modern refinery.

Oil rigs, it seems, employ something close to 100 people (adjusted figure: 27) each.

Information on paper mills is not forthcoming.

Some Irish pharmaceuticals plants seem to employ between 79 and 167 (adjusted figures: 21 and 45) people.

The Linthorpe Pottery employed in 1885 between 80 and 100 workers (adjusted figures: 21 - 27)

The Blanefield printworks is reported to have employed between 250-300 people (adjusted figures: 67 - 80) in 1869.

Quarrying seems to have been labour intensive: the Calstock quarry, which opened in 1808 employed 700 people (187) in its unspecified heyday (perhaps the mid 19th century). It also reports that a single worker could cut about a ton of stone in a day, giving a daily (Simutrans-monthly) output of some 700 tonnes (although this seems to have been a particularly large quarry).
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.

prissi

There is also the problem that at an early stage it is highly unlikely that all towns in the vicinity can fullfill the production demand. Since in Simutrans a chain is only creating as much transport as its weakest like, but the destinations are chosen randomly, a too low base production will make industry games almost impossible.

jamespetts

Quote from: prissi on May 06, 2013, 09:40:47 PM
There is also the problem that at an early stage it is highly unlikely that all towns in the vicinity can fullfill the production demand. Since in Simutrans a chain is only creating as much transport as its weakest like, but the destinations are chosen randomly, a too low base production will make industry games almost impossible.

Ahh, thank you for pointing that out. Abandoning the system of having workers live in particular towns, as discussed above, might help here, but in the meantime I shall follow your advice and not set the base production too low: thank you.
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.