WorldspotterPL
Topic Author
Posts: 256
Joined: Thu May 27, 2004 2:40 am

Airline Hub Benchmarking

Fri Oct 25, 2019 9:42 am

Hi guys,

I have been writing down a little bit of research on hubs in the last year. I wanted to set up a blog or something, but now I really just want to get it out there and discuss with you guys (its 2017 data and will get older and older), so here goes. Don't mistake me for a wannabe researcher or a journalist, it's all just for fun - so please be gentle with feedback. My biggest worry is that I do not have enough graphs - often I have to quote my lists and excel files. I hope to still manage to follow.


Hubs – connecting one billion passengers a year

In most cases, flying means getting on an aircraft at your home airport and getting off the aircraft and leaving the airport at your destination. This is the case in around three quarters of all flights – because if there is sufficient demand and the right aircraft equipment to offer a direct flight between two airports, there will be at least one direct service. Of the four billion global passengers that boarded airplanes in 2017, three billion took a direct flight from origin to destination. But still the massive number of one billion had to change planes. This is where hubs come into play. They enable people to go from Tirana to Hong Kong or from Little Rock to Nagoya with only one stop. Only as soon as we have invented small one-person aircraft that can economically connect Tirana with Hong Kong or Little Rock with Nagoya will hubs become obsolete. This is my answer to a question I often get asked: why do we need hubs and why passengers that do not even want to go to – say – Amsterdam have to even further congest the airport’s infrastructure. Of course there are more profound approaches to answering this question, such as: only with connecting passenger can most of the long haul flights be filled in order to make them economically viable (with the aircraft technology we have today).

This leads to the very interesting phenomenon (or: paradox) that hubs incur – connecting passengers, that easily make up 70% of most long haul flights, are needed to fill planes at a certain contribution margin. Yet, in most cases it is origin or destination passengers who pay the fares that allow an airline to make a flight an economic success. But why do passengers that only take one flight (and thus cause much lower costs) pay more than passengers that take two flights? Well, you have it right there. Because the trip from A to B without the hassle of having to change planes is worth much more to a passenger, airlines can charge a premium for direct passengers. Thus the latter cross-subsidise the in many cases lower paying but in all cases more costly connecting passengers. This is why Air France will ‘steal’ passengers on Munich to New York from Lufthansa’s direct flights by offering lower prices just as Lufthansa will do the same on Paris – New York via Munich or Frankfurt (or Zurich, Vienna, Brussels or even Dusseldorf for that matter). Of course not only the added second flight involved is driving up costs – connecting passenger also add complexity to the flight processes, e.g. handling of luggage, rebooking or even hotel costs in the event of delays and missed flights.

Yet, the interesting field on costs and yields of hubs is not central to my article. Following aviation media and forums for centuries I feel there is a clear gap in the very basic descriptive statistics about hubs. It sounds almost too trivial, but my goal is to shed light upon a few very basic questions: what hubs are there in the world? What does very basic descriptive data tell us about their differences and commonalities? Do all American hubs work the same way or do all hubs of a similar size work the same way, be they in Asia or Africa? How can we cluster them into logical groups of hubs? And what does all of this tell us about the importance of geography of hubs or the strategy of airlines and alliances?


A New Definition of Hubs

To even try to answer these questions, hubs need to be defined first. As a hobby aviation researcher (luckily with MIDT-Data at hand) I find it hard to take that even the big aviation data providers seem to define hubs as more or less large airports with connecting passengers (irrespective of airlines). Others calculate theoretical Connectivity Indices with potential connections based on flights and minimum connecting times over all airlines. The scientific literature does not solve this problem either – mostly lacking the adequate data on passenger flows and O&D city pairs. On forums like this one I feel the need to answer to every second topic with my data-set. I believe there is a clear gap in the public aviation knowledge for numbers about true hubs – not simply airports. Not hubs by some vague ‘large airport’-sytle definition, but by a very specific one, namely

‘the operation of one airline with its (alliance and other code-share) partners at one airport’.

I believe that hubs need to be defined this way since hubs don’t simply happen (except maybe at exceptional mega-cities such as London or New York). They are planned in banks of arriving and departing flights by airlines and their partners. Every aviation geek would agree that Icelandair’s operation in KEF is one of the world’s truest hubs, even though by absolute passenger numbers the airport would not be dubbed a hub by many. On the other hand, Garuda Indonesia’s Jakarta operation is located at a very large airport (often dubbed a hub), but the airline’s operation does not qualify as a hub in 2017 (per my definition) since the share of connecting passengers was at only 9%.

For my humble research, I have defined that for an airline-partners-airport combination to be a hub, it needs to have

- at least one million connecting passengers per annum

and
- at least a 10% share of connecting passengers

How did I come up with these numbers? First, the basis for every hub definition has to be connecting passengers simply because they make a hub a hub. Since this is no doctoral thesis, I took the liberty of trying around with definitions from my gut feeling of many years in the industry and the definition I chose roughly includes all the hubs I feel it should include and excludes those that I think it should. The line needs to be drawn somewhere, but not every airport can be a hub even though every airport has the odd connecting passenger. I found that only a combination of absolute and relative connecting passengers includes all the small and larger hubs that deserve to be called a hub. Defining a hub only by – say – 30% connecting passengers minimum would have penalised airline/airport operations in very large metropolitan areas – United’s Newark hub for example would have been ruled out. At the same time, defining hubs only by absolute connecting passengers – 2 million for example - would have punished small but intentional hubs such as [email protected] or [email protected] Sure there are borderline cases – it hurts to rule out the two Jakarta hubs of Garuda and Lion Air, both at only 9% connecting share but with 2.1 million and 1.8 million absolute connecting passengers. Also, Virgin Atlantic’s very interesting long-haul operation at Heathrow did not make it into my list – it only just misses the one million connecting passengers mark although its share lies at 24%. The line has to be drawn somewhere.

Image
This graph shows the world's Top 30 hubs (by my definition) by total daily connecting passengers, hubs by continents follow.

Image
Hubs in Africa

Image
Hubs in Asia

Image
Hubs in Europe

Image
Hubs in the Gulf

Image
Hubs by airlines in the big 3 alliances in North America

Image
Hubs by non-aligned airlines in North America

Image
Hubs in South America

Applying this definition to 2017 MIDT Data I have come up with a very interesting dataset that – as far as I am concerned – for the first time lists all 154 hubs there are in the world with passenger numbers (connecting passengers and O&D passengers) and connecting passenger share to start with. For my deeper analyses I have defined and calculated further KPIs such as

- population of metropolitan areas that serve as catchment for the hubs
- avg. distance flown per passenger via each hub (city pair distance via hub)
- number of city pairs flown via each hub in different size and distance categories
- number of passengers on city pairs via each hub in different size categories and distance categories
- also, I have plotted waves of arriving and departing passengers for many hubs, indicating both the continents, to which flights go, and the height and width of waves.


Prior to clustering airline hubs in order to deeper analyse and compare them, I will introduce basic numbers and statistics about global connecting passenger flows. This should later help embed single hubs into the global aviation system.



