Proposal - Cargo Class refinement

[andythenorth]

The current specification of cargo classes could be refined.

Specifically with respect to bulk, which, as used in many sets, covers both 'bulky' and 'suitable for hoppers'.

This represents a problem for cargos such as logs, scrap metal and sugar cane, which are bulky and suitable for carrying in open wagons and trucks, but are not suitable for hoppers.

It is tiresome to impose the need for newgrf authors to handle these cargos explicitly, and negates the purpose of classes somewhat.
It is not ideal for players to have odd refit choices, some of which result in change requests to newgrf authors.

Is this not capable of being resolvable adequately within the current cargo label framework?

[Eddi]

the way i see it, there are currently three cases where the current classes are insufficient.

1. fine grained: These are typically seen in silo wagons for fine-grained bulk cargo like cement or fertilizer, but not for "large grained" bulk cargo like coal and ore, which technically can be poured, so it is suitible for hoppers, but does not flow well by itself, so cannot be put in silo wagons.
2. lightweigt: cargo like straw, cotton or sugar cane, that is usually not poured, but cannot be counted as "pieces" either. the wagons used for these kinds of cargo are usually very large, because they are limited by loading gauge, not by axle weight. they may be transported in open wagons or specialised cage wagons, but not in hoppers or silo wagons.
3. scrap metal: this is kind of a corner case, as it not really fits into any kind of class: it's not "piece" in the sense that it can be counted in regularly sized units (crates, palettes, ...), it isn't "bulk" [in the sense of the specs] because it can usually not be poured, it isn't lightweight either. scrap metal can only sensibly be transported in open wagons, not in hoppers or silo wagons, but neither in closed wagons, due to its irregular size

[andythenorth]

I appreciate that alternative cargo schemes have been discussed before and are not going to happen, but here's another one for the deadline:

- Cargos would set 'base' classes: bulk, piece goods, liquids (also PAX, mail for completeness, possibly express)
- Vehicles would indicate which base classes they transport

- Vehicles would also indicate various 'technical provisions', they support e.g. refrigerated, hazmat, hopper, silo, covered / sheltered, over-sized (some of these are enhancements, some limitations). I can't see a a requirement for vehicles to deal with excluding properties here.
- Cargos would AND and/or NOT with the technical provisions. This would be a property of the cargo, not the vehicle.

The scheme ends up almost the same, but the greater burden of support is moved to cargo authors, and reduced for vehicle authors. A significant number of vehicles and cargos could probably use just the 'base' classes.

The attraction for me is two-fold. First we could clean up some problematic cargo classes.

Second, we move the burden of work to the right place. A small number of industry sets have been created, and ECS encourages collaboration between sets to keep cargo definitions sane. Each set contains max. 32 cargos. A large number of vehicle sets have been created, and can contain many vehicles. More vehicle sets are being worked on than industry sets, and that seems likely to continue. It is hard work to provide reliable cargo support with the current system. Vehicle authors should just have to provide for what their vehicle *does* support and not what it doesn't.

Vehicle authors who wanted more fine-grained support would do as they have to do now: use cargo labels for explicit cargo support, which works well.

I wondered if there was a a way this could be implemented technically as an alternative schema (via new properties), thereby preserving backwards compatibility, e.g. new schema is only used where both cargo and vehicle implement the new properties.

Examples:
Coal might require bulk.
Logs might require (bulk or piece goods), NOT (hopper or silo)
Grain might require (bulk or piece goods) AND covered / sheltered

An open wagon would provide bulk, piece goods
An open hopper would provide bulk, hopper
A covered hopper would provide bulk, hopper, covered / sheltered

The cargo author would test and find that this leaves many silo wagons refittable to coal; so they might then add NOT silo to coal cargo.

So:

• it's broadly the same as the current scheme, but moves some of the work to cargo definitions
• it's unlikely to happen
• it's probably nothing that hasn't been discussed before

However I thought it was worth writing down.

[DJ Nekkid]

As its 4am right now (and i probably have a 'too high pro-mille (10th of a thousund part alcholol/blood-tingy), this post is most likely gonna be edited tomorrow, and to me is the 'express' cargo as much as a pain as the bulk cargo.

