News:

Simutrans Chat Room
Where cool people of Simutrans can meet up.

Aircraft pakfile errors

Started by passengerpigeon, April 12, 2022, 06:11:57 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

passengerpigeon

Whilst running an airline network on the Bridgewater-Brunel server, I have noticed the following errors pertaining to aircraft specifications:
- Several aircraft have minimum takeoff distances that are obviously wrong, making them almost unusable in the early game. The Ford Trimotor requires 1,725m of runway, the JU-52 requires 1,850 and the Vickers Viscount requires over 2,400; I suspect that the first two are the result of unit confusion, since a runway of just under 2,000 feet, rather than metres, sounds like a realistic minimum takeoff distance for both. Also, the Lockheed L-049 needs a lot more runway than any of the later Constellations and the DC-6 needs less runway than the earlier and smaller DC-4.
- The Douglas DC-6B is set to carry Very Low passengers by default.
- A few planes such as the de Havilland Dragon and Beech 18 don't have enough power to ever reach their stated top speed.
- Not a bug, but the aircraft apron without a jetbridge on it, which cannot be built after 1977 currently, should never go obsolete; remote stands without bridges are still very widespread including at brand-new airports and the bridge gates look funny when not placed right next to a terminal.

Spenk009

For the power figures and runway lengths, most of these values are readily available figures. The issue with the runways can be safe lengths of runway and inconsistent data for each aircraft. For power, there is a limitation/bug in the physics engine, where increasing power does not increase actual top speed. Also Iirc, cruising speed is used for top speed as this is where planes are flown. If you can list what is wrong with evidence why it's wrong, a pull request is very likely to be accepted.

The DC-6B's wrong class assignment is a result of the dat file still using undefined passengers, instead of specifying the class. I would suggest changing the class to "High" and incorporating the comfort value that already exists.

I agree with the unjust ending of the non-jetway apron. If stations/stops had a "transfer time" and  "loading time" feature, the player could be given a choice.

jamespetts

Thank you for the report - I have removed the non-jetway apron retirement and added class definitions to the DC-6B. As to the minimum runway lengths - do you have data as to the correct minima for these aircraft? There is no point in changing these data unless we can be sure that we are changing them to the correct 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.

passengerpigeon

#3
Quote from: jamespetts on April 15, 2022, 12:50:48 PMThank you for the report - I have removed the non-jetway apron retirement and added class definitions to the DC-6B. As to the minimum runway lengths - do you have data as to the correct minima for these aircraft? There is no point in changing these data unless we can be sure that we are changing them to the correct values.

According to https://www.aopa.org/news-and-media/all-news/2010/may/01/tin-goose and https://www.ju-air.ch/en/about-ju-air/fleet/, the minimum runway length is 450m for the Junkers and 426m for the Trimotor. The latter figure apparently comes from the manufacturer; the former is from an airline that doesn't provide a source for the number but I don't think it would be far off. The website http://www.vickersviscount.net/Pages_Technical/700%20Series%20Specification%20Performance%20Weights%20and%20Dimensions.aspx says that the Viscount requires 1609m to take off with a full load at 30 degrees Celsius, or 1481m with an air temperature more typical of Britain. Try as I might, I can't find any takeoff distance figures that are definitely for the L-049 Constellation as opposed to later Connies, but I can't find anything suggesting it is much different from the 4,600ft/1402m required by the later models either.

passengerpigeon

#4
Al
Quote from: Spenk009 on April 13, 2022, 02:56:45 PMFor power, there is a limitation/bug in the physics engine, where increasing power does not increase actual top speed. Also Iirc, cruising speed is used for top speed as this is where planes are flown. If you can list what is wrong with evidence why it's wrong, a pull request is very likely to be accepted.

I know that cruise speed is used as the top speed, but the issue is that several planes aren't given enough power to reach their assigned cruise speed, and as such cruise slower and accelerate sluggishly, much like a train without enough motive power. The planes in question are:
- de Havilland Dragon Rapide
- Lockheed Model 10 Electra
- Fokker F27-100 and -200
- Aero Commander 520 and 680
- Sud Super Caravelle 10B
- Vickers Vanguard
- Beechcraft 18

Another bug regarding speed is that the speed limit implemented to prevent overland supersonic flight is too low; it should be raised from 950kph to 976kph. At the moment the Convair 880, with a cruise speed of 976kph, is slowing down over land even though it is subsonic. Also, both the Convair 880 and 990 should emit smoke particles, as they do in real-life.

