Improved airports

[jamespetts]

I have been working on some improvements to airports this week-end whislt I am waiting for Bernd to finish working on tracking down network desyncs related to the braking physics code. The results are on my -devel branch (and see the Pak128.Britain-Ex Github repository for the compatible pakset).

The new features are:
(1) aircraft all have a minimum runway length in meters, and cannot take off/land from a runway shorter than that length; and
(2) all airports must have control towers.

The second feature is disabled by default, as many paksets do not define control towers (is_control_tower=1 in the .dat file for an extension building). This feature is enabled by setting allow_airports_without_control_towers = 0 in simuconf.tab. Airports without control towers show a purple status colour and have [NO CONTROL TOWER] appended to their names in the schedule and information windows.

As to the first feature, the minimum runway length for each type of aircraft is set by specifying minimum_runway_length=[a number in meters] in the individual aircrafts' .dat files. The figure is displayed in the depot/replacer window (only when it is dealing with aircraft, however). The runway length is measured using the meters per tile setting.

These changes involve changing the pakset and saved game formats, so this is incompatible with previous -devel builds (as ever, this does not affect release builds).

I hope to see some rather more realistic looking airports in the next server game! Incidentally, as to the difficulties encountered previously with airports not having enough capacity, this has not been fixed with changes to the code, but the following pakset changes should between them alleviate the issue:
(1) larger aircraft (thanks to Giuseppe);
(2) a lower passenger factor; and
(3) a lower proportion of passengers travelling long distances;
and the following user tips should likewise assist:
(1) do not use aircraft for short journeys (approx. 150km); and
(2) for large airports, have multiple terminals (each a separate stop, linked with land transport), with its own runway.

I should be very interested in any thoughts on these matters, and anyone who is able to compile from -devel to test.

[ӔO]

Nice work!
Multiple terminals and avoiding short hops sounds like a good idea.


Were one way signals were added to the pakset? I've realized that you can totally jam an airport if there is only one runway and too many airplanes trying to take off or land.

[jamespetts]

Thank you!

Do you mean one way signs for runways or taxiways? Taxiways I don't think enforce collision detection/occupancy, and runways are inherently one way for taking off/landing. How would you imagine one way signs working in airports?

[ӔO]

One way signs for both, although just on the taxiway is fine.
You can designate landing only and takeoff only if the taxiway to the runway is one way only. This would improve traffic flow, as the airplanes tend to just use the nearest runway possible.

For reasons that I don't really understand, sometimes an airplane can queue itself on the runway for a parking spot at a terminal. This causes the runway to become occupied and any airplane that wants to take off, can't. This would never happen if there is one runway for landing and one for take off.

There's already quite a good one way sign in pak128 that fits in just fine with pak brit.

[jamespetts]

Ahh, I see what you mean, a landing only runway and a takeoff only runway - isn't that what they do at Heathrow? You don't happen to have a link to the sources of the Pak128 one way sign, do you...?

[ӔO]

hmmm, I think you might have to ask fabio for the link to the sources.

All I know is that it's part of a "airports.misc.pak" and needs to be extracted.
The easily identifiable double arrows should be "roadsign.airport_oneway.pak".

[isidoro]

Giving a game function to control towers is very nice.  But make them compulsory is a bit too much imho.

Other possibilities:

   
• Make them compulsory only after certain date (they need radio, don't they)
• Make them compulsory only for big airports (big may mean with several runways, with at least one of the runways longer that a certain length (which, btw, would allow to indirectly connect this to point 1 if you force heavy planes from a date on to use longer runways))...
• Make them optional but, if there, waiting time for landing/taking off is shorter
• ...

Regarding the length of the runways topic, how would you mark that an airplane cannot take off/land due to the runway being too short?  Or would this be calculated when routing only?

[jamespetts]

AEO -

thank you for that. Is there a screenshot of it somewhere that might tell me whether it really fits in properly with Pak128.Britain colours?