I could probably do some preprocessing work with the goods cargo, but the bulk and liquid cargo is a paint to get 100% correct. As (afaik)v is i.e. grain a 'bulk' cargo, but not a 'covered' cargo. And a 'silo' cargo should be both? It could be carried in hoppers, but it should also be covered by a tarpantaiullin to prevent rain from letting the grain 'go bad'.

bleh! tttyall 2morrow!

[andythenorth]

Following some discussion on irc:
- no new properties
- there seems to be a case for discussing a small number of additional classes, mostly relating to differentiating 'bulk' further
- a callback provided for cargos to determine whether refitting is allowed, with a small number of associated vars

This will improve the possibilities for support of unmaintained vehicle sets by cargo set authors. Many of the older sets don't use cargo classes, or use them incompletely. A callback would be able to use various methods to work around this.

It would also be quite simple to establish conventions for vehicle set authors to follow with respect to which classes to set for which vehicle types, e.g. for open wagons, flatbed trucks, heavy cargo planes etc. Cargo set authors already collaborate on classes in a similar way.

[andythenorth]

Following discussion (thanks to those who helped / pointed and laughed)...

Proposals for two new classes.

1. 'clean'. This is primarily for use with the auto-refitting / refitting cost cb added to Openttd around r23087. 'Clean' was originally suggested as 'food' then expanded to be more generic. This allows for more sane control refitting by class without knowledge of specific cargoes, e.g. oil tanker to milk tanker (milk would require 'clean' + have a cost); fish to cotton (cotton would require 'clean').

2. 'neo-bulk'. This is applied to cargos like logs, automobiles, sugar cane. Currently these are usually treated as piece goods or bulk, but this can cause odd results, such as logs travelling by box car, or sugar cane travelling by hopper. 'neo-bulk' is the term used by the shipping industry for these types of cargo, which are bulky but not 'bulk' (they don't pour or flow). Adding this class would be sufficient to allow newgrf authors to distinguish this aspect clearly without knowledge of specific cargos. E.g. open wagons and flat wagons would typically allow neo-bulk, but vans would exclude it.

Proposal to clarify 'oversized' class (bit 10).
From spec, I read this as being similar to US 'high and wide' or 'dimensional load'. E.g. things requiring specialist wagons, like heavy flatcar, or well wagon, possibly with speed restrictions etc. Is this the intention?

Proposal to add some of these classes to default cargos, including 'clean' for Grain / Wheat / Maize / Fruit, and 'neo-bulk' for Wood. (There may be others I missed).

Question: are livestock properly 'neo-bulk'? Cargo classification pages online suggest this is a valid designation, but does it help with newgrfs?

I intend to modify some FIRS cargo labels and classes to suit this. Will post in FIRS thread. Also will add support for new labels to HEQS and FISH.

Comments?

[Eddi]

Proposal to clarify 'oversized' class (bit 10).
From spec, I read this as being similar to US 'high and wide' or 'dimensional load'. E.g. things requiring specialist wagons, like heavy flatcar, or well wagon, possibly with speed restrictions etc. Is this the intention?

there may be two distinct possiblities here:

1. "things that are piece goods, but are too large to sensibly fit in a closed wagon"
   e.g.: cars, machines, wood (logs), steel (rods, pipes)
2. "things that are so large, that their weight hardly matters"
   e.g.: wool, milk (cans), straw/fibre crops/sugar cane, cars

[andythenorth]

Proposal to clarify 'oversized' class (bit 10).
From spec, I read this as being similar to US 'high and wide' or 'dimensional load'. E.g. things requiring specialist wagons, like heavy flatcar, or well wagon, possibly with speed restrictions etc. Is this the intention?

there may be two distinct possiblities here:

1. "things that are piece goods, but are too large to sensibly fit in a closed wagon"
   e.g.: cars, machines, wood (logs), steel (rods, pipes)
2. "things that are so large, that their weight hardly matters"
   e.g.: wool, milk (cans), straw/fibre crops/sugar cane, cars

Or...things like transformers, fractional distilling columns, windfarm components, aircraft fuselages, locomotives, heavy equipment, buildings etc...