In addition, I don't think any of the available aircraft should require more than 3,000m of runway, because Boston's longest runway is around that length and it has a direct A380 flight to Dubai. Situations in which a civilian aircraft would need more than that are limited to high altitudes (altitude not currently being simulated) and maybe, for the oldest jets, operating out of sea-level airports at completely full passenger and fuel load in boiling hot weather - conditions that are unlikely to actually be met in a game based on Britain where the maximum practical map size is well under the top range of any jet in the game.

One more suggestion for airports is that the obsolescence date for the original control tower building should be removed; if you want to build a small regional airport after it becomes unavailable, you need to use the huge modern-looking tower which appears out of place.

jamespetts

My apologies for the delay in replying to this: I must confess that I had missed the replies to this.

I have now done some work in this respect. I think that I have dealt with the aircraft performance issue; with most of the smaller aircraft, the issue was using the default air resistance, which is too high for smaller aircraft. However, the Caravelle 10B had an incorrect power rating, which I have also fixed.

As to the overland subsonic limit, I have raised this to 990km/h - the US apparently set its limit to mach 1, although aircraft generally did not fly in the transonic range, which was anything from mach 0.8 (circa 950km/h) upwards.

As to runway lengths, I know that most of these have been researched and calibrated very carefully by Giuseppe over the years, so I am reluctant to change them without being able to be very confident that the replacements are part of a consistent system applicable to all aircraft. I have altered the runway length for one or two very small aircraft where Giuseppe had not calibrated this and where I had not previously had data.

I have removed the obsolescence for the original control tower building, although we could very much do with a new small integrated control tower/terminal building, perhaps based on that at London City Airport so that players no longer need to build art deco style terminals into the 21st century.
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.

passengerpigeon

Thanks so much for fixing these bugs; the speed glitch with the Fokker F-27 was greatly slowing down the replacement of piston propliners and first-generation Viscounts in my Bridgewater-Brunel airline's fleet before because it was still slower than them. Unfortunately it appears that the Beechcraft King Air 200, Fairchild Metroliner, de Havilland Twin Otter, Dornier 228 and Embraer ERJ-145 are still affected by this same bug - sorry I didn't report them earlier, but Bridgewater-Brunel hadn't advanced to that time yet; I made a test save that allows buying obsolete vehicles and think that those are all of the planes with this error.

Quote from: jamespetts on May 24, 2022, 11:27:55 PMAs to runway lengths, I know that most of these have been researched and calibrated very carefully by Giuseppe over the years, so I am reluctant to change them without being able to be very confident that the replacements are part of a consistent system applicable to all aircraft.

I think that the runway lengths already are inconsistent - as for the Boeing 727-200, for instance, I looked at Boeing's documentation and found that it does indeed list the required takeoff distance at 3,250m, but this is for absolute maximum passenger and fuel load - considering that many intercontinental widebody airports don't even have a runway that long, and the 727-200 notably served airports such as New York-LaGuardia without any serious range or capacity restrictions, this figure seems to be more theoretical than anything and I doubt any aircraft was ever sent out that heavy in practice. I suspect Giuseppe looked at similar manuals for other Boeing aircraft and used this same theoretical figure based on absolute maximum load. On the other hand, the Caravelle III, which was known as an underpowered dog in real life, only requires 2,013m to take off - this figure seems to instead be based on a more reasonable passenger load rather than 100% MTOW. Similarly, the Boeing Stratocruiser needs 7,000 feet to take off at full load, not the in-game 1,100 metres. As I stated earlier, I think that all planes' takeoff distances should be calculated based on a standard load rather than a theoretical maximum (considering that the maximum practical map size, until the advent of international portals, is about the size of the British Isles) and 10,000ft (3,048m) should be the longest takeoff distance required by any aircraft in the game, with the possible exception of the Concorde, as this is considered an adequate runway length for intercontinental airports in real life, and runways no longer than this have served the 707, 747-100 and other more capable widebodies departing for international destinations without a problem.

jamespetts

Thank you for that. I have now recalibrated the air resistance for those aircraft, too, which I think fixes the problem.

