City based demand

[Seattle]

There will be no "submitted routes" as the idea is not to mimic what real airlines do and where they fly, the demand must be fully generated..

Okay, then you can use the B option for those tiny airports in cities with a major one.

[Kazari]

I think a reasonable "sphere of influence" might be based on the size of the airport. That would be easiest to program.

Another possible way to calculate/complicate it would be to take the traditional sphere of influence and add (1/(recommended fare-actual fare). This would mimic the effect of a lower fare on people's willingness to travel further by showing larger demand.

The problem is that Sami then has to account for that on EVERY city pair and every shared airport influence zone.

A compromise, Sami, would be to calculate all those "extended" influences fewer times a day than you do the immediate market of a fare.

[Sigma]

I don't see the ultimate point of having these small airports remain small?

Again, the entire point is not to mimic reality.  Why should Burbank be perpetually limited to being only a regional player?  Why are some people stuck on making sure that these small airports stay small?  I fail to see the point of that.  If there's a big airline using LAX as a major domestic hub, why shouldn't I be able to "sneak in", start an airline in Burbank, and work to steal at least a sizeable percentage of their business away?  Why shouldn't we be able to make Burbank the next LAX?  At least provided it has the infrastructure (within the confines of what we have modeled in the game) to do so.  If they really are, literally, "small" airports -- with short runways and limited hours, then they'll be self-limiting as a major player for that reason alone.  But, outside of that, there's no reason we should be attempting to limit their potential.

Now, as the game progresses and we finally get connecting passengers, then LAX will naturally had an advantage for international connections at least.  But there's little reason why Burbank, or any of the countless others, shouldn't be equally viable as a destination airport for that municipal region.  This already happens now -- people don't fly to Burbank or Ontario or John Wayne because they want to go to those immediate areas, they do it because it's cheaper (often MUCH cheaper) and easier then the behemoth that is LAX.  Anyone who's flown into LA knows that LAX only makes sense if you're connecting somewhere or have to go somewhere right nearby -- any of the countless surrounding airports make much more sense most of the time and are growing everyday for just this reason.   Southwest, one of the most successful airlines in the world, did precisely this at Love field just a stone's throw from DFW, even with all the restrictions that were placed upon their operating out of it.  The strategy of using smaller airports nearby these huge mega-airport is practically the only viable one for any new carriers around the globe because the world's major airports are already packed full.  

[Seattle]

You cannot compare DAL/DFW to what your saying because DAL was the first airport there and Southwest chose to stay there + its much closer to the center of Dallas.

I do see your point, but there are reasons why the airports that are big... are big. LAX is very centrally located and therefore can be looked at as Los Angeles Airport, while the other 4 airports serve specific purposes to there near by communities.

For example SNA/ONT are close to Disney Land and all those other parks.

and, I think the point is to mimic reality to a certain point... just not exactly, as it is now.
It just isnt feasible to think that BUR could ever handle as many pax/flights as LAX can. [from an airport standpoint and a geographical standpoint.].

[type45]

I think there are one more difficult case happened at China: the Pearl River Delta region, having HKG, CAN, MFM, SZX and ZUH within 100NM  HKG's long international demand is not just based on local demand, but also some Taiwan or SEA connection, also competing with CAN for demands amount South China, and the Cross-Strait demand from Taiwan with MFM; CAN has inter-relation with SZX in domestic and some international demand; SZX also has a big international demand and a number of short haul international demand; MFM has competition with HKG for Cross-Strait and SEA routes; ZUH is the smallest but also having competition on domestic with others. All of them have a long runway and long running hours. Also there are no domestic demand for HKG and MFM because they are independent administrative regions of China and have their own boarder.

I don't have the idea to solve it, but I wonder the methods above can help to handle the problem

[Kazari]

I like the kernel of Sigma's idea. I also think would be easier to program.

Let the permitted departures govern how large an operation can get.

The only problem with it is that then you have to program what the sphere of influence of an airport is. So if you operate out of Newbergh (KSWF) -- on Long Island -- how would YOU measure the difference in influence between that, KLGA, KJFK and KENR? I would say that while fares alone would -- and do get many people to go out there, that as you grow further from the "center" that your influence must wane. There's a reason people fly out of LaGuardia -- and it's not always fare.

My problem would be determining WHERE that center is. Sami's going to have to take all these communities he has mapped and somehow assign airports to them.

Should that "influence sphere" be based strictly on x NMs from the airport? Then traffic is determined by the population within x NMs (with arithmetically diminishing percentages as you go out)? Or are there other factors to consider?