Global (Connecting) Passenger Flows


Global Passenger Flows


Before we shift our focus towards hubs and connecting passenger flows, let’s look at the biggest overall passenger flows (including connecting traffic). Notably, the top global city pairs flown are all short domestic routes. Most of them are between the two dominant metro areas of a country, such as Beijing to Shanghai, Sydney to Melbourne, Rio de Janeiro to Sao Paolo, or Delhi to Mumbai. At a distant number one, however, is an odd one: flights between South Korean Capital Seoul (Gimpo and Incheon Airports) to the country’s main holiday island Jeju. An incredible 18,000 passengers take this trip per day each way. At number two and three, Japanese routes Tokyo to Sapporo and Fukuoka respectively follow. At only half the volume of the likes of Beijing to Shanghai, London to Dublin is the first European city pair at just over 5000 passengers per day each way. At the same time, it is only the second non-domestic city pair in the list just behind Tokyo - Seoul.
As for country pairs, China to China and USA to USA clearly top the global list at 15% and 14.3% of all passenger flows. Several domestic flows follow, such as India and Indonesia (both 3%), Japan (2.7%), Brazil (2.3%), as well as Australia, Turkey, and Russia (around 1.5% all). At number 20 of the list, the largest international passenger flow between two countries between the UK and Spain at 1.2% or roughly 60,000 passengers per day in each direction.

(Connecting) Passenger Flows via Hubs

As for passenger flows with a connecting hub, let’s also look at city pair flows before we look at country pairs. The largest city-pair flow on the list with at least one stop – one could have guessed it – is the original ‘Kangaroo Route’ London to Sydney at 782 passengers per day each way. Although most one-stop flights in the world could be flown non-stop if economically viable, London-Sydney is one of the few relevant city pairs that actually require a stop for refueling. Now, as part of ‘Project Sunrise’, Qantas is looking at options to fly the route non-stop from the 2020s onwards on Boeing 787s, 777-8, or Airbus A350s. For a very long time, East Asian hubs Singapore, Bangkok, and Hong Kong have famously dominated this ‘Kangaroo’ flow. Today, at just over 200 passengers per day each way, Dubai is the top hub between the British Capital and Australia’s largest metro. Fellow gulf hubs Abu Dhabi (92) and Doha (51) are also forces to be reckoned, but Singapore (182) is still the clear runner-up. Hong Kong (81) and Bangkok (41) have fallen behind. To give you an idea of how hubs work: thick flows such as these are very rare. In my dataset there are over 85,000 city pairs with at least one passenger per day each way. Of these, not even 700 are 100 or more passengers thick. More than 110,000 have 10 or fewer passengers. And then there are thousands more flows with below one passenger per day that do not even show up in my data. Hubs cater for all these tiny traffic flows and this is what I wanted to stress when I mentioned earlier that hubs will exist until we invent little one-seater aircraft that can fly long haul missions.
Back to our list of global connecting passenger city pairs: before LON-SYD’s little sister Kangaroo-flow London – Melbourne shows up at number five on the list, three US coast to coast flows follow. Los Angeles and San Francisco to Washington D.C. are number two and three; Los Angeles to New York is number four. These American flows are notably all quite evenly split up between the at least ten hubs in the US with the likes of Denver, Chicago, Charlotte, Atlanta, or Phoenix leading most ranks. You can already see that the US is the country of hubs, a fact my analysis will clearly show.

Also in the top ten of global connecting passenger flows is New York to Israel’s main metro Tel Aviv. Aeroflot’s Moscow Sheremetyevo hub is the dominant player in connecting the city with the largest Jewish population in the world (New York at over one million) to the country with the largest Jewish population (Israel at 6.5 million). To give you a rough idea of how a flow like this one – a flow that can be and is flown directly – is distributed: roughly 1,500 total passengers fly between the two cities per day in each direction. 1,000 fly non-stop on the direct routes provided by El Al, United, and Delta. Around 500 choose to fly via a hub, mostly via Aeroflot’s hub at Moscow (117 passengers), other important hubs include [email protected] (55), [email protected] (47), and Air [email protected] CDG (46).

When we go on to look at country pair flows, the US leads by a mile (even though adding up the European countries would be a helpful measure to compare). Domestic US to US connecting traffic makes up an incredible 26% of global connecting traffic. Of these flows, Delta and American Airlines share more than a quarter each. Low-Cost Carrier Southwest is notably number three at 22% ahead of network carrier United at 17%. Back to country pairs: At a tiny 1.6% each, USA-Canada and USA-Mexico follow the mega US domestic connecting passenger market at number two and three. China to China (number four) and Brazil to Brazil (number five) have a mere 1%. Other important domestic connecting passenger markets are Indonesia, Russia, and Australia. The former is a growing country made up of hundreds of islands, thus making air travel necessary even at short distances; the other two are simply vast land masses. The thickest intercontinental connecting passenger flow between two countries is USA to India at 0.9% followed by USA to UK at 0.8%.

This is how the four billion global passengers were roughly split up between the flows they flew on in 2017. Shifting our focus towards hubs, we have seen on what city and country pairs the one billion connecting passengers traveled. But before I will go deeper into analysing single hubs and hub groups, they need to be clustered. This is what I will do in the next chapter.


Clustering Airline Hubs

The 154 global hubs of 2017 are as different as they are fascinating. Some are very small and cater for only a few very thick city pairs: Eva Air’s operation at Taipei takes more than 150 people from Ho Chi Minh City to Los Angeles every day. Yet, there are much bigger hubs such as Lufhansa’s Munich operation where the largest city pair is Rome to Hamburg at a mere 23 daily passengers. A hub with a deep network like [email protected], however, has tens of thousands of city pairs with less than one passenger per day, while Eva [email protected] only has a fraction of this number. Some hubs connect their average passengers with a combination of two flights under 1000km – American Airlines does so at its Washington Reagan (DCA) hub, and so does LATAM at Sao Paolo’s secondary Congonhas Airport (CGH). Others, such as Singapore Airlines, famously connect Europe to Australia and other long haul flows at an average of over 9000km flown per passenger. Then there are hubs at the same global location that work completely differently. Dubai – the second largest hub in the world – connects all continents with a mega-fleet of wide-bodied aircraft. Nearby Gulf Air at Bahrain Airport specialises in connecting the Gulf and Middle East to India on A320s. Then there is the hub that eclipses every other hub with its sheer numbers: Delta’s Atlanta operation that makes Hartsfield-Jackson Int’l Airport, with six parallel runways, the world’s largest in terms of passengers.

First, in order to cluster global airline hubs (by my definition) I have plotted them on two axes (see graph below). The X-Axis contains total connecting passengers per day of the airline/airport operation; the Y-Axis indicates the percentage of connecting passengers vs. Origin & Destination (O&D) passengers of the same operation. Note that throughout my analyses I count connecting passengers twice – once on the inbound flight and once on the outbound flight they connected to – in order to stay in line with the general approach to airport and airline statistics.

Image
All 154 hubs plotted, airport codes follow

The results show the majority of airline hubs scattered in a curve (‘hub-curve’) from the bottom left where there are many hubs in one place up to the top right hand corner where the density of hubs becomes very thin. On the bottom left are hubs where only a few thousand total daily connecting passengers represent merely a 10% share of an airline’s passengers at an airport. These only just classify as hubs in my definition. In the top right hand corner are hubs that combine a large number of absolute connecting passengers with a high proportion of total passengers at the airline’s operation at this airport. Additionally there are a few hubs that leave the curve – some to the top left corner and some to the bottom/right or top/left hand sides of the curve. Notably there are no hubs in the bottom right corner – for this would mean that there are hubs of several hundreds of millions of passengers. Today, Delta Airlines and its (SkyTeam-) partners’ Atlanta hub is the largest hub with around 80 million passengers, coincidently at the world’s largest airport with more than 100 million passengers in total.