Isidoro -

why is it too much to have control towers compulsory? After all, they are in real life! Note also that the feature can be disabled.

As to the second question - I have only been able to make it work calculating at the routing stage, so aircraft get an orange tooltip, "runway too short" when they try to take off/land. I'd like to be able to have a more sophisticated method of doing this, ideally enabling the runway length to be shown in the airport's information window, but I'm not sure how it might be done at present.

[ӔO]

here is what it looks like in pak brit.

[jamespetts]

Thank you - that is helpful!

[dom700]

Whats the typical runway length you are having in mind? Im just wondering how this change would look like for pak128 with the A380 having HUGE runways.

[greenling]

Jamrespetts
You want build in zones for Landing and Starting?

[jamespetts]

Whats the typical runway length you are having in mind? Im just wondering how this change would look like for pak128 with the A380 having HUGE runways.

I try to find the real runway length for the actual aircraft in question, and then use Simutrans-Experimental's meters per tile setting to translate that into tiles.

[ӔO]

The runway length required at full load for the larger aircraft is typically 3500m.

An-225 needs 3500m at full load.
747 series needs slightly less than that. Typically around 3200m to 3300m
A380 needs even less than that. Its lifting capabilities just outperform 747 by a wide margin.

For sizes of the larger airplanes, you can see here: http://upload.wikimedia.org/wikipedia/commons/5/52/Giant_planes_comparison.svg
A380 is very stubby with an advanced wing design that allows it to have some impressive lifting capabilities, which also keeps runway length to a minimum. Boeing 747 and An-225, on the other hand, are built with 70's technology, which are not as good, and require the extra length.

Your typical short runway is under 2200m, while the longer ones can range from 3400m to 4000m.
Most of the major airports were rebuilt with 747 in mind, since that came about in the 1970's. Quite a lot of them only have runways that are 3200~3500m long. Quite a lot of airports do get modernized over the years and one thing they do is add runway length to add a margin of safety.

Runway lengths over 4000m should, typically, not be necessary for commercial aircraft, but they do exist. For instance, high altitude airports. Since the air is thinner higher up, the runway length needs increase to compensate for the lack of performance. I don't think it's necessary to program altitude and air thickness, but I only mention it for the sake of completeness.

[jamespetts]

AEO,

interesting information. May I ask - where do you find the information on these minimum runway lengths? I ask because the lengths seem to be greater than my research has so far indicated.

[ӔO]

@james

There is usually a specification publication on various airplanes floating around.
except for airbus, who don't seem like they want to publish their data too much.

Boeing is pretty good about that: http://www.boeing.com/commercial/airports/plan_manuals.html
Just select an airplane and then select "airplane performance"

What I do know about A380 is that they had to design it so that it would work with most airports without too many modifications, which means it had to perform just as well or better than a 747.

[prissi]

Looking at those specs (like for the 707) you find all length between 2300 and 4500 (extrapolated) for wet and dry wether, different engine types and airport elevations.

Maybe a range is gamewise rather better to implement and easier to research.

[jamespetts]

Interesting information - thank you!

[Milko]

Hello James

I can help? 
I have a lot of information on aircraft that I designed (typically data will ricavavo technical manuals).
I read that you mean to use the track length in meters and turn it into tiles, you are not likely to have slopes that are too long?
For example, with m / = 200 tiles, a 3.5km long runway was made ​​of 17 tiles, are a little too much?
You have a makeobj windows for testing?

Giuseppe

[Carl]

Since the new default value for meters per tile in pak.britain will (IIRC) be 125, even a 2.5km runway would be 20 tiles. That sounds like a lot, but it sounds perfectly acceptable to me. Requiring proper airport infrastructure will help to balance air travel in relation to other modes of transport (and why shouldn't airports be realistically sized?)


I had only one small comment on your initial post, James. I think that flights of <150km are relatively common in the USA as part of hub-and-spoke systems. For instance, Boston airport has flights to nearby Barnstable, Vineyard Haven and Provincetown -- all of which are about 100km away.