As to runway length - if I am to adjust this, what I will need is a consistent and sensible way of going from readily available data for the whole range of aircraft in the pakset to the in-game number, since we do not simulate varying takeoff and landing weights. I should note that we cannot assume for these purposes that all aircraft will be going on short journeys with minimum fuel, since it is planned in due course to introduce an overseas journey feature, which would involve simulating aircraft and ships travelling to long-distance off-map destinations and back, making full use of the range parameter of existing aircraft. Currently, I do not know what such a consistent system would be.
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.

passengerpigeon

Quote from: jamespetts on May 25, 2022, 11:43:34 PMThank you for that. I have now recalibrated the air resistance for those aircraft, too, which I think fixes the problem.

As to runway length - if I am to adjust this, what I will need is a consistent and sensible way of going from readily available data for the whole range of aircraft in the pakset to the in-game number, since we do not simulate varying takeoff and landing weights. I should note that we cannot assume for these purposes that all aircraft will be going on short journeys with minimum fuel, since it is planned in due course to introduce an overseas journey feature, which would involve simulating aircraft and ships travelling to long-distance off-map destinations and back, making full use of the range parameter of existing aircraft. Currently, I do not know what such a consistent system would be.

I think that a better way of doing things would be to look up examples of the longer routes operated by each aircraft type, and then out of these long routes, which airports had the shortest runways, to get a good idea of what runway length was needed in practice to depart with a nearly full load. This is relatively easy to research, at least for newer planes (the old ones apart from the Viscount, JU-52 and Trimotor seem pretty accurate already). For instance, Airliners.net says that the 727-200 operated flights in excess of four hours from Minneapolis-St Paul to the Caribbean and Mexico, flying to Ixtapa with a runway of only 2500m and an even longer trip to Montego Bay-Sangster with a runway of 2,653m, in both cases configured in a high-density layout. The same website says that the 747-100 operated from Minneapolis to Tokyo-Narita, at the upper limit of its range, with the longest runway at Minneapolis being 3,355m. Because of the fact that the 747-100 had a particularly poor power to weight ratio, I still think that runway requirements for the other widebodies and the Boeing 707 should be capped at 3,048m, if no information can otherwise be found confirming that they can take off nearly full in a shorter distance.

jamespetts

Quote from: passengerpigeon on May 27, 2022, 12:46:25 AMI think that a better way of doing things would be to look up examples of the longer routes operated by each aircraft type, and then out of these long routes, which airports had the shortest runways, to get a good idea of what runway length was needed in practice to depart with a nearly full load. This is relatively easy to research, at least for newer planes (the old ones apart from the Viscount, JU-52 and Trimotor seem pretty accurate already). For instance, Airliners.net says that the 727-200 operated flights in excess of four hours from Minneapolis-St Paul to the Caribbean and Mexico, flying to Ixtapa with a runway of only 2500m and an even longer trip to Montego Bay-Sangster with a runway of 2,653m, in both cases configured in a high-density layout. The same website says that the 747-100 operated from Minneapolis to Tokyo-Narita, at the upper limit of its range, with the longest runway at Minneapolis being 3,355m. Because of the fact that the 747-100 had a particularly poor power to weight ratio, I still think that runway requirements for the other widebodies and the Boeing 707 should be capped at 3,048m, if no information can otherwise be found confirming that they can take off nearly full in a shorter distance.
I am afraid that I do not think that this would be workable: it would be too difficult to get either consistent or precise enough figures for all types of aircraft using this sort of technique, and would be excessively prone to error (especially where airports have had runways lengthened) and would be far too labour intensive compared to a solution based on aircraft data.
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.

1993matias

My apologies for digging up an old topic. One way to satisfy the requirement of easily available runway data would be to use minimum runway length requirements. If the only other option is to use maximum possible runway requirement in a worst-case scenario, then this might actually make more sense. Aircraft rarely fly at 100% allowed weight, thus rarely using maximum runway requirements.

Minimum runway length should be available for most popular aircraft types on wikipedia, which usually takes data from Jane's All The World Aircraft, a reputable source.