[Eddi]

i should have expressed myself more clearly, i meant to prepend "in order to make this useful for cargos that are actually defined in one or more industry current sets..."

[andythenorth]

I updated the wiki with new classes:
http://newgrf-specs.tt-wiki.net/wiki/Ac ... s_.2816.29

[michael blunck]

I updated the wiki with new classes:

You´re too fast, Andy. Given the fact that there are only two classes left now, the right to exist for these proposed classes has to be discussed in a broader audience (newGRF set developers) and in a much more detailed way.

["clean"]
IMO, this is a clash of categories. Firstly, it is based on a specific refitting process, but all other cargo class terms are based on special transportation needs of the cargo itself. Secondly, it seems to be merely introduced to change the refitting costs, and not to make newGRF developers´ lives easier. I.e., if you want refitting to "milk" to be more expensive, you don´t need to add a new cargo class altogether, but you can do it by playing with refitting cost now.

Moreover, I don´t see the usefulness of "clean transportation" at all. Both in-game and in RL things like "foodstuffs" or "textiles" are packaged appropriately. No need to complicate the current system any further. Refrigerator cars are usually only used for fish, food, fruit (or milk). They´re never used for "dirty" cargoes, hence no need for "clean" (even in addition to REEF).

["neo-bulk"]
Again, I don´t see how introducing this class would add to a better exposure of current cargoes or in less work for the newGRF developer?

Logs, coils, pipes, cars, paper rolls, cellulose, cement, and a lot of other cargoes are also marked as "neo-bulk" in RL, but can easily be handled in-game as "piece goods" (or bulk). In fact, for the earlier years this makes even more sense because of low quantities being transported back then. And IMO, the term as such would be generally only useful for today´s "mass transportation" by ships, but not for trains or trucks.

E.g. open wagons and flat wagons would typically allow neo-bulk, but vans would exclude it.

Sorry, but I don´t see the point. In my sets, open wagons are generally enabled both for "bulk" (ore, sand, grain, ...) and "piece goods" (steel products, automobiles, ...). Flat wagons and box vans are generally not enabled for "bulk". Hoppers, OTOH, don´t load "piece goods".

"Cars" are transported in closed vans, not only in the past but especially today (the DB just bought a whole new fleet of closed car transporters for some new Daimler car models). So, this class is quite useless, because its constituents are not only transported by open wagons or flatbeds but also by box vans.

In general, I don´t think that refinement of existing cargo classes would lead any further, simply because by using cargo subtypes, there´s too many variants of most cargoes, possibly requiring different means of transportation.


And BTW, refering to a post of Eddi, ECS has already introduced a new cargo class, namely:

-----------------------
bit=10, value=800, "powder-shaped material" (bulk cargo needing
special silo-wagons, e.g. equipped for compressed air unloading; for
cargoes like cement, plaster, lime granulates, coal dust, plastic
granulate or powder, PVC, metal powders, salt, sugar, semolina, ...)
-----------------------

regards
Michael

[Michi_cc]

You´re too fast, Andy. Given the fact that there are only two classes left now, the right to exist for these proposed classes has to be discussed in a broader audience (newGRF set developers) and in a much more detailed way.

<snip snip>

And BTW, refering to a post of Eddi, ECS has already introduced a new cargo class, namely:

Where's the broad, detailed discussion about this ECS cargo class then?


-- Michael Lutz

[frosch]

I prefer some example to get away from the abstract discussion.

From my point of view there are two classes which need clarification:

• Oversized/overweight
• Neo-bulk

While the "neo-bulk" definition is clear from the wiki, the "oversized/overweight" definition is not. I can give it these three possible definitions:

• Stuff that is only transportable via flatbed-wagons, not open-wagons. This would fit the example cargos "Glass" and "Vehicles".
• Stuff that need loading using a crane, i.e. suitable for flatbad wagons, open-wagons. This would be essentially the same as "neo-bulk".
• Stuff that is only transportable by vehicles specialize to transport exactly that cargo. While this actually matches the definition closely, it is utter non-sense. Such cargos should have no cargo classes at all, and the vehicles should specifically set them.