[Sigma]

Perhaps the "sphere" could vary by population density.

So if you're out in the middle of South Dakota your sphere, at its limits, would be well over 100 miles in radius.    But if you're at NYC, that sphere only radiates, say, 10 miles.  And within that sphere, as you radiated further from the center, the population would be less likely to travel to your airport until it got to 0.  In a perfect world you'd consider geographical barriers, but that would be difficult.  Considering political barriers would also be nice, and would be easier than geographic ones, but still quite difficult I'd imagine.

The purpose would be to represent greater transportation options as wellas the fact that, the greater the density, the more resistance to travelling further (traffic, time to travel X distance, etc).

The problem then lies in what "density" do you use.  You can't use that of the area immediately around the airport, because an airport in a rural area near an urban one could end up with a large sphere that fully-encompasses an urban area (or in many cases, several urban areas if they were close enough and the airport rural enough to be granted a large sphere).  

As an aside, we have "Airport Size" in the game now.  While we should start with the figures that are already in there now so that "large" airports are already large even when the game starts, that figure ultimately should be variable based off the number of routes/flights flown out of an airport (or, even better, the number of pax going through).  And certain costs need to be a function of that number -- particularly landing/pax fees.  Perhaps it could even change how often or how many slots are added each year.  This would make smaller airports attractive for start-ups, particularly later in the game due to their lower costs as well as having available slots, but it would prevent an airline from jumping into a municipal airport smack downtown somewhere that happens to have decent infrastructure, and turn it into a major bustling airport without having to pay the higher fees of those in 'larger' airports.  If an airline really was filling up every slot, the airport would have to increase their fees to cover the operating and capital improvement costs necessary to provide service.

[Seattle]

and we also have to take into account wealth/affluence.

BOM might have apopulation of 20 million.... but has way less traffic than SEA with 4 million people.

I think it would way too complicated to use a sphere of influence system based on pop. density because of what Stigma said earlier.

However it makes the most sense....

[Unbornio]

Don't limit their capabilities please? LGW should have a chance to be LHR  . So let them serve the same population. It'll be very interesting. And... we won't be always seeing LHR at the top.

[Filippo]

I have an idea.

We calculate the catchman area for each airport based on it size (infrastructural)
Eg: Since LHR has a bigger infrastructure than LCY, LHR is bound to have a much bigger catchman area than LCY.

The cathman area will be depicted as a circle on a map. The circle will be colour-coded from green to red (like the load factors), where green is an area which is very likely to use the airport, and red can only be attracted to the airport (eg by low fares).

There will be 3 circles for each airport: domestic, international, intercontinental. Obviously, this is because the catchman area varies according to the flights people have to take. The intercontinental catchamn area is bound to be bigger than the domestic one.

Where the circles overlap, the airport which circle results greener in that area is bound to take the majority of the traffic. Though, airlines can compete on this with fares or other factors such as comfort etc. Thus, airlines based at different airports will be able to compete with each other for city to city based demand as smaller airports can now "steal" traffic from bigger airports as they usually have the advantage of being closer to city centers.

[JonesyUK]

How about just sharing demand based on distance between airports?
For example:
Domestic flights: demand combined for airports within 25 Miles
Short Haul: Combined for airports within 50 Miles
Long Haul: Combined for airports within 100 Miles

So LHR and LGW would share all demand, but they wouldn't share domestic traffic with STN, whilst all three would share international traffic

Distances could be varied by country, as the UK is more densly populated than the US for example.

Large airports could have a preference modifier based on their current size so LHR would be 5x, LGW 4.5x, STN 4x, etc which influences passenger choice along with price, time, seat quality, etc

I don't know much about progamming, but I imagine it'd be fairly simple to combine the existing data rather than creating a whole new model?

[Sami]

Thinking into this a bit again for change (..after a year).

The population data, growth rates, GDP data and it's history are mainly all OK and needed elements to make this work. But that alone does not give anything yet, as the model has to be more specific and go into more details to be able to create better regional variations and realistic scenarios about popular tourist or business destinations. Plan is to completely exclude the current demand system, and all real-world passenger demand data and make the system behave naturally on it's own by giving it only values of the population and travel needs - combined with available airlines and their tickets.

A big project all in all that one should not expect in a year at least, if ever. All is still just testing.

Anyways ... I figured that the easiest way to incorporate this "regional knowledge" into the system is to create a world of squares (1x1 lat-lon spacing). There are some 130 000 such squares inside the globe I believe but land only covers some 40 000 of them. So piece of cake, just check each of these and check if the square contains areas that are a) major business locations, b) popular holiday destinations (include main seasons for travel too; all year; or only part year?), c) capital city of country or other major administrative center, d)  ... something else that needs to drive the demand up from any other area.