Image
The hub-curve

Clustering these hubs into groups for the first time, lacking professional clustering software, I have chosen a hands-on approach simply using proximity on the plot of connecting passengers and their share at a hub. Later, I added data about City population and average flown distances via hubs to help sharpen the clusters.

Image
All 154 hubs clustered into groups


‘Regular’ hubs on the ‘hub-curve’: Small Hubs to Super Hubs

First, however, I have defined five clusters of hubs that roughly make up the ‘hub-curve’ I have mentioned above. These groups become smaller as hubs get bigger from bottom left to top right. Clusters range from the largest group of the ‘Small Hubs” in the bottom left corner to the two ‘Super Hubs’ of Emirates and Delta Airlines at Dubai and Atlanta respectively. Don't worry, I will go through the groups on by one, starting with the largest hubs.

Image


‘Super Hubs’

- 70% - 75% connecting passengers share
- over 100,000 daily connecting passengers

Members (2):

Americas (1):
- [email protected] (ATL)

Asia/Pacific (1):
- [email protected] (DXB)


Image

First: those two ‘Super Hubs’ are massive; especially Atlanta could have easily deserved a category on its own. However, a comparison is very interesting. 80,000 actual people (counted one at landing and once at take-off to get to 160k passengers) change planes at ATL on Delta every day, nearly 60,000 at Dubai on Emirates. Yet, the two hubs could not be more different in the way they work (see image below), even though they are both located in similarly large metros (between four and six million inhabitants). Delta’s Atlanta Skyteam hub has an average weighted passenger-distance of only 2,879km (note that all global hubs have an average of around 5,000km). Around 90% of connecting passengers are domestic, a further 6% Caribbean and Central American. In stark contrast, Emirates’ Dubai hub averages nearly fourfold the distance at 10,242km with its wide-body only fleet of Airbus A380s and Boeing 777-300ERs. Only 12% of connecting passengers use [email protected] for regional flying between cities of the Gulf and the Middle East while nearly two-thirds alone (65%) use it on the global North-West to South-East-Corridor between Europe, Southeast Asia, the Indian Sub-Continent, and Australia. Dubai – along with the other Gulf-based hubs and Istanbul – is perfectly located to cater for these traffic flows. As a country-pair, however, USA to India is [email protected]’s strongest (5.4%) and India is its source of the highest number of connecting passengers (9.3%) followed by the UK (8.6%). The strongest city-pairs via [email protected] are routes from Muslim mega-cities Karachi (25 million) and Jakarta (29 million) to Jeddah – gateway to the place of Muslim pilgrimage town Mecca. These are some of the thickest one-stop city pair flows there are in the world at around 150 connecting passengers per day each way.

Image
The world's two absolutely dominant hubs could not work more differently: Emirates' Dubai hub (above) and Delta's Atlanta hub (below) shown here by their waves of arriving and departing flights on a random day in 2017. Destinations are coloured in by continents (Western Europe being blue, Eastern Europe light blue, North America red, Central America pink, South America purple, Asia yellow, Australia green, Africa grey, Middle East brown). Not only is it obvious that [email protected] is very domestic and [email protected] very intercontinental (check out the waves of pax going from blue (Europe) to green/yellow (Asia/Pacific!!)), also the peaks are more than twice the height at ATL - Delta using much smaller aircraft to many more destinations. Also, [email protected] has much fewer and longer peaks with only three connecting waves per day in which (mostly long haul) passengers have longer connecting times as compared to [email protected] The graph for the Emirates hub also makes the point that the hub could not exist if it wasn't for the 24h operations at the airport.
Image


‘Very Large Hubs’

- 55% - 75% connecting passengers share
- 70,000 - 100,000 daily connecting passengers

Members (5):

Americas (2):
- [email protected] (CLT)
- [email protected] (DFW)

Europe (3):
- [email protected] (AMS)
- [email protected] (FRA)
- [email protected] (IST)


Image
The 'Very Large Hubs' by size of connecting passengers and coloured in by alliance

Moving down the ‘hub-curve’ come the five ‘Very Large Hubs’. They consist of two US-hubs – notably two American Airlines oneworld hubs, Dallas Fort-Worth and Charlotte – and three European hubs, Frankfurt (Lufthansa), Amsterdam (KLM), and Istanbul (Turkish Airlines). Similar to [email protected], both American Airlines hubs are located in the southern states of the USA and they, too, are very regional hubs. [email protected] has a share of 96% of connecting passengers to and from North and Central America as well as the Caribbean (at 3639km average weighted passenger-distance). At 99%, [email protected] (2,313km) is even less connected to the world, only a mere 1% of passengers connect to Europe.

The European hubs, on the other hand, all average around 6,000km distance per passenger, even though Turkish Airlines’ hub at Istanbul works very differently than the other two. [email protected] and [email protected] feature a similar set of passenger flows with Europe equally important at 67% of connecting passengers (AMS focuses more on Western Europe, especially the UK, while FRA has a larger share of Eastern Europe-bound passengers). North America (around 14% for both) and Far East Asia (around 6% for both) follow, while Amsterdam then has a larger focus on ex-colonies in Africa, South America, and the Caribbean. Frankfurt specializes more in connecting traffic to the Middle East and India. Notably, the domestic share of connecting passengers of Turkish’s IST hub at 11% is smaller than the UK’s share of passengers at KLM’s AMS hub (12%). Frankfurt, in comparison, has 16% domestic connecting passengers.

Turkish Airlines’ main hub at Istanbul Ataturk (IST) has a very broad connecting traffic pattern. While Europe is its most important market, at (46%) it is not as important as it is to the competition in this cluster. Its traffic flows are much more evenly split with shares of passengers to and from North America, Gulf, India Sub-Continent, Southeast Asia, and Africa all making up between 4 and 8 percent. Making efficient use of its geographic location, [email protected] uses narrow-body equipment on many more medium-haul route than its competition and thus serves more destinations in all parts of Africa, the Middle East, and Central Asia than any other hub. An analysis of the distribution of connecting passengers over different total flight distances via the hubs underlines this observation: While [email protected]’s and [email protected]’s curves peak at passengers flying intra-European 1,000-2,000km and vanish to almost nothing for city pairs between 3,000 and 6,000km, [email protected] peaks between 2,000 and 5,000km. On the long-haul end of the scale, all three hubs interestingly peak at the same category of 9,000-10,000km. AMS and IST show a nearly parallel height pattern between 8000km and 12000km while Frankfurt – with its strong focus on North America and Far East Asia – carries almost double the amount of connecting passengers between 8,000 and 10,000km.

Image

Image
[email protected] (above) has five clear hub waves while [email protected] (below) has four that are much less visible. Note that Turkey is listed as being Eastern European in my dataset, that is part of why [email protected] is dominated by light blue colours. The Turkish hub resembles [email protected] much more than it does its continental European competitors. This of course has to do with the airport being open 24h like DXB, unlike the likes of FRA or AMS.