To add more examples to the discussion I have prepared some table on the wiki, which adds new classes to the default cargos.
The table has currently these issues:

• It assumes that there is actually no vehicle set released, which uses the newer classes "covered/sheltered" and "overweight/oversized". So no GRF breaks when the cargo classes of default cargos are changed.
• "Oversized/overweight" is most likely uses incorrect due to the unclear definition.
• Toyland cargos are quite mood to even start with.

Any opinnions or corrections on these assumptions?

[andythenorth]

You´re too fast, Andy.

Feel free to revert, I may have been too eager.

Meanwhile, here's a nice link to some quotes from Grace Hopper http://en.wikiquote.org/wiki/Grace_Hopper

[andythenorth]

Maybe we should just have namespaced cargo labels By default to the set defining them (firs.foo, manual-industries.bar, ecs.ham, default.eggs etc), but with the ability for a set to reuse cargos that are declared in another set's namespace. Meh.

Anyway, for a vehicle author, I understand the cb route as working something like this (order might vary). For any given vehicle:

1. explicitly allow certain known cargos via labels (in the set CTT)
2. explicitly disallow certain known cargos via labels (in the set CTT)
3. check classes, use advanced varact 2 or branching to allow or disallow various combinations of class specific to the capabilities of the vehicle.
4. depending on the vehicle type, finally either allow or disallow all unknown classes. E.g. open wagon, cargo ship might allow anything, other vehicles might disallow*.

*Allowing all is more future proof, but will have a not-ideal result where future new classes are intended to be restrictive, e.g. 'vehicle must be red' or whatever.

This last point might suggest that adding new restrictive classes is not wholly wise....classes are supposed to be future proofing vehicle/cargo set compatibility by enabling vehicles to refit to unknown cargo. How does a vehicle author allow future refitting of unknowns, whilst also future proofing against as-yet-unknown classes which are intended to prevent refitting?.

--

Separately to this, I find it very unclear what cargo set authors are supposed to do when setting classes with prop 16. When multiple classes are to be set, does the spec require that these classes are AND or OR?

The documentation of train prop 28/29 suggests very clearly that the spec requires multiple classes to be AND.

I believe that there are numerous cases where multiple classes are being defined based on OR.

I suggest that if spec is AND, as per train prop 28/29 docs, then FIRS has been doing it entirely wrong and needs to change.

If the spec is "it doesn't matter whether it's AND or OR" then I suggest the spec is broken.

[PikkaBird]

This last point might suggest that adding new restrictive classes is not wholly wise....classes are supposed to be future proofing vehicle/cargo set compatibility by enabling vehicles to refit to unknown cargo. How does a vehicle author allow future refitting of unknowns, whilst also future proofing against as-yet-unknown classes which are intended to prevent refitting?.

Hear hear. Personally I think it unwise to create a cargo without any of the "classic" bits 0-6 set, as those are the only classes most vehicle sets will check for. Bits 7 and upwards should be considered only as additional information, for the benefit of sets which want to provide a larger-than-strictly-necessary range of vehicles or vehicle appearances. "Neo-bulk" cargos should always also have "piece goods" set, for example.

[Eddi]

And BTW, refering to a post of Eddi, ECS has already introduced a new cargo class, namely:

-----------------------
bit=10, value=800, "powder-shaped material" (bulk cargo needing
special silo-wagons, e.g. equipped for compressed air unloading; for
cargoes like cement, plaster, lime granulates, coal dust, plastic
granulate or powder, PVC, metal powders, salt, sugar, semolina, ...)
-----------------------

regards
Michael

so, where is that in the specs?