Did you lose track already?

When we combine this additional data that needs to be manually inserted (to this help will be needed) with the city&population data  ... and that combined again to data of available airports .. etc ... and that we again would combine into a database of country relations (ie. friendly or unfriendly countries) and such other data ... then we could calculate each square separately, and determine the travel needs for people from that square to somewhere in the world thus making the demand values.

User would basically look still for demands from airport to airport; but the system will then just calculate the catchment area of that airport as the base population and check their travel needs from all squares that are within the reach and combine the data into the demand values. However this sounds easier since there are for example country borders that in some cases can be crossed and in other cases cannot, there are mountains, there are seas etc which may still all be inside the same square but essentially impassable... So that 1x1 lat-lon square size may be too much. But reducing it to 0.5x0.5 would bring quite much more work and requires more processing. The size of the squares are 111x111 km per side in equator and getting narrower when coming closer to poles, for example 71x111km at 50N...  (oh, and also, what about short island hops that happen inside that same 100x100km square?)

Pic #1: Here's part of USA and of it's ~5500 squares just for fun.

Pic #2: Squares combined with population data pointers (each dot is a city unit with population number and yearly pop. growth rate number).

Feel free to brainstorm.

[attachment expired]

[Sami]

About airport catchment areas  ... it is usually defined as travel time to the airport, instead of distance. Normal max. catchment area limit is 2 hour travel time.... But with the available data I believe such system is hard to make up, but would be nice.

Here's some catchment areas purely based on 100km range ring. As you can see Helsinki ring would go all the way through the sea to Tallinn even, and that's a trip people don't take unless they fly far away.

[attachment expired]

[JumboShrimp]

Wow, excellent thread.

Now I understand what you meant by regional demand.  It can be quite complex.  It introduces a concept of competition between airports.  Player based in some airport can steal local demand from another player in neighboring airport.  It is very interesting, but as I said, quite complex.

Deciding between:
1. regional demand
2. connecting traffic
from the point of view of game playability, I think connecting traffic adds more value.  But, unfortunately something on the regional demand has to be done to get realistic demand figures, in order to do the connecting traffic.

I envisioned only a half step on regional demand that is far easier to implement than a full fledged regional demand.  It could be done by fixing local demand to an airport with more realistic figures and concentrating more heavily on connecting traffic.  This way,  LGA, JFK and EWR would all be assigned some local demand, and this local demand would stay within the same ratio for the length of the game.

What does a full fledged demand buy you?
- more perfect assignment of demand to the airport
- competition between airports
- potential of airport to grow in influence over surrounding squares
- saving a half step in implementation, since 1/2 + 1/2 tends to be > 1

Again, for purpose of game playability, these are much smaller than connecting traffic, ability to grow the slots of your airport as you turn it into a hub...  For the purpose of running hubs, connecting traffic is more important than perfection of determining local demand.

I have a lot of ideas on the regional demand as well, and I will post them in follow-up posts.

[JumboShrimp]

Comment on demand in general:

Suppose we jump forward and assume we have figured out a local population demand for flying to places.  Suppose the demand (aggregate demand for flying) boils down to a single numerical figure for the airport (or a square).  How should the demand for a route between 2 airports (squares) be determined?

I think it should be a combination of 2 things:
1 - simple population demand, people visiting families, minor business travel.  If we have a numerical value for aggregate demand for airport A (or square A) and numerical value for aggregate demand for airport (or square) B then good model for A-B demand is a simple multiplication of those numerical values.  Suppose you have A-B pairs with following values:

<pre>
  A  |  B  | Demand |
   0 | 200 |     0  |
   1 | 199 |   199  |
  10 | 190 |  1900  |
  50 | 150 |  7500  |
 100 | 100 | 10000  |
</pre>

This way, demand for a route from a low population place anywhere (and back) would always be tiny.

2. Demand to an attraction.  This would be the second component of demand for A-B route.
Suppose A has an attraction with some rank of attractiveness.  The demand for this 2nd component would be the magnitude of the attraction at A * numerical value of B.

This way, demand from B (a major population center) to A (tiny island in Caribbean with almost no population but beutiful beaches) would be a function of beuty of the beaches at A multiplied by population representation of B.

[JumboShrimp]

I just had an idea how to figure out the vacation / convention / business center demand.  A good proxy for all of these combined would be hotel capacity in the area.  And if that is hard to get, just number of hotels would probably do.