Image
Look at [email protected]! This is the same scale as FRA and IST above! Isn't it fascinating how hubs run so differently can cater for a very similar amount (and precentage) of connecting passengers? American's largest hub (yes, more connecting passengers than Dallas!) has strong similarities to next door neighbour [email protected], albeit being even more clear cut between banks of arrivals and departures. In all my research I have not seen anything as clinical as [email protected]


‘Large Hubs’

- 50% - 60% connecting passengers share
- 25,000 - 50,000 daily connecting passengers

Members (8):

Americas (6):
- [email protected] (PHX)
- [email protected] (DTW)
- [email protected]/St. Paul (MSP)
- [email protected] (ORD)
- [email protected] (IAH)
- [email protected] (DEN)

Europe (2):
- [email protected] (FCO)
- [email protected] (MUC)


Image

To the bottom left of the ‘Very Large Hubs’ lies a group of eight ‘Large hubs’. Apart from Chicago, which is on the overlap to the ‘Large Mega City Hubs’, the group of hubs has a homogenous set of metropolitan areas in which they are located. Starting with Munich at two million inhabitants, the group averages around four million with Houston, home of United’s hub at Bush International Airport, the largest at six million. The Windy City, with a metro of 9.2 million Chicagoans almost a mega city, finds itself in this group and not in the group’s ‘Mega City Hubs’ sister group, since its profile of average connecting passenger distance at 3,650km is very similar to that of the other regular ‘Large Hubs’. The eight ‘Large Hubs’ are a very homogenous group with around 3,500km on average and all in between 3,000 and 4,000km. The ‘Large Mega City Hubs’, in comparison, average at more than double, around 7,500km distance flown per connecting passenger.

A look at their location explains the similarities between the members of the ‘Large Hubs’. Not only are the metro areas of similar size, they are also all located within the continent. I thought about naming this group ‘Large Continental hubs’. In contrast, the majority of mega cities and their hubs lie on the edge of continents at the great oceans, which in part explains their connecting traffic patterns (see ‘Mega City Hubs’). Of the six (!) US-hubs in this small group of ‘Large Hubs’, half are Mid-Western hubs ([email protected], [email protected]/St. Paul, and [email protected]). Houston lies on the coast of the Mexican Gulf but can be dubbed ‘continental’ with regard to the large oceans and continents. Phoenix is the closest to an Ocean within the group, but still has the entire 40 million-state of California between itself and the Pacific. Interestingly, over in Europe, Munich (Lufthansa) lies as cozily within the European continent as Denver (United) does in North America. Applying the same oceans and continents logic to Rome (Alitalia), the Italian capital can also be called continental. Rome’s southern central location within Europe makes it a natural gateway for (North-) Europeans towards Africa (in theory), yet the African markets are almost negligible compared to the major Asian and American markets.

The distance per passengers patterns of the ‘Large Hubs’ represent the continental nature of their locations. All eight hubs have their by far largest peak at 1,000-4,000km, with the two European hubs peaking earliest at 1,000-2,000km, notably on par with Detroit. The other five US-American hubs – due to further distances in the much larger continent – peak later, between 2,000 and 4,000km. Munich is very Europe-centric at 82% of total connecting passengers, much more than big-sister hub FRA (see above). Fittingly dubbed ‘northern-most Italian city’, MUC is even more reliant on traffic from Italy (12%) than [email protected] is famously on traffic from the UK (see above). [email protected] also has a very strong Eastern European presence with Romania and Poland at a larger share than European heavy-weights France and the UK (!). Italy’s major hub, [email protected] - coincidently a competitor within the same hub-cluster, suffers from Munich taking much of the wealthy North-Italian traffic on its North-bound journeys to the US and Asia. The FCO hub has, being centrally located within Italy, found its niche in serving traffic between Southern and Northern Italy and its islands. A (by European standards) staggering 21% of connecting traffic stays within the country, a number matched only by four hubs in Europe that we will see in later clusters.

I find these numbers so fascinating, so let me quickly take you through them on a short excursion on domestic hubs: First, at 22% domestic share, the largest hub in Europe’s largest country: Aeroflot’s Moscow (SVO) hub that today still follows its historic role in connecting the enormous Russian landmass. Aeroflot was, in 1991, the world’s largest airline and three times larger than the biggest airlines of the Western world. Secondly, Air France’s secondary hub at Paris Orly (ORY) at 22% sports the same amount of domestic to domestic flying. However, when adding in the French ‘départements et régions d’outre-mer’ (most importantly La Réunion, Martinique, Guadeloupe, and French Guiana), Air [email protected] becomes a 75% France to France hub. Then there are the two Scandinavian hubs of [email protected] (23% Norway to Norway connecting traffic) and [email protected] (21% Sweden to Sweden connecting traffic) that often serve as an alternative to motorways or trains by connecting very small airports in very challenging geographic (and often weather) conditions. More on the latter three hubs can be found in the chapter on ‘Small Hubs’.

Back to our ‘Large hubs’: On the long-haul side of the distance per passengers curves, only [email protected] and to a lesser extend [email protected] have noteworthy peaks. Munich has nearly 50% more volume but both peak at the typical 7,000-8,000km distance for Europe-US-East Coast/Mid-West traffic, while Munich also has a strong point at 9,000-10,000km where Europe-West Coast and Europe-Far East-Asia traffics lie. The rest of the cluster has small long haul peaks with Delta’s Asian focus at Detroit (see image below) being one observation at 11,000 km to 12,000 km flown.

Image

Image

Image
[email protected], [email protected], and [email protected]: three of the 'Large Hubs' and their waves. Denver is clinical, similar to Charlotte, Munich has six broader waves. Delta's Motor City hub has - compared to all red hubs like Denver or Charlotte - a few yellow spots here and there. These represent its focus on Asian destinations.



Ok guys, this is it for now. I will continue with the medium sized and the small hubs the coming week. Then I will go on to the special hubs like the Mega City hubs, the Pure Connectors and the Niche Hubs - granted you guys want to read more of my stuff on boring old hubs :)

Best regards for now!
Paul
 
User avatar
mikegigs
Posts: 232
Joined: Mon Jul 14, 2014 8:03 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 11:30 am

Wow... this is fascinating! If you really did all that work yourself (and I don't doubt that you did), then I am extremely impressed. Some of the info is somewhat obvious (DL's ATL hub is BIG and domestic-connection focused), but other information (about the "niche hubs" and "pure connectors") I never would have found out on my own. The comparison to departure banks is super cool too. I think we rarely look at connection statistics in such an in-depth way.

Thanks for sharing!
Airports: BOS, JAX, JFK, EWR, LGA, CVG, ATL, CLT, DCA, IAD, STT, PVD, ALB, MCO
Aircraft: 733, 735, 73G, 738, 752, 717, A319, A320, MD-88, E190, E175, E145, CRJ-200, CRJ-700, Q400
Airlines: B6, CO, DL, US, NW, WN, DH
...a good start but a long way to go!
 
User avatar
Midwestindy
Posts: 4299
Joined: Sun Mar 12, 2017 3:56 am

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 11:49 am

Great stuff!
Status for 2019/2020: AAdvantage Platinum, Delta Gold, Southwest A-List
 
User avatar
DL747400
Posts: 819
Joined: Mon Sep 01, 2008 4:04 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 11:57 am

I am in a big data coma.
From First to Worst: The history of Airliners.net.

All posts reflect my opinions, not those of my employer or any other company.
 