[Eddi]

from an (ongoing) discussion, here is my current opionion:

the current "cargo classes" scheme is lacking a definition of whether the classes are meant as "one of them must apply" (logical OR) or "all of them must apply" (logical AND). for both cases there are examples where it would be useful (e.g. this cargo appears as "bulk" OR it appears as "piece goods" (usually not at the same time), while other cargo may be "piece goods" AND "needs special protection (armored)")

i hereby present my proposal:
split the current cargo classes into two sets:

• a basic set describing the general consistency:
  • countable (formerly piece goods, including passengers and mail)
  • uncountable (formerly bulk)
  • liquid
  these would be ORed, i.e. a cargo defining itself as "liquid, countable" would be possible to transport in an open wagon (as barrels/cans/whatever) or in a tank wagon (as liquid).
• a specific set of additional requirements, e.g.:
  • self-moving (passengers)
  • alive (livestock)
  • pourable (coal, ore, etc.)
  • not pourable (wood, cars, etc.)
  • refrigerated
  • armored
  • large (machines, wood, steel (rods/pipes/rails))
  • light (wool, fibre crops, etc.)
  • heavy (steel, machines)
  • covered
  • ...
  these would be ANDed, i.e. a wagon must provide facilities for handling "large" AND "heavy" objects in order to load steel.

[wallyweb]

Having followed this discussion and a related one, I became concerned about added layers of complexity.

...

I like this proposal. It is elegant in simplicity and logic.

[michael blunck]

From both old and ongoing discussions, here is my current opinion:

Proposal

With more and more "generic" cargoes being introduced, their graphical representation needs more and more cargo subtypes and their allocation needs additional entries in the vehicle´s cargo bit mask, thus the established way to primarily use cargo classes seems no longer feasible.

E.g., FIRS has lots of generic cargoes: FMSP ("farm supplies") could be tractors or other farm vehicles and equipment, fertilizer in bulk, or tools and other small machines, each requiring different cargo graphics, different means of transportation, and hence appropriate cargo mask entries. BDMT ("building materials") could be either lumber, stone or bricks, steel construction material, cement, etc. All requiring different cargo graphics and different means of transportation, from box vans, open wagons and flat beds to special silo-wagons.

Now, this proposal deals with giving up the system of "cargo classes" and to introduce a system of "wagon classes".

Best bet would be to use the "UIC classification of goods wagons" instead of re-inventing the wheel. In this way, the scheme should fit for goods wagons from a lot of countries, and the scheme would by untroubled by exotic entries as we now have (hazard, livery refit tricks). The new standard would be based on existing (railway) vehicle classes. (At first, this may sound strange for trucks, ships and aircraft, but o/c even a truck or a ship could signal "I´m a tanker and I´m able to protect goods against moisture as well".)

This is the original UIC scheme (from 1986):

E - ordinary open high-sided wagons
These are used primarily for transportation of bulk goods that are not moisture-susceptible and can usually be tipped, dumped or shovelled (typical freight would be coal, scrap metal, steel, wood, ..).

F - special open high-sided wagons
These are mainly self-discharging hoppers which use gravity-unloading, but also side-tipping wagons for all sorts of bulk goods (coal, coke, ore, sand, gravel).

G - ordinary covered wagons
These are the "standard" railway wagons, for transportation of all moisture-susceptible goods, usually packaged in boxes, sacks and barrels. Also for moisture-retentive bulk material, as well as for express goods and mail. In early times used for cattle transport as well.

H - special covered wagons
These are either (older) wagons for small livestock and cattle, or havong been used for automobile transportation, as well as modern sliding wall wagons for palletised goods.

I - refrigerated vans
These are special wagons with cooling equipment for food, fruit, fish, etc.

K - ordinary flat wagons with separate axles
These are 2-axle wagons (with or w/o stakes) for loads too large for box cars, like vehicles, machinery, containers, long piece goods like lumber or steel beams and reinforcement fabrics, but also for "bulky" loads (wood) if equipped with sturdy end-walls.

L - special flat wagons with separate axles
These are heavy capacity flatcars, depressed center flatcars, automobile transporters, container wagons, intermodal trailer transporters, etc.