[el_slapper]

<150km air transport is really special.

I like the idea of correct runway sizes. 20 squares will force me to think harder. At the same time, I already imagine the newbie hordes swarming the forum & complaining that air transport is broken. The runway size should be clearly visible from many screens, on the main screen(maybe a tooltip), the depot, etc..... interface has to be extremely friendly, on that one.

[jamespetts]

Interesting discussion. I ought to note that 150km was a very approximate figure, and there may well be exceptions to general rules in any event, but the basic point that aircraft ought be used only for long distances still holds, I think.

It would indeed be most interesting to see realistically sized and designed airports - much more fun than a two tile long runway with no control tower, I think!

[wlindley]

The 150km figure ought to be derived from some ratio of maximum, average, or speedbonus speeds... so it will change with technology as the game progresses.

Should the same sort of minimum distance calculation hold also for maglev trains?  Or indeed for horse-drawn omnibus when compared to the speed of walking... isn't there already something like that? perhaps simply generalize it.

[ӔO]

I would suggest checking some of the high traffic areas.


The only one I know of, which has quite a fierce competition on rail, air and road, is Tokyo to Osaka. That's around 500~520km.
It is approximately 3hrs by train, 1hr by airplane and 7hrs by bus.

-3hrs by train. Quite obvious and this is already programmed inside experimental. Next to no waiting time. Fair is slightly cheaper than airplane.
-1hr by airplane. Around 1hr of waiting time, and another 1hr of traveling to get to and from the airports to major rail hubs. That's 3hrs total. What's more, there is a limit to how many flights can be flown in a day.
-7hrs by bus. This is the slightly mysterious one for experimental. What it is, is a cheap, usually overnight, ride. It's very cheap compared to the other two, but the time consumed traveling is done while sleeping. It is quite popular, despite its long travel time.

For britain, I'm guessing London to Edinburgh or London to Paris? I'm not entirely sure where all the airports are for UK, nor do I fully understand where the high speed trains run.

[Carl]

The only "official" highspeed line in the UK is the one to Paris -- though the west coast/east coast main lines (and others?) run at 125mph.

While it's true that there's a limit to the amount of flights that can be run in a day, the limit is pretty high. You can run an hourly air service -- I believe this more or less exists on routes like Heathrow to New York -- and there's no reason in principle why you couldn't double that, if demand were there.

[jamespetts]

We can't do cost comparisons in Experimental because that would be a bit too complicated (to code, for the computer to do quickly enough for good performance in large games, and for players to do without the game ceasing to be fun), but a the journey time based computations of rail versus air are well set out. London to Edinburgh, London to Glasgow and London to Manchester (as well as London to Paris) are good examples of similar journey time comparisons, although in the Anglo-Scottish routes, the rail journey is considerably longer than three hours, and for the London to Manchester route, the flight is shorter than an hour, I think.

[kierongreen]

Unless you are connecting to other flights, London - Paris is now dominated by Eurostar (TGV train). By air it's a bit under 3 hours (45 minutes flying, 2 hours checkin and getting to airport), Eurostar is a bit over 2 hours. Coach isn't really competitive, maybe 7 hours or so as the coaches have to go via Eurotunnel (checkin time 30+ minutes) or via the ferry (checkin time again, and longer crossing of the Channel).

London - Edinburgh or Glasgow is more balanced. The journey as the crow flies is further than to Paris, the railways are slower speed, but they are more direct. Flying is about 3 hours with travel and checkin, train is 4 1/2, coach is 8 hours. Coach is extremely cheap, plane is often cheaper than the train, but not always. As with the example given in Japan, coach travel can be done overnight on this distance. All are used extensively to cater for different types of passengers, and all run regularly (at least hourly for trains and planes during the day). Coaches are more bunched up in terms of timetables, lots leave London or Edinburgh/Glasgow late at night and arrive early in the morning, and again, lots leave early in the morning and arrive early evening.