User avatar
EightyFour
Posts: 79
Joined: Sun Dec 24, 2017 10:35 am

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 12:08 pm

I don't have much to contribute expect saying, great work, great thread, and that I look forward to the next installment.
 
AlexBrewster03
Posts: 101
Joined: Fri Jul 20, 2018 3:12 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 12:09 pm

God god that’s a lot of fascinating information. Hats off to you for this wonderful post. The one thing ill say is that I’m very surprised UA does not have a single ‘Very Large’ hub
 
JAMBOJET
Posts: 151
Joined: Tue Dec 11, 2018 6:23 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 12:17 pm

wow! Thanks so much for sharing! Such an amazing post! Thank you!
 
ncflyer
Posts: 1240
Joined: Tue Sep 05, 2000 7:03 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 12:29 pm

Looks cool but is there a cliff notes version?
 
BooDog
Posts: 137
Joined: Fri Nov 09, 2007 1:44 am

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 2:58 pm

I'm going to read every word of this post after I get some coffee in me. It looks GREAT.
B1B - best looking aircraft ever.
 
nickpo
Posts: 2
Joined: Sun Feb 28, 2010 2:20 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 3:09 pm

Impressive and fascinating! Thanks a lot for sharing.
LatinAirliner
 
User avatar
chepos
Posts: 6950
Joined: Sat Dec 02, 2000 9:40 am

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 3:16 pm

Excellent work, insightful and intelligent post such as this is why this can still be a great site.


Sent from my iPhone using Tapatalk
Fly the Flag!!!!
 
LAXdude1023
Posts: 5918
Joined: Thu Sep 07, 2006 3:16 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 3:22 pm

Excellent work! Thank you for sharing.
"I dance and laugh among the rotten"
 
User avatar
DeltaMD90
Posts: 8532
Joined: Tue Apr 15, 2008 11:25 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 4:17 pm

Glad I opened this, I thought it would be some noob asking about the difference between a hub and focus city ;)

Well, I know what I'm doing this weekend...
 
User avatar
tlecam
Posts: 1462
Joined: Tue Jul 23, 2013 1:38 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 4:23 pm

Wow. So interesting! Thanks!
BOS-LGA-JFK | A:319/20/21, 332/3, 346 || B:717, 735, 737, 738, 739, 752, 753, 762, 763, 764, 787, 772, 744 || MD80, MD90
 
PAA25
Posts: 6
Joined: Wed Dec 07, 2016 8:47 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 4:38 pm

Wow, this is amazing--thanks for marshaling and visualizing the data! For the visualizations, what software are you using? (Looks like it could either be Tableau or R Shiny.)
Last edited by PAA25 on Fri Oct 25, 2019 4:50 pm, edited 1 time in total.
 
User avatar
proudavgeek
Posts: 14
Joined: Thu Aug 25, 2016 9:29 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 4:40 pm

Very detailed and amazing..I wish I can read it word by word.... Just curious,, did you use Tableau or PowerBI or something else for all the visualizations? Look cool...
 
FSDan
Posts: 2688
Joined: Mon Jan 03, 2011 5:27 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 6:15 pm

Excellent analysis! Thanks for taking the time to share!

A few observations I found interesting:
  • EWR isn't the mega connecting hub it's sometimes portrayed as. Lots of passengers, but weighted much more towards O&D when compared to hubs with similar total passenger numbers or departure counts such as MSP and DTW.
  • B6 really appears to be an O&D airline, although with these statistics being a few years old I wonder if the share of connecting traffic at FLL has grown significantly?
  • In a way, it's surprising to see that almost all of the Northeast Asian hubs such as ICN, PEK, PVG, NRT/HND, etc. stack up as relatively small hubs on the world scale as far as connecting traffic goes.
  • Good point regarding how AZ's FCO hub struggles to catch the high value North Italian traffic bound for intercontinental destinations. Just one part of why the airline has been such a basket case, I suppose.

A suggestion regarding future data that would be really interesting to see: you alluded to the fact that there's a real difference between hubs that connect a plethora of city pairs with relatively less volume on any given pair, versus those that "specialize" in a few high volume traffic flows. It would be really interesting to see a ranking of relative "generalist" hubs vs "specialist" hubs - perhaps by # of city pairs accounting for X% of total connecting volume, with a lower number indicating a more specialized hub?

Also, a question on the hub bank wave graphs... I think they are interesting to highlight the overall structure of each hub, but it isn't clear to me whether each "stack", if you will, is scaled by number of destinations, seat capacity, or something else.
This is my signature until I think of a better one.
 
User avatar
Pudelhund
Posts: 214
Joined: Tue Jan 15, 2019 11:06 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 6:37 pm

Posts like this are the main reason I keep coming back here.
 
User avatar
lightsaber
Moderator
Posts: 18429
Joined: Wed Jan 19, 2005 10:55 pm

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 11:41 pm

This data is facinating. Like all good data, it takes time to process.

My first thought was amazement on the huge fraction of direct flights. I expected a higher fraction connecting than 25%, something closer to a third of traffic.

It isn't until congested LHR where a hub is majority O&D.

I couldn't help compare the ME3:
DXB with 74% connecting
DOH with 85% connecting
AUH with 85% connecting

to compare
IST 58% connecting
ADD 78% connecting

Only IST has a strong defensible (tough to bypass) position.

Lightsaber
IM messages to mods on warnings and bans will be ignored and nasty ones will result in a ban.
 
User avatar
UPlog
Posts: 443
Joined: Sat Jan 27, 2018 5:45 am

Re: Airline Hub Benchmarking

Fri Oct 25, 2019 11:47 pm

The Hubs of Asia chart is missing percentage numbers.
 
User avatar
ScroogeMcDuck
Posts: 14
Joined: Wed Jun 22, 2016 7:50 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 5:17 am

This is a great read. Thanks!
 
ncflyer
Posts: 1240
Joined: Tue Sep 05, 2000 7:03 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 10:39 am

Finally read it. Terrific. One thing I’ve never understood. It is widely speculated that ATL and CLT are tremendously profitable, perhaps even at the top at least in the USA. Yet they have low share of o&d, I am actually shocked ATL is lower than CLT as ATL is a much larger metro. The author correctly points to the premium that non connecting pax pay. Can someone resolve the apparent contradiction, that low o/d airports are on top! Maybe it’s a sheer volume play.
 
Max Q
Posts: 7858
Joined: Wed May 09, 2001 12:40 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 11:05 am

Interesting


Shame it’s a little short on detail..
The best contribution to safety is a competent Pilot.
 
c933103
Posts: 3892
Joined: Wed May 18, 2016 7:23 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 11:11 am

I thought PVG and PEK have <10% passengers that are connecting?
Peace cannot counter violence when violence are being used by the powerful without mercy. #HongKong
But there is one possible exception. That is if the world could come together and make those who use violence lose their power. #China
 
MIflyer12
Posts: 6197
Joined: Mon Feb 18, 2013 11:58 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 11:21 am

ncflyer wrote:
Finally read it. Terrific. One thing I’ve never understood. It is widely speculated that ATL and CLT are tremendously profitable, perhaps even at the top at least in the USA. Yet they have low share of o&d, I am actually shocked ATL is lower than CLT as ATL is a much larger metro. The author correctly points to the premium that non connecting pax pay. Can someone resolve the apparent contradiction, that low o/d airports are on top! Maybe it’s a sheer volume play.


