[-] Connecting passengers - an idea on how to do it

[fark24]

I've heard that adding connecting passengers is a feature in development so I don't know how far along the concept is (i.e., whether it is still in the conceptual phases or whether there is already a plan of how to do so). However, I had an idea that I thought I'd share just in case it has not yet been worked out.

For simplicity, how about a 1-stop same-aircraft connection system? For instance, let's say you are an airline based at ORD and you have an aircraft that flies ORD-LAX-ORD and then it flies ORD-JFK-ORD. Currently, when the aircraft departs from LAX, it would only be carrying passengers who want to go to ORD. However, there is a pool of passengers at LAX who would also like to go to JFK but they will not be aboard the flight.

My solution would be for the program to perform a daily calculation for each airport. Have it look for any available 1-stop connections via the same scheduled airplane. Should there be any 1-stops available, assign those flights the "tech stop penalty". The remaining seats would be filled with however many connecting pasengers that airline won the right to fly (if any).

For example, let's say that demand for LAX to JFK is 500 passengers every day, 400 non-stop seats available & 600 1-stop seats available. But the 1-stop airlines fill 400 of their 600 available seats with passengers heading to their primary destination - leaving 200 seats available. The program then calculates how to distribute those 500 passengers across the 600 available seats.

The 1-stop airlines would get the technical stop penalty applied to them. They would then compete amongst each other and amongst the non-stop carriers based on the usual factors such as flight length (which, of course, is likely longer and further penalizes them), aircraft type, seat type, company image, etc.

Let's say there are 3 airlines that fly 1-stop through their hub from LAX to JFK - one at ORD, one at SEA, and one at IAD. The airline based at ORD would have a geographical advantage as the flight time would potentially be shorter - possibly giving them a larger share of the connecting passengers. The airline in SEA would be at a geographical disadvantage but they could make up for it in part by flying a larger aircraft, having a higher company image, etc.

This would also encourage more realistic route planning. I know many of my routes will tend to be back & forth in the same direction instead of following a north-south or east-west flow. For instance, an airline based in DFW might have an aircraft scheduled to fly DFW-PHX-DFW and then DFW-DEN-DFW. Instead, the connecting passengers feature would be more likely to encourage a routing such as DFW-PHX-DFW followed by DFW-ATL-DFW because there is a better chance the airline would be competitive for passengers going from PHX to ATL. PHX-DFW-DEN is about 778nm off of the great circle distance between PHX-DEN but PHX-DFW-ATL is only about 11nm off from the great circle distance between PHX-ATL.

Another beauty of this system is that it would make for viable connection hub airports. You could, for instance, build major operations in STL, DUS, PTY, etc. by scheduling efficient North-South and/or East-West routings and competing for connecting passengers.


Any thoughts on this kind of a system? I know it's not completely realistic since real-life passengers do connect on different aircraft and sometimes need more than one connection. But I think this would be a good start.

[Sanabas]

This would also encourage more realistic route planning. I know many of my routes will tend to be back & forth in the same direction instead of following a north-south or east-west flow. For instance, an airline based in DFW might have an aircraft scheduled to fly DFW-PHX-DFW and then DFW-DEN-DFW. Instead, the connecting passengers feature would be more likely to encourage a routing such as DFW-PHX-DFW followed by DFW-ATL-DFW because there is a better chance the airline would be competitive for passengers going from PHX to ATL. PHX-DFW-DEN is about 778nm off of the great circle distance between PHX-DEN but PHX-DFW-ATL is only about 11nm off from the great circle distance between PHX-ATL.

Is that actually more realistic planning? Doesn't seem like it to me, but I'm just guessing. If a direct flight isn't available, how often to you go via somewhere else, but get to sit on the same plane at the middle airport? How often would you have to get off the plane, hang around for anything up to a few hours, before getting on for the next leg?

As for the idea itself, I wouldn't want to see that sort of very limited implementation of such a big feature. It's the sort of thing that should get modelled in more depth or not at all, I think.

[exchlbg]

I think using seats on one airplane for different tasks than just A-B traffic is the main problem here.
You have to create six legs out of four with different prizes, so flight creation pages have to be redesigned. And since you want to know how your plane is doing on every single leg and in on combined legs, aircraft and flight overviews also have to be redesigned, not to mention Sami´s side of again new code tasking of PAX distribution.
So implementation wouldn´t be much easier than for general connecting passengers system.
As much as I´m waiting for that connection thing, I´m aware I will have to wait a little longer...

