Re: [sumo-user] Air traffic
If you use the 'real' runway/taxiway layout eg as obtained from OSM then you have several 'fast' exits and other more right angle turns from a given runway. With a minimal route for the aircraft - a "from" runway, stop at stand, then a "to" runway, I find you need a couple of specific "via" edges to get realistic routing. However with this and an 'aircraft' landing at ~110kt sumo is quite happy to slam on the brakes to meet an exit closer to the target terminal - I found you needed to set taxiway 'priorities' to ensure the aircraft didn't do this and didn't zigzag across what is really a one way system. Having set those for one landing/takeoff direction I found they just didn't work when you changed direction. Doubtless much fiddling might get it to work but I didn't think it worth the effort, given that in reality different traffic rules come into play when the runway reverses. ie a different net file seemed fine. (all you're doing is amending priorities and closing/opening a couple of connections) Obviously if you use traci to act as 'ground control' and 'tower' then you don't need different files. Cherrs Div Original Message On 13 May 2024, 18:24, wrote: > Dear Div, > > do you have an idea why this is not possible? I have always thought that > reversing the direction of a runway is just a matter of different routing of > my airplanes. I have never tried that in practice, though. > > Jan > > On Mon, 13 May 2024, at 12:00 PM, The div via sumo-user wrote: > >> One more caveat wrt change of runway direction: >> Heathrow for example uses one runway for landing, one for takeoff and >> switches these round in the middle of the day. This can be handled easily by >> using different routes. However I found that change of direction from eg 27R >> to 09L, because the wind direction has changed, could only be modelled by >> use of different network files. This is because different exits/entries/ >> taxiways come into play. Whilst being very specific about routes could >> handle this it offers no real 'modelling' >> >> Original Message >> On 13 May 2024, 10:39, The div via sumo-user < sumo-user@eclipse.org> wrote: >> Hi, I have question about the new vClasses "aircraft". >>> Are there plans to add air traffic? Or what it is generally good for? >>> >>> I've used sumo to model airports and some drone activities (just out of >>> interest) - my observations: >>> a) Modelling airport ground movements and multi-modal traffic (passenger, >>> bus, rail, cargo, catering, etc) works well today, with the caveat that >>> aircraft tug/pushback behaviour isn't really practical - you can do it at >>> airports with small numbers of stands but at larger airports it's just too >>> messy. >>> >>> b) if you're not using traci to behave as ground control/tower control >>> modelling spaced takeoff/landing needs some fudging and won't be correct. >>> >>> c)it works because in larger airfields ground movements of all traffic, >>> including aircraft, follow strict routes, even across the apron - so you >>> can set up roads/railways mapping these routes. Modelling smaller fields >>> with free movement across the apron won't work. >>> >>> d) passenger loading/unloading times on transit buses/tube lines/aircraft >>> has to be set to unrealistic values because load/unload is at a single >>> point. I haven't yet tried the doors features of v20 that may fix this. >>> >>> e) I looked at airway modelling and decided it wasn't really sensible/could >>> not envisage a use case that would model anything that excel wouldn't: >>> i) distances are huge relative to road distances - meaning you have a long >>> duration simulation for a single vehicle travelling hundreds of miles >>> ii) airway transit is about changing flight level - you need to handle this >>> to get climb/descent timings right >>> iii) using the gui, even just to check behaviour, simply doesn't work on >>> this scale. >>> >>> f) if you assume that eventually drones will be subject to similar airway >>> restrictions as aircraft to avoid collisions then it should be feasible to >>> model them using sumo vehicles - because the range of 'flight levels' >>> available and distances travelled will be relatively small. >>> >>> g) if you assume that drone traffic will be uncontrolled and reliant on >>> collision avoidance, then using sumo vehicles won't be practical because >>> drones will try to travel in straight lines between launch and destination. >>> (ie there won't be a road network) >>> >>> h) I have had reasonable success modelling this latter style of drone >>> behaviour using a sumo poi to model a drone and traci to manage flight. >>> >>> Cheers >>> Div >>> >>> ___ >>> sumo-user mailing list >>> sumo-user@eclipse.org >>> To unsubscribe from this list, visit >>>
Re: [sumo-user] Air traffic
Dear Div, do you have an idea why this is not possible? I have always thought that reversing the direction of a runway is just a matter of different routing of my airplanes. I have never tried that in practice, though. Jan On Mon, 13 May 2024, at 12:00 PM, The div via sumo-user wrote: > One more caveat wrt change of runway direction: > Heathrow for example uses one runway for landing, one for takeoff and > switches these round in the middle of the day. This can be handled easily by > using different routes. However I found that change of direction from eg 27R > to 09L, because the wind direction has changed, could only be modelled by use > of different network files. This is because different exits/entries/ taxiways > come into play. Whilst being very specific about routes could handle this it > offers no real 'modelling' > > > > > > > Original Message > On 13 May 2024, 10:39, The div via sumo-user < sumo-user@eclipse.org> wrote: >> >> > Hi, I have question about the new vClasses "aircraft". >> Are there plans to add air traffic? Or what it is generally good for? >> >> I've used sumo to model airports and some drone activities (just out of >> interest) - my observations: >> a) Modelling airport ground movements and multi-modal traffic (passenger, >> bus, rail, cargo, catering, etc) works well today, with the caveat that >> aircraft tug/pushback behaviour isn't really practical - you can do it at >> airports with small numbers of stands but at larger airports it's just too >> messy. >> >> b) if you're not using traci to behave as ground control/tower control >> modelling spaced takeoff/landing needs some fudging and won't be correct. >> >> c)it works because in larger airfields ground movements of all traffic, >> including aircraft, follow strict routes, even across the apron - so you can >> set up roads/railways mapping these routes. Modelling smaller fields with >> free movement across the apron won't work. >> >> d) passenger loading/unloading times on transit buses/tube lines/aircraft >> has to be set to unrealistic values because load/unload is at a single >> point. I haven't yet tried the doors features of v20 that may fix this. >> >> e) I looked at airway modelling and decided it wasn't really sensible/could >> not envisage a use case that would model anything that excel wouldn't: >> i) distances are huge relative to road distances - meaning you have a long >> duration simulation for a single vehicle travelling hundreds of miles >> ii) airway transit is about changing flight level - you need to handle this >> to get climb/descent timings right >> iii) using the gui, even just to check behaviour, simply doesn't work on >> this scale. >> >> f) if you assume that eventually drones will be subject to similar airway >> restrictions as aircraft to avoid collisions then it should be feasible to >> model them using sumo vehicles - because the range of 'flight levels' >> available and distances travelled will be relatively small. >> >> g) if you assume that drone traffic will be uncontrolled and reliant on >> collision avoidance, then using sumo vehicles won't be practical because >> drones will try to travel in straight lines between launch and destination. >> (ie there won't be a road network) >> >> h) I have had reasonable success modelling this latter style of drone >> behaviour using a sumo poi to model a drone and traci to manage flight. >> >> Cheers >> Div >> >> >> >> >> ___ >> sumo-user mailing list >> sumo-user@eclipse.org >> To unsubscribe from this list, visit >> https://www.eclipse.org/mailman/listinfo/sumo-user >> ___ sumo-user mailing list sumo-user@eclipse.org To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
Re: [sumo-user] Air traffic
One more caveat wrt change of runway direction: Heathrow for example uses one runway for landing, one for takeoff and switches these round in the middle of the day. This can be handled easily by using different routes. However I found that change of direction from eg 27R to 09L, because the wind direction has changed, could only be modelled by use of different network files. This is because different exits/entries/ taxiways come into play. Whilst being very specific about routes could handle this it offers no real 'modelling' Original Message On 13 May 2024, 10:39, The div via sumo-user wrote: >> Hi, I have question about the new vClasses "aircraft". > Are there plans to add air traffic? Or what it is generally good for? > > I've used sumo to model airports and some drone activities (just out of > interest) - my observations: > a) Modelling airport ground movements and multi-modal traffic (passenger, > bus, rail, cargo, catering, etc) works well today, with the caveat that > aircraft tug/pushback behaviour isn't really practical - you can do it at > airports with small numbers of stands but at larger airports it's just too > messy. > > b) if you're not using traci to behave as ground control/tower control > modelling spaced takeoff/landing needs some fudging and won't be correct. > > c)it works because in larger airfields ground movements of all traffic, > including aircraft, follow strict routes, even across the apron - so you can > set up roads/railways mapping these routes. Modelling smaller fields with > free movement across the apron won't work. > > d) passenger loading/unloading times on transit buses/tube lines/aircraft has > to be set to unrealistic values because load/unload is at a single point. I > haven't yet tried the doors features of v20 that may fix this. > > e) I looked at airway modelling and decided it wasn't really sensible/could > not envisage a use case that would model anything that excel wouldn't: > i) distances are huge relative to road distances - meaning you have a long > duration simulation for a single vehicle travelling hundreds of miles > ii) airway transit is about changing flight level - you need to handle this > to get climb/descent timings right > iii) using the gui, even just to check behaviour, simply doesn't work on this > scale. > > f) if you assume that eventually drones will be subject to similar airway > restrictions as aircraft to avoid collisions then it should be feasible to > model them using sumo vehicles - because the range of 'flight levels' > available and distances travelled will be relatively small. > > g) if you assume that drone traffic will be uncontrolled and reliant on > collision avoidance, then using sumo vehicles won't be practical because > drones will try to travel in straight lines between launch and destination. > (ie there won't be a road network) > > h) I have had reasonable success modelling this latter style of drone > behaviour using a sumo poi to model a drone and traci to manage flight. > > Cheers > Div___ sumo-user mailing list sumo-user@eclipse.org To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
Re: [sumo-user] Air traffic
> Hi, I have question about the new vClasses "aircraft". Are there plans to add air traffic? Or what it is generally good for? I've used sumo to model airports and some drone activities (just out of interest) - my observations: a) Modelling airport ground movements and multi-modal traffic (passenger, bus, rail, cargo, catering, etc) works well today, with the caveat that aircraft tug/pushback behaviour isn't really practical - you can do it at airports with small numbers of stands but at larger airports it's just too messy. b) if you're not using traci to behave as ground control/tower control modelling spaced takeoff/landing needs some fudging and won't be correct. c)it works because in larger airfields ground movements of all traffic, including aircraft, follow strict routes, even across the apron - so you can set up roads/railways mapping these routes. Modelling smaller fields with free movement across the apron won't work. d) passenger loading/unloading times on transit buses/tube lines/aircraft has to be set to unrealistic values because load/unload is at a single point. I haven't yet tried the doors features of v20 that may fix this. e) I looked at airway modelling and decided it wasn't really sensible/could not envisage a use case that would model anything that excel wouldn't: i) distances are huge relative to road distances - meaning you have a long duration simulation for a single vehicle travelling hundreds of miles ii) airway transit is about changing flight level - you need to handle this to get climb/descent timings right iii) using the gui, even just to check behaviour, simply doesn't work on this scale. f) if you assume that eventually drones will be subject to similar airway restrictions as aircraft to avoid collisions then it should be feasible to model them using sumo vehicles - because the range of 'flight levels' available and distances travelled will be relatively small. g) if you assume that drone traffic will be uncontrolled and reliant on collision avoidance, then using sumo vehicles won't be practical because drones will try to travel in straight lines between launch and destination. (ie there won't be a road network) h) I have had reasonable success modelling this latter style of drone behaviour using a sumo poi to model a drone and traci to manage flight. Cheers Div___ sumo-user mailing list sumo-user@eclipse.org To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
Re: [sumo-user] Air traffic
Dear Petr, this originated some time ago as an idea for modelling mixed airport traffic, which emerged during my discussions with my colleagues from the Department of Air Transport at CTU FTS (airport traffic management seems to be quite an issue for many airports). I posted a question to this list at around mid-November 2022 and on 2022-11-23 a contributor (The div ) posted an update that was able to simulate movement of airplanes on the tarmac. AFAIK there are students working with that subsystem at CTU FTS, I can probably provide contact if you are interested. What I can imagine is that you may update SUMO to use a separate layer of edges that describe e.g. drone airways over a city, but you would have to extend the current object-following behaviour to e.g. allow for vertical separation etc.. I do not think that it would be reasonable to use SUMO for country-wide simulation of air traffic. Jan On Mon, 13 May 2024, at 9:30 AM, Petr Chytil via sumo-user wrote: > Hi, I have question about the new vClasses "aircraft". > Are there plans to add air traffic? Or what it is generally good for? Are > there some other attributes like CarFollowing models? > > Thank Petr > ___ > sumo-user mailing list > sumo-user@eclipse.org > To unsubscribe from this list, visit > https://www.eclipse.org/mailman/listinfo/sumo-user > ___ sumo-user mailing list sumo-user@eclipse.org To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
Re: [sumo-user] Air traffic
We noticed that people are using the existing capabilities of SUMO for airport and air traffic simulation and wanted to support them in this. There are no special car following models for aircraft yet but there might be completely different movement models in the future (similar to how pedestrians have their own movement model). Am Mo., 13. Mai 2024 um 09:31 Uhr schrieb Petr Chytil via sumo-user < sumo-user@eclipse.org>: > Hi, I have question about the new vClasses "aircraft". > Are there plans to add air traffic? Or what it is generally good for? Are > there some other attributes like CarFollowing models? > > Thank Petr > ___ > sumo-user mailing list > sumo-user@eclipse.org > To unsubscribe from this list, visit > https://www.eclipse.org/mailman/listinfo/sumo-user > ___ sumo-user mailing list sumo-user@eclipse.org To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