ATL's O&D is a smaller fraction of a much larger total passenger number. ATL has more O&D.

In a quick read I didn't find anything in this analysis that confirms profitability. We can infer that hubs where the dominant carrier is trying to grow are profitable: ATL and DFW, for examples. If they weren't profitable (as profitable as alternatives, as profitable as their average) they would try to redeploy the assets - like what DL did with CVG and MEM to BOS/SEA/LAX.
 
MaxTrimm
Posts: 249
Joined: Thu Nov 19, 2015 2:43 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 2:18 pm

This is fascinating, and I admire the (surely) great amount of time put into this. Can't wait to see the small hub report.
 
dmstorm22
Posts: 567
Joined: Wed Jun 15, 2016 1:49 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 4:49 pm

This is great, gonna take my time to digest it all.

Fantastic stuff.
 
grbauc
Posts: 1456
Joined: Sat Mar 28, 2015 9:05 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 9:15 pm

OP Great post and job compiling and analyzing the information.

FSdan your right EWR is not the connection hub I tend to believe it is.
 
grbauc
Posts: 1456
Joined: Sat Mar 28, 2015 9:05 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 9:17 pm

MIflyer12 wrote:
ncflyer wrote:
Finally read it. Terrific. One thing I’ve never understood. It is widely speculated that ATL and CLT are tremendously profitable, perhaps even at the top at least in the USA. Yet they have low share of o&d, I am actually shocked ATL is lower than CLT as ATL is a much larger metro. The author correctly points to the premium that non connecting pax pay. Can someone resolve the apparent contradiction, that low o/d airports are on top! Maybe it’s a sheer volume play.


ATL's O&D is a smaller fraction of a much larger total passenger number. ATL has more O&D.

In a quick read I didn't find anything in this analysis that confirms profitability. We can infer that hubs where the dominant carrier is trying to grow are profitable: ATL and DFW, for examples. If they weren't profitable (as profitable as alternatives, as profitable as their average) they would try to redeploy the assets - like what DL did with CVG and MEM to BOS/SEA/LAX.



Good point.
 
ncflyer
Posts: 1240
Joined: Tue Sep 05, 2000 7:03 pm

Re: Airline Hub Benchmarking

Sat Oct 26, 2019 10:54 pm

While nothing in the thread raises profitability it’s been raised in this website in countless places that ATL is a money making machine and that CLT is doing extraordinarily well for AA. Given how much the airport have grown, I buy that, but it goes against the fact that the airports are so reliant on connecting traffic, since as OP points out fare premiums and lower costs are generally for point to point.

Now money making machine could be just because the hubs are so damn huge, but I measure money making as a % of revenue too.
 
2travel2know2
Posts: 2829
Joined: Mon Apr 26, 2010 3:01 pm

Re: Airline Hub Benchmarking

Sun Oct 27, 2019 1:14 am

Great stuff, great research..
I'm not on CM's payroll.
 
User avatar
JBusworth
Posts: 173
Joined: Fri Sep 08, 2017 8:55 am

Re: Airline Hub Benchmarking

Sun Oct 27, 2019 3:21 am

This is absolutely fascinating! Great job. I am looking forward to hearing more.

I noticed that QF's ex-Dubai Kangaroo route stop is listed as one of the niche hubs. It would be interesting to see how their shift back to Singapore stops has changed the equation on the Kangaroo route and subsequently what the flows would look like once they get Project Sunrise underway.
 
ETinCaribe
Posts: 473
Joined: Tue Dec 15, 2009 11:57 pm

Re: Airline Hub Benchmarking

Sun Oct 27, 2019 3:17 pm

The passion of A.netters is awe inspiring. Great analysis.
 
WorldspotterPL
Topic Author
Posts: 256
Joined: Thu May 27, 2004 2:40 am

Re: Airline Hub Benchmarking

Mon Oct 28, 2019 10:48 am

Hey guys, thank you very much for your kind words and positive feedback. Being able to share your passion with so many fellow aviation geeks makes it even more fun and I will start working on the second batch straight away. But let me quickly to through your specific questions.

mikegigs wrote:
Wow... this is fascinating! If you really did all that work yourself (and I don't doubt that you did), then I am extremely impressed. Some of the info is somewhat obvious (DL's ATL hub is BIG and domestic-connection focused), but other information (about the "niche hubs" and "pure connectors") I never would have found out on my own. The comparison to departure banks is super cool too. I think we rarely look at connection statistics in such an in-depth way.

PAA25 wrote:
Wow, this is amazing--thanks for marshaling and visualizing the data! For the visualizations, what software are you using? (Looks like it could either be Tableau or R Shiny.)

proudavgeek wrote:
Very detailed and amazing..I wish I can read it word by word.... Just curious,, did you use Tableau or PowerBI or something else for all the visualizations? Look cool...


- I did it all by myself with Excel and - very good guess - Tableau

FSDan wrote:
A suggestion regarding future data that would be really interesting to see: you alluded to the fact that there's a real difference between hubs that connect a plethora of city pairs with relatively less volume on any given pair, versus those that "specialize" in a few high volume traffic flows. It would be really interesting to see a ranking of relative "generalist" hubs vs "specialist" hubs - perhaps by # of city pairs accounting for X% of total connecting volume, with a lower number indicating a more specialized hub?

Also, a question on the hub bank wave graphs... I think they are interesting to highlight the overall structure of each hub, but it isn't clear to me whether each "stack", if you will, is scaled by number of destinations, seat capacity, or something else.


Thanks, that is a very good point you are making! I will have a look at how much work it would need but a comparison between specialised and generalist hubs is on my list! To address your other point: the waves are simply stacked by flights - each cell represents a destination flown in that time window, coloured in by the continent the destination is on. Each cell has the same size irrespective of aircraft etc.

UPlog wrote:
The Hubs of Asia chart is missing percentage numbers.


- I have no Tableau licence at the moment so I cannot go back to changing those images :(

ncflyer wrote:
Looks cool but is there a cliff notes version?


- there is no cliff notes version unfortunately :)

ncflyer wrote:
Finally read it. Terrific. One thing I’ve never understood. It is widely speculated that ATL and CLT are tremendously profitable, perhaps even at the top at least in the USA. Yet they have low share of o&d, I am actually shocked ATL is lower than CLT as ATL is a much larger metro. The author correctly points to the premium that non connecting pax pay. Can someone resolve the apparent contradiction, that low o/d airports are on top! Maybe it’s a sheer volume play.


Very good question, I would be interested in an answer from someone within an airline, too. I guess part of the answer is that even on one-stop itineraries, hubs have their monopolies - especially the huge ones with all their connecting airports - since in some cases they are the only ones that can connect a very small airport economically to their network that others cannot.

c933103 wrote:
I thought PVG and PEK have <10% passengers that are connecting?


They are as airports, but I am talking about the specific China Eastern/SkyTeam operation at PVG and the Air China/Star Alliance operataion and PEK. I think this is a much better definition of a hub.

JBusworth wrote:
This is absolutely fascinating! Great job. I am looking forward to hearing more.

I noticed that QF's ex-Dubai Kangaroo route stop is listed as one of the niche hubs. It would be interesting to see how their shift back to Singapore stops has changed the equation on the Kangaroo route and subsequently what the flows would look like once they get Project Sunrise underway.


Thanks! I hope I will find the time to put historic (I have 2010 -2017 lying around) and very recent data together. But as others have rightly pointed out, this is a hell lot of work. Then we could talk about a lot of very interesting analyses such as the one you are suggesting. I would love to see, how the hubs evolved throughout the big US mergers for example, or what the dynamics have been between the IAG or Lufthansa group hubs in the last decade.


Part 2 of the analysis will follow this afternoon hopefully.
Best regards and thanks again for all your kind words.
Paul
 
Sokes
Posts: 459
Joined: Sat Mar 09, 2019 4:48 pm

Re: Airline Hub Benchmarking

Mon Oct 28, 2019 6:25 pm

Thanks for all your work. You must really love aviation.

WorldspotterPL wrote:

(Connecting) Passenger Flows via Hubs

...The largest city-pair flow on the list with at least one stop – one could have guessed it – is the original ‘Kangaroo Route’ London to Sydney at 782 passengers per day each way.
...
Los Angeles and San Francisco to Washington D.C. are number two and three; Los Angeles to New York is number four. These American flows are notably all quite evenly split up between the at least ten hubs in the US with the likes of Denver, Chicago, Charlotte, Atlanta, or Phoenix leading most ranks. You can already see that the US is the country of hubs, a fact my analysis will clearly show.


Why would Los Angeles and San Francisco to Washington D.C connect via a hub if they are the second and third most popular city pairs in the world?


WorldspotterPL wrote:
Yet, there are much bigger hubs such as Lufhansa’s Munich operation where the largest city pair is Rome to Hamburg at a mere 23 daily passengers. A hub with a deep network like [email protected], however, has tens of thousands of city pairs with less than one passenger per day, while Eva [email protected] only has a fraction of this number. ...
Dubai – the second largest hub in the world – connects all continents with a mega-fleet of wide-bodied aircraft. Nearby Gulf Air at Bahrain Airport specialises in connecting the Gulf and Middle East to India on A320s.
...
Only 12% of connecting passengers use [email protected] for regional flying between cities of the Gulf and the Middle East while nearly two-thirds alone (65%) use it on the global North-West to South-East-Corridor between Europe, Southeast Asia, the Indian Sub-Continent, and Australia.
...
Moving down the ‘hub-curve’ come the five ‘Very Large Hubs’. They consist of two US-hubs – notably two American Airlines oneworld hubs, Dallas Fort-Worth and Charlotte – and three European hubs, Frankfurt (Lufthansa), Amsterdam (KLM), and Istanbul (Turkish Airlines). Similar to [email protected], both American Airlines hubs are located in the southern states of the USA and they, too, are very regional hubs.
...
The European hubs, on the other hand, all average around 6,000km distance per passenger, even though Turkish Airlines’ hub at Istanbul works very differently than the other two. [email protected] and [email protected] feature a similar set of passenger flows with Europe equally important at 67% of connecting passengers
...
To the bottom left of the ‘Very Large Hubs’ lies a group of eight ‘Large hubs’. ... Starting with Munich at two million inhabitants, the group averages around four million ... The eight ‘Large Hubs’ are a very homogenous group with around 3,500km on average and all in between 3,000 and 4,000km.
...
A look at their location explains the similarities between the members of the ‘Large Hubs’. Not only are the metro areas of similar size, they are also all located within the continent. I thought about naming this group ‘Large Continental hubs’.
...
The distance per passengers patterns of the ‘Large Hubs’ represent the continental nature of their locations. All eight hubs have their by far largest peak at 1,000-4,000km, with the two European hubs peaking earliest at 1,000-2,000km, notably on par with Detroit. The other five US-American hubs – due to further distances in the much larger continent – peak later, between 2,000 and 4,000km. Munich is very Europe-centric at 82% of total connecting passengers, much more than big-sister hub FRA (see above).
...
Paul


Munich has 1,5 million inhabitants, the metropolitan region 6 million.
I'm surprised to see Munich's largest connecting city pair has 23 daily passengers. Great info.
Istanbul has a European side, but Turkey is basically Asia.

I assume you mean the longer leg of the journey of a ‘European Very large City Hub’ is around 6000 km distance flown per connecting passenger?
If somebody flies US to India via Frankfurt, he fits somehow into the 6000 km average. But if 67% of connections in Frankfurt are from Europe the average has to be below 6000 km.
Even Stuttgart-Frankfurt-Singapur or Stuttgart-Frankfurt-LA won't be 6000 km average.

Continuing with this assumption:
I wonder why Munich and Frankfurt or Dubai and Bahrain are so different?
I assume if all long distance passengers would be sent via Frankfurt Lufthansa could use more A321s within Europe and more B777s instead of A330s intercontinental. I further assume there is quite some demand from many European cities to Munich, but not so much O+D as to be able to offer attractive connections or to use good CASM planes. By having Munich as a second hub these connections become possible, even as long range CASM becomes worse.
Wikipedia:
Munich: 413.000 flight movements/ year, 46 Mio passengers/ year
Frankfurt: 512.000 flight movements/ year, 70 Million passengers/ year

Do you have numbers about passengers/ flight movements per year for your hubs? I suspect more hubs are better for RASM, but lead to worse CASM and that there is an optimum somewhere. This optimum probably also depends on spending power?
More hubs have more cost, but offer more direct connections. Is therefore the high price for direct connections justified?
Did the B787-8 cause any airline to split a long range hub into two? Or are there simply a few more direct long range connections from existing very large hubs? Large hubs 3500 km versus very large hubs in Europe 6000 km sounds as if the B777X has a bright future.

I know from Frankfurt that building a further runway was politically opposed. I'm just speculating. All these thoughts are only assumptions.

Dubai and Bahrain sounds like a political agreement to me. Why would Emirates not connect more in the region? Why would Bahrain not connect more intercontinental?
Anyone knows more?
Why can't the world be a little bit more autistic?
 
blockski
Posts: 602
Joined: Wed Jun 15, 2016 8:30 pm

Re: Airline Hub Benchmarking

Mon Oct 28, 2019 6:57 pm

Sokes wrote:
Why would Los Angeles and San Francisco to Washington D.C connect via a hub if they are the second and third most popular city pairs in the world?


I'm guessing by 'Washington, D.C." the author means DCA, which has only a handful of direct flights to SFO and LAX because of the DCA perimeter rule.
 
CIDFlyer
Posts: 2116
Joined: Wed Apr 13, 2005 7:19 am

Re: Airline Hub Benchmarking

Mon Oct 28, 2019 11:12 pm

very interesting! just curious though Im not seeing AA at ORD on the list?
 
muralir
Posts: 107
Joined: Mon May 27, 2013 3:44 pm

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 1:34 am

This is great stuff, and I feel the need to read it twice or three times to digest all the information :-)

One question: ORD is fairly unique in serving as a massive hub for 2 different airlines. How does that factor into your calculations? IMHO, combining the operations would likely put ORD into the super-hub category as you define it.

Also, it appears that UA and AA use ORD in fairly similar patterns (lots of domestic feed from secondary markets feeding large hub-to-hub and hub-to-point operations), with the sole exception being perhaps that AA appears to rely more on their alliance partners for their international ops. Would you agree? Or do you see a bigger difference between the two operations?
 
Blerg
Posts: 2626
Joined: Tue Jan 09, 2018 11:42 am

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 6:19 am

Great work! I'm also especially happy to see Air Serbia and BEG on that list. They are a small carrier that's slowly building a solid hub operations at Belgrade airport. Especially this year they launched a whole bunch of new destinations (KRR, HEL, NCE, BCN, MAD...) and all have been extremely successful both in terms of finances and loads. Some like BCN have already received extra frequencies for next summer.

For example when Adria Airways went bankrupt, they were the only airline to release rescue fares which resulted in most flights in the next two weeks being upgraded from the Atr to a high density B733 (144 seats). Yesterday they inaugurated their 13.15 departure after they increased LJU from 14 to 17 this winter season. Even today several weekly flights get upgraded to the 733, this past Saturday that was the case with both daily flights.Overnight they became the number one airline in Slovenia in terms of frequencies (17 from BEG and 2 from INI).

Belgrade airport is growing nicely as well. In September the airport welcomed 645.000 passengers which is 11.6% more than last year. In 2019 they expect to handle some 6.1 million, up from 5.6 last year. Air Serbia is definitely an airline to follow as they are silently expanding without much PR or fanfare.
 
WorldspotterPL
Topic Author
Posts: 256
Joined: Thu May 27, 2004 2:40 am

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 7:34 am

Hi guys,

Part II online now!!

https://www.airliners.net/forum/viewtopic.php?f=3&t=1433981



blockski wrote:
Sokes wrote:
Why would Los Angeles and San Francisco to Washington D.C connect via a hub if they are the second and third most popular city pairs in the world?


I'm guessing by 'Washington, D.C." the author means DCA, which has only a handful of direct flights to SFO and LAX because of the DCA perimeter rule.


You misunderstood. My list in this case was 'The largest city-pair flow on the list with at least one stop'. So LA-D.C. (all airports) is the second largest one-stop flow in the world, the city pair with the second most pax traveling via a hub, only SYD-LON is larger. I hope that answers your question.

CIDFlyer wrote:
very interesting! just curious though Im not seeing AA at ORD on the list?


If you look closely at the chart with all airports and groups you will see a little AA logo and ORD next to it within the group 'Medium-sized Mega City Hubs' (which I will pick up in Part III) - and you will also find it in the graph about US alliance hubs.

muralir wrote:
One question: ORD is fairly unique in serving as a massive hub for 2 different airlines. How does that factor into your calculations? IMHO, combining the operations would likely put ORD into the super-hub category as you define it.


Not quite, but good point - when added up, [email protected] has similar absolute numbers of transfer pax as [email protected] - but with significantly more O&D. So this hub would lie somewhere below [email protected] and on the right hand side of [email protected] on the plot - maybe this hypothetical hub would have been the only 'Very Large Mega City hub' - definitely no 'Super Hub' because of the total number of connecting pax.

muralir wrote:
Also, it appears that UA and AA use ORD in fairly similar patterns (lots of domestic feed from secondary markets feeding large hub-to-hub and hub-to-point operations), with the sole exception being perhaps that AA appears to rely more on their alliance partners for their international ops. Would you agree? Or do you see a bigger difference between the two operations?


Hopefully I will find the time to make comparisons like this one. I have not had a closer look at the two yet.

Blerg wrote:
Great work! I'm also especially happy to see Air Serbia and BEG on that list. They are a small carrier that's slowly building a solid hub operations at Belgrade airport. Especially this year they launched a whole bunch of new destinations (KRR, HEL, NCE, BCN, MAD...) and all have been extremely successful both in terms of finances and loads. Some like BCN have already received extra frequencies for next summer.

For example when Adria Airways went bankrupt, they were the only airline to release rescue fares which resulted in most flights in the next two weeks being upgraded from the Atr to a high density B733 (144 seats). Yesterday they inaugurated their 13.15 departure after they increased LJU from 14 to 17 this winter season. Even today several weekly flights get upgraded to the 733, this past Saturday that was the case with both daily flights.Overnight they became the number one airline in Slovenia in terms of frequencies (17 from BEG and 2 from INI).

Belgrade airport is growing nicely as well. In September the airport welcomed 645.000 passengers which is 11.6% more than last year. In 2019 they expect to handle some 6.1 million, up from 5.6 last year. Air Serbia is definitely an airline to follow as they are silently expanding without much PR or fanfare.


Interesting insights - thanks! To be honest, I was surprised to see BEG becoming a hub by my definition! But exactly that is the beauty of a research like this. People always talk about Haneda being a double hub - but my data shows both don't make it (by my definition). And a hub like BEG no one would have had on their radar.

Best regards,
Paul
 
konkret
Posts: 307
Joined: Thu Jun 23, 2016 9:26 pm

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 9:42 am

The information that SVO is the top connecting hub between JFK and TLV is pretty surprising to me.
 
zakuivcustom
Posts: 2904
Joined: Sat Jun 10, 2017 3:32 am

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 1:21 pm

First, thanks for clutching all the data. a.net definitely need more of this data-driven analysis.

The classification was also right on in a sense that some of the airports in certain groups doesn't surprised me at all. For example, under "niche hub" there's [email protected] ("Niche" being N. America to/from Europe), [email protected] (East Asia to Europe), or even [email protected] (Eastern Europe), [email protected] (Mini-ET) and [email protected], but all of which has a decent local demand (at least compare to the "pure connectors"). The "pure connector" are the usual suspect, i.e. [email protected], ME2 (EY and QR), [email protected], along with BAH and MCT.

Not surprised at the "Mega City Hubs", either. LHR & CDG in Europe, ORD and YYZ both being fairly large metro with a good geographic location in the "middle"; then there's BKK, SIN, HKG, and also ICN (albeit smaller). Lastly, you got the largest CN3 hub ([email protected], [email protected], [email protected]) standing out, with Shanghai's location "in the middle" (Between "Northern" China aka Beijing and surrounding province and "Southern" China especially the coast like Fujian and Guangdong) along with its relative location further to the east (aka best located for traffic between "interior" China and Japan/North America) making it the most "hub" of the major mainland Chinese airports.

Otherwise, it's just crazy how much ATL, CLT, and DFW stands out, and also how UA doesn't have a "true" fortress hub similar to [email protected] (and CLT) and [email protected]

WorldspotterPL wrote:
Interesting insights - thanks! To be honest, I was surprised to see BEG becoming a hub by my definition! But exactly that is the beauty of a research like this. People always talk about Haneda being a double hub - but my data shows both don't make it (by my definition). And a hub like BEG no one would have had on their radar.


I'm not surprised about HND at all. At the end of the day it's located on an "edge" geographically, which limits any international connection to at best a "niche" hub. TPAC traffic is just not that big compare to TATL anyway (Which is why CDG and LHR make it on your list but not HND or NRT).
Free Hong Kong! Free China!
 
blockski
Posts: 602
Joined: Wed Jun 15, 2016 8:30 pm

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 1:25 pm

WorldspotterPL wrote:
blockski wrote:
Sokes wrote:
Why would Los Angeles and San Francisco to Washington D.C connect via a hub if they are the second and third most popular city pairs in the world?


I'm guessing by 'Washington, D.C." the author means DCA, which has only a handful of direct flights to SFO and LAX because of the DCA perimeter rule.


You misunderstood. My list in this case was 'The largest city-pair flow on the list with at least one stop'. So LA-D.C. (all airports) is