Greetings
Christian

[fark24]

Is that actually more realistic planning? Doesn't seem like it to me, but I'm just guessing.

I'm thinking of the Southwest Airlines routing here. While they are technically a point-to-point airline and not so much a hub-to-hub airline, they do offer a lot of 1-stop or multiple stop flights on the same aircraft. The flights often even keep the same flight number. I've been on flights were we just sit on the airplane as some passengers depart and some new passengers board.

It has worked for Southwest as they (1) try to keep their ground times short and (2) have a high brand loyalty (think high CI) and (3) generally do not require a lot of "back-tracking".

A percentage of people do fly their 1-stop flights vs. non-stop flights offered by other airlines in real life. They can compete on price and CI so long as the difference in travel time is within reason.

Also with respect to realism, look at the real-life airlines that try to use geography to create hubs. Copa does it with PTY and Finnair does it with HEL. However, in airwaysim, it is difficult to get a viable "hub of the Americas" going like Copa has or a Europe-Asia gateway like Finnair just based on the local demand.

As for the idea itself, I wouldn't want to see that sort of very limited implementation of such a big feature. It's the sort of thing that should get modelled in more depth or not at all, I think.

I do agree that same-plane routing (outside of Southwest) is generally limited as most airlines transfer passengers between planes, maintain scheduling banks, etc. But that would be incredibly hard to compute.

I guess I'm more of the school where I don't think seeking perfection should stand in the way of an incremental improvement.

[LemonButt]

None of this is possible without a revenue management system in place.  The only routes where connections are really possible are either really small airports with lots of demand or really small airports with little demand.  For example, if an airline is flying LAX-JFK, the demand will likely be satisfied by non-stop flights.  If you are trying to compete with a layover in STL for example, you'd have LAX-STL-JFK.  The problem is you can't compete--the demand is completely satisfied by non-stop routes.  Thus a revenue management model has to be used where airlines are charging a premium for nonstop flights on routes with heavy 1 stop competition.  That means if it cost $1000 to fly LAX-JFK direct with no competition, the price for a connection in STL or FAR or DTW would be $1000 (to be profitable), but those flying direct could raise their prices for the convenience of flying direct

For example, I flew from Rome to Charlotte in May.  A direct flight was $2000/each whereas I was able to fly from Rome to Dublin to JFK to Charlotte (2 stops) for only $750 on Aer Lingus/jetBlue.  Aer Lingus/jetBlue still turned a profit with me buying a ticket, but had to charge much less to offset the "convenience" of flying direct to get my business.

The actual code for connecting pax should be fairly simple being that it is largely based on airport connectivity--as an airport's destinations grow, so does the potential to connect.  An airport with 200 destinations served versus 20 is going to have exponentially more traffic with connecting passengers.  Instead of trillions of calculations calculating every possible route for a passenger, it will be a matter of using an arbitrary multiplier for all routes based on airport "connectivity".

Any real time calculations on effective routing are going to be problematic.  Logically, a layover would be in an airport that gets you closer to your destination versus further away, but in the real world this doesn't happen.  For example, to fly to Las Vegas from Asheville, you have to fly 120 miles east to Charlotte (away from your destination) before flying 2000 miles west towards your destination.  As a result, you can't really come up with arbitrary sectors to search for hubs and establish connections.

[swiftus27]

Isn't this completely what city based planning is all about?

[RibeiroR]

Well, we have a lot of connect flights requests...
The most requested feature of all times

[lowjhg]

Bumping up a old thread because i can't seem to create a new one.

So basically I currently have something that might be able to benefit those who have a large amount of bases.

For all my examples, i shall use my NZ airline in GW2 for example, where i have level 4/4 categorisation in both AKL and CHC.

Currently, the system only allows for any base with the requisite infrastructure,
( A ) : Flying long haul from either CHC or AKL to LAX direct without tech stop
( B ) : Flying long haul from AKL with tech stop in CHC to LAX but without the ability to pick up/offload pax in CHC.

So in the interest of case B, why not allow the picking up and offloading of pax on tech stops routed through? Surely it presents no adverse implications as the base requirements are met for the aircraft type as well. If the x-CHC pax demand can be complemented by x-AKL pax demand, it would make flying an existing route more economically viable than just basing frames in CHC and fly x-CHC LH ops without being complemented by x-AKL pax.