*O - open multi-purpose wagons (composite open high-sided flat wagon)
*R - ordinary flat wagons with bogies
*S - special flat wagons with bogies

T - goods wagons with opening roof
These are relatively new wagon types, with different opening roof systems (rolling, sliding, hoods), primarily used for moisture-sensitive bulk cargoes like cement, gypsum, lime, potash, or cereals. They´re also used for large "bulky" piecegoods to be loaded by cranes.

U - special wagons
These are mainly silo-wagons for transportation of pulverulent cargoes (both food: salt, sugar, semolina, ...; and non-food: cement, plastic granulate, plaster, ...); but also very special low-loader wagons (e.g. "Schnabel" cars).

Z - tank wagons
These are used for transportation of all types of liquid and gaseous commodities (fuel, light fuel, gasoline; tar, lubricants, heavy oil; petrochemical products, gases, chemical products; foodstuff).

In addition (beyond the UIC scheme), we´ll need special wagons for transportation of passengers (coaches) and mail (mail vans). OTOH, the UIC scheme for goods wagons has dual entries for flat wagons with and w/o bogies (* ::= K,L -> R,S). In game terms, this distinction is unneeded, because it does not affect the cargo types being transported. Same goes for UIC class "O".

So, in this way, the proposed scheme would look like this:

A - passenger coaches
D - mail vans, also usable for "valuables"
E - ordinary open high- and low-sided wagons
F - special open high-sided wagons
G - ordinary covered wagons, box cars, vans
H - special covered wagons, livestock transporters, sliding wall wagons
I - refrigerator wagons
K - ordinary flat wagons
L - special flat wagons, automobile transporters
T - goods wagons with opening roof
U - special wagons, silo wagons, low-loaders
Z - tank wagons
----------
total 12 bits

Now, how to map these onto the existing cargo classes? For new cargoes, this shouldn´t be a problem, maybe only for the original cargos.

Mapping of original cargo classes:

Code: Select all

bit value	old cargo class	new cargo class
0   1	passengers	-> A
1   2	mail		-> D
2   4	express
3   8	armoured
4   10	bulk		-> E
5   20	piece goods	-> G
6   40	liquid		-> Z
7   80	refrigerated	-> I
---------
8   100	hazardous
9   200	covered/sheltered	-> T
10  400	oversized/overweight	-> U
11  800	pulverulent		-> U

Classes "express" and "armoured" could stay as before (for compatibility reasons), but in fact, there´s not much use of them in the new system. "Armoured" was always very dubious in terms of RL, and "express" is a rather meaningless cargo class, given the fact that the game cannot handle "prioritized" goods in a special way at all.

How does the new scheme work?

As before, these are only bits set by cargo action0 prop16.

TTD original cargoes: s.a.

ECS cargoes:

Code: Select all

label	name	    old class			new class
BRCK	bricks	    20 piece goods		[G], (E)
CERA	ceramics   20 piece goods		G
CERE	cereals	    210 bulk, covered/sheltered	E, T, (F)
CMNT	cement	    A10 bulk covered/sheltered, powder	E, T, U
DYES	dyes	    860 piece goods, liquids, powder	G, Z, U
FERT	fertilizer	    30 bulk, piece goods		E, G, (F, T)
FICR	fibre crops 30 bulk, piece goods		E, G, 
FISH	fish	    84 express, refrigerated	I
GLAS	glass	    420 piece goods, oversized	G, U
LIME	lime stone 10 bulk			E
OLSD	oil seed	    210 bulk, covered/sheltered	E, T, (F)
PETR	petrol	    40 liquid			Z
PLAS	plastic	    860 piece goods, liquid, powder	G, Z, U
POTA	potash	    210 bulk, covered/sheltered	E, T, (F)
RFPR	refined pr. 40 liquid			Z
SAND	sand	    10 bulk			E
SULP	sulphur	    210 bulk, covered/sheltered	E, T, (F)
TOUR	tourists	    05 passengers, express		A
VEHI	vehicles	    420 piece goods, oversized	[G], (K, L)
WDPR	wood pr.   30 bulk, piece goods		E, [G] (K, T) 
WOOL	wool	    220 piece goods, covered/sheltered G, T, (H)

A similar scheme could be easily set up for FIRS.

regards
Michael