Demand Model

[POTKC]

As suggested in the Airport Database thread. I'd personally suggest demand should always stay the same, but slot prices should vary depending on time of day. Consequentially, arrival/departure slot prices should be charged per flight, not leased for some massive sum permanently.

[USA]

Slots are only used at select airports where governments restrict the number of flights. Most all other airports operate solely on gate space availability.

[Aussie]

...I'd personally suggest demand should always stay the same...

Just on this point: wouldn't that cause issues with having to constantly update the stats? A dynamic demand system (although hopefully not as crazy as CA's) would allow for more flexibility, and would mean that players can take alternate paths, rather than sticking to the European/Asian/American hubs where the current real world traffic is. Although starting with real figures is probably a better idea than just randoms.

That's just my opinion tho, I've only really discovered that this exists in the past hour so I'm definatley no expert on how it's going to run.

[chaitanya]

...

...

Hey Aussie! You are here!

[POTKC]

...I'd personally suggest demand should always stay the same...

Just on this point: wouldn't that cause issues with having to constantly update the stats? A dynamic demand system (although hopefully not as crazy as CA's) would allow for more flexibility, and would mean that players can take alternate paths, rather than sticking to the European/Asian/American hubs where the current real world traffic is. Although starting with real figures is probably a better idea than just randoms.

That's just my opinion tho, I've only really discovered that this exists in the past hour so I'm definatley no expert on how it's going to run.

Welcome to the forum, Aussie! Long time no see! By "always stays the same" I meant doesn't fluctuate with time. Also, USA, by slots I meant whatever you call the thing CA did. Landing rights? Something like that.

[Frenchie]

The question is, do we want to encourage diversity in our worlds (having airlines in Africa and South America), or do we want to strive for realism and emphasize as mentioned above, the three continents where the market has sprawled, namingly Asia, Europe and North America.

If we opt for diversity (quite interesting when it comes to roleplay), then the obvious answer would be a 'toned down' version of CA's model for demand (not as fluctuating, but still quite randomized).

However, if we opt for realism (which I think could be an interesting turn in our community's history as it would therefore make more sense to start a new airline platform rather than just overhaul CA's model), then we would need a new system. Might I suggest having a threshold system? By this I mean we would have approximate figures of demand for the continent and regions (to simplify the work rather than implementing demand in every airport) and then have an upper and lower threshold in which these figures could fluctuate, say by 15% max. This would ensure a degree of variability (as seen in the real world) whilst still having a general realistic figure and demand based off real world facts.

[Unknown98]

However, if we opt for realism (which I think could be an interesting turn in our community's history as it would therefore make more sense to start a new airline platform rather than just overhaul CA's model), then we would need a new system. Might I suggest having a threshold system? By this I mean we would have approximate figures of demand for the continent and regions (to simplify the work rather than implementing demand in every airport) and then have an upper and lower threshold in which these figures could fluctuate, say by 15% max. This would ensure a degree of variability (as seen in the real world) whilst still having a general realistic figure and demand based off real world facts.

One problem I see with that is it still wouldn't address the issue of flying A380s to places in the middle of nowhere. Plus we already have demand information for over a thousand airports, there can't be much farther to go in collecting that data.


Here's my plan for creating a demand model. It may not be perfect and there's probably going to be lots of different ideas as to how we can improve it, but I think we should start with this and see where it takes us. Then we can go from there.

1st: We come up with a formula that calculates % of daily demand based on time of day. For example:

00:00-02:00: 4%
02:00-04:00: 3%
04:00-06:00: 6%
06:00-:08:00: 12%

...And so on, all the way to 24:00 so that the percentages equal up to 100%. But we need to make it in a way so that there's not a big jump between times. In the example above, there shouldn't be a 6% jump from 06:00 to 06:01. It needs to be a smooth curve, but I'm not entirely sure how to do that since calculus was a very long time ago and I don't remember much of it lol.

So how/why does that work? It estimates the percent of total daily demand during certain time intervals. It simulates the "popularity" of the airport throughout the day. It adds another element to flight scheduling by breaking down the demand for each airport into different times. If you schedule a regional flight that takes off at 1am and lands at 4am, it will be difficult to fill. It also takes competition into play, so if the market is saturated during the most popular times of the day, maybe you can create a flight later or earlier in the day where there is more demand left.

It was brought up that in some countries it may not necessarily work like this. Perhaps we can create a modified demand model for different regions or countries if that is the case.

So how is demand for a particular route actually calculated? I'm still working on this in the excel model but here's the idea:

[(demand for origin + demand for destination) / 2] * ratio

The demand would be the percent of daily demand as explained above.

The ratio is the ratio of demand between the origin and destination airports. This combats against flying to remote locations in the middle of nowhere. Even if you depart from a very popular airport, if you are flying to somewhere with no demand, you will have very little passengers. The calculation is a bit complicated but this is how it works:

If the departure airport and destination airport have the same exact demand (theoretical, would never actually happen) the ratio between the two would be 1.

If either airport had exactly half as much demand as the other (or 1.5 times as much), the ratio would be 0.5. Essentially it represents the percent of difference in demand between the two airports.

The closer the two airports are in demand, the less of a penalty you will get.

Hopefully I can upload the excel sheet soon which will allow you guys to play around with it and input different airport demands and such so you can see actual numbers.

Then there will be a lot of other modifiers like marketing, image (basically what rating was in CA), competition, and probably more. But I haven't thought much about those yet.

[POTKC]

The problem with such demand fluctuations is that they would be very interesting to deal with if an actual flight schedule could be set up, but with the realtime schedule-less base of Airline Enterprise, where all flights have to be started manually, to take advantage of/properly work around the demand fluctuations one would have to wake up at the correct time of night and/or sit in front of their computer 24/7 to start flights at the correct time.

[USA]

If I remember correctly we were going to try a hybrid system that still required people to check on their airlines (I.e. Fly a given itinerary) but so long as they did that on whatever interval we set it would almost function as scheduled flights.

[Unknown98]

So initially I was just going to create a system where you click "fly itinerary" and it would start flying at that moment. But I hadn't thought about time zone differences and such, and if we're going to work demand at different times into the demand formula, there would need to be some kind of scheduling involved. So here's what I was thinking:

For each route/itinerary you create, you select a local departure time for the first flight. Then everything else is calculated and automated. Once the entire itinerary is flown, you have to hit the "fly itinerary" button again. The plane won't start flying the itinerary until the selected departure time, though. So if you hit the button at 10pm, but the first flight you have set up on that route doesn't leave until 6am, then it won't take off until 6am.