Elsewhere, across England the vast majority of public transport is split between coach (very cheap) and train (usually expensive, but some cheap tickets). Comfort is usually far better on trains than coaches though, unless you happen to get stuck on an overcrowded train of course (these actually aren't that common and are usually predictable). Air travel doesn't compete on the high traffic routes within England (e.g. London - Birmingham or Manchester) as the distances are too small (only lightly used routes from North East to South West are worth flying).

Intercity lines in the UK run mostly at 100mph or more - a lot of the Great Western, East Coast and West Coast have been upgraded to 125mph. There is a prospect that conventional lines could be relatively easily upgraded to 140mph also (this has been tested before). Due to relatively high speeds on conventional railways High Speed Rail within the UK is unlikely to drastically reduce journey times, on the longest journeys to Scotland for example it will probably only reduce them by 1 hour. However congestion is a major problem on British Railways - during the 1960's severe cutbacks left a network now unable to cope with todays levels of traffic.

The transatlantic routes don't have an hourly service - the flights tend to all run at more or less similar times, so there will be maybe 20 flights leaving late morning from various airports in the UK over the period of about 2 hours, getting into various airports in the US in the mid-afternoon. Hence you don't really get a choice as to what time of day you fly at, just where you are flying to and from (and with which airline).

[jamespetts]

Hello James

I can help? 
I have a lot of information on aircraft that I designed (typically data will ricavavo technical manuals).
I read that you mean to use the track length in meters and turn it into tiles, you are not likely to have slopes that are too long?
For example, with m / = 200 tiles, a 3.5km long runway was made ​​of 17 tiles, are a little too much?
You have a makeobj windows for testing?

Giuseppe

Oops - almost forgot to reply to this. I don't think that the runway lengths would be too long: as Carl says, long runways are realistic. In fact, the next version of Pak128.Britain-Ex will have 125m/tile, so the runways will be even longer: a 3.5km runway will be 28 tiles long. This will, as Carl points out, force people to put down realistic airport infrastructure, and make airports more resource intensive (and less suitable to be placed very near cities) than they have been in Simutrans in the past. This is a good thing, I think.

Yes, your help in adding minimum runway lengths would be very welcome! You don't actually need a -devel makeobj to code the values into the .dat files, as the standard release makeobj will simply ignore the strings. Adding the minimum runway length settings from your information for all aircraft (and replacing my values if I have got them wrong) would be very useful indeed (especially if you could do it on your Github branch so that I can simply merge your changes).

If you want to test it, however, you will need not only the latest -devel makeobj, but a binary build from the latest -devel branch, as the existing release candidate is too old to be able to read files written. I can upload these for you if you'd find that helpful?

[kierongreen]

Another interesting side effect of forcing runways to be 30 tiles or so long is that planes take over half a minute of real time to taxi down this length of runway. Even time taken to take off and land (demanding full use of runway) will be 10 seconds or so at least. That's ontop of the taxiing time of course. Capacity at airports will be very much dependent on having parallel taxiways to keep the main runway clear. Maybe two runways will minimise need for taxiing but I'm not sure. Of course, all this extra time will add to total journey time as well. For comparison, a two track railway in simutrans can support high speed trains running every five seconds (in standard simutrans anyway, I'm not sure whether braking distances change this for experimental?), and those trains will also be carrying more passengers each than any airliner.

So, capacity will be vastly reduced from at present, while maintenance costs and land required will be much greater. Looking at it just now it would seem that late on in the game air travel will also be restricted to extremely long distance routes as effectively they will start off with over 100 tile extra to travel over trains, and added to this is the time taken to travel this distance. So for distances of less than around 600 tiles (75km) planes will take longer than high speed trains to cover this distance. With transfer times and getting to airports you can probably double this. So for map sizes of less than 1000x1000 planes will not be practical, and you'll need to be playing map sizes of well over 2000x2000 (once you take into account positions of cities) for there to be a significant advantage of having any air routes at all. I'm not sure how friendly this is to people wanting to play smaller maps...

In a way this is all realistic of course (and I agree that having to build huge airports is good), but rebalancing will be essential to ensure that there is a place for air travel within simutrans-experimental.

[ӔO]

28 tile runway will be quite something to see in action.

Although, there should still be a possibility for smaller airports for commuter turboprops and regional jets. These kinds of airports would only require, approximately, 16 tile or 2000m runways.

[sdog]

With 125m per tile, any maps below 2000 tiles length would be a very tiny sandbox anyway. It is also a good idea to have very oblong maps, to allow long distances, without having too much area.

>20 tiles runways still seem quite a bit extreme. That's about 1/6 of the size of a typical map size when simutrans started.

I wonder if simutrans' approach is still effective with such large map sizes, and such a narrowly spaced grid.

-7hrs by bus. This is the slightly mysterious one for experimental. What it is, is a cheap, usually overnight, ride. It's very cheap compared to the other two, but the time consumed traveling is done while sleeping. It is quite popular, despite its long travel time.

This likely only works due to the incredible ability of japanese to sleep anywhere and for any time, when they have a chance to do so.

[dom700]

Another interesting side effect of forcing runways to be 30 tiles or so long is that planes take over half a minute of real time to taxi down this length of runway. Even time taken to take off and land (demanding full use of runway) will be 10 seconds or so at least.

Well, you should still place the runways in such a way that the aircrafts only taxi one field on the runway after landing and before take-off.
Even though I am heavily using city airports in my games, airports which typically just take 9 fields, and some industries even have their own airports because they are so extremely cheap (compared to a 100 tile railway track), I am actually looking forward to this

This likely only works due to the incredible ability of japanese to sleep anywhere and for any time, when they have a chance to do so.

Hmm, Ive been doing this on a rather regular basis. All those bus companies apparently have customers, who dont care how they get around in europe as long as its cheap. The only problem with sleep Ive had, was the police

[Carl]

FWIW, here's what a realistically-proportioned Heathrow looks like on my 118-meters-per-tile GB map. Given the similarity in scale, this is the kind of scale airports will have to be on in the new version of pak128.Britain. The runways are 3.9km -- 30 tiles -- long.

[Milko]

Oops - almost forgot to reply to this. I don't think that the runway lengths would be too long: as Carl says, long runways are realistic. In fact, the next version of Pak128.Britain-Ex will have 125m/tile, so the runways will be even longer: a 3.5km runway will be 28 tiles long. This will, as Carl points out, force people to put down realistic airport infrastructure, and make airports more resource intensive (and less suitable to be placed very near cities) than they have been in Simutrans in the past. This is a good thing, I think.

Assuming that the size (in tiles) of the airport does not vary with change in km / tiles while the length of the runway would vary with the variation of parameter, it would better define the minimum length of the track using the tiles in place that km?

If you want to test it, however, you will need not only the latest -devel makeobj, but a binary build from the latest -devel branch, as the existing release candidate is too old to be able to read files written. I can upload these for you if you'd find that helpful?

I only need yhe makeobj.

Thank's
Giuseppe

[el_slapper]

125m per tile? mmmh, if I want to simulate my favorite travel, Paris to Wroclaw(family of my wife), 1400km, it would be, errr, 1400*8=11200 tiles. that's not even a big travel. Buses do the travel in 18 hours for a 70 €uro ticket. Wizzair makes it slightly costlier, but that's only 2 hours of flight(plus one hour to get to Beauvais airport from Paris).

Of course, I guess the idea is to simulate Britain alone. Are there many flights other than London-Manchester-Edinburgh-Glasgow-Dublin in British islands?

Google maps tells me London-Glasgow is 402 miles - 643 km. That's already more than 5000 tiles. I need to repair my 8-cores computer if I want to play that..... and the map does not even link northern scotland. Isn't 125m/tile a little overkill, at least with current computers?