[Sami]

Faster just click all that data in manually (by clicking the desired spot at googlemaps and then entering the data into appearing popup, and pressing save) than try to incorporate that data from various sources I'd say.

[Kazari]

A couple of things.

1. I think the squares are too large. I think your idea about slicing them in half is a good one.

2. I think it would be more realistic if the catchment area of any given airport were directly proportional to the marketing spent by an airline in that airport. OR by the number of flights departing from that airport. In your example Somero residents would have an equal affinity for Turku, Helsinki or Nokia (?); in the system I'm suggesting the airport with the most spent out of it would draw them out more.

3. I understand JumboShrimp's diagram and I even get the idea of using Google to post features on a per-square basis. But I'm bothered by the randomness of grid-to-grid travel. If I understand this correctly, the odds of a person in Tallahassee going to the grid square of Macon or to a square in Malaysia are roughly equal as long as they have the same population? How does the international/domestic mix apply here, if at all?

[JumboShrimp]

2. I think it would be more realistic if the catchment area of any given airport were directly proportional to the marketing spent by an airline in that airport. OR by the number of flights departing from that airport. In your example Somero residents would have an equal affinity for Turku, Helsinki or Nokia (?); in the system I'm suggesting the airport with the most spent out of it would draw them out more.

Interesting idea - catchmant area for player would depending on marketing of that player.

I like the idea of the catchment area being variable.  The greater percentge of the travel demand being fullfilled by the airport (player) should in turn grow regional influence of the airport (or player)

3. I understand JumboShrimp's diagram and I even get the idea of using Google to post features on a per-square basis. But I'm bothered by the randomness of grid-to-grid travel. If I understand this correctly, the odds of a person in Tallahassee going to the grid square of Macon or to a square in Malaysia are roughly equal as long as they have the same population? How does the international/domestic mix apply here, if at all?

The way it should work is that you start with demand of airports (or squares).  Multiply a pair of them to get the demand for the route.  Then, apply a distance multiplier.  Let's say the sweet spot is 500 miles, zero demand is at 50 miles.  So the demand would start at 0 at 50 miles, go up to 100% at 500 miles, and then start a long descent.

I thought of a solution of the islands.  Instead of going down from 500 miles, go up instead.  So each island airport would have an "IsIsland" flag, and a different multiplier would be applied.

[attachment expired]

[forsberc]

1. I think the squares are too large. I think your idea about slicing them in half is a good one.

I think would agree that the squares are too large for Metropolitan Areas. However, for most of the world they are fine. For example, Siberia in Russia will not have any demand for flights. I think the squares should be a factor of population density. The larger the population density, the smaller squares are used. The smaller the population density, the larger squares are used. This will probably allow much easier coding and time management should this be implemented.

2. I think it would be more realistic if the catchment area of any given airport were directly proportional to the marketing spent by an airline in that airport. OR by the number of flights departing from that airport. In your example Somero residents would have an equal affinity for Turku, Helsinki or Nokia (?); in the system I'm suggesting the airport with the most spent out of it would draw them out more.

I agree with this statement. However, it should be a total of all Airlines operating out of the airport. Meaning, it should incorporate route marketing. If an airline is not based at an airport, but flies there and thus has route marketing on that particular route, that marketing money spent should be taken into account.

I believe the Airport's Image (to define it) could be based on:

1) marketing,
2) company images of all airlines that operate there and
3) average pricing.

The airport's image could then become the "multiplier" for demand like which was discussed earlier. This multiplier could then increase (or decrease) the sphere of influence of the airport itself.

Here's some catchment areas purely based on 100km range ring. As you can see Helsinki ring would go all the way through the sea to Tallinn even, and that's a trip people don't take unless they fly far away.

Finally, in order to help solve the connection problems and passengers traveling between countries (or islands) to get to the airport, you can do the either of the following:

A) Have border crossings reduce demand by 50%. This is because of customs and the lack of will of the traveler to change countries unnecessarily. I feel even this 50% reduction could still generate (and steal) more passengers if the airport in question had a multiplier of 150%-175%.

-or-

B) Make the airports in question a ratio of airport images divided by 2. Say Airport A is 80 and Airport B is 10 (using the current scale of -100 to +100). The math being 8/2= 4. Thus Airport A's multiplier is 4 times as large as Airport B and thus has a much larger draw.
If traveling to the airport means the traveler must cross a border, then the ratio could be divided by 2 or 3. Thus, assuming that Airport B is in a different country than A, Airport A's multiplier is only 2 or 4/3 as large as Airport B.

I hope that this all makes some sense. I feel this question/ problem is definitely on the right track to being solved. 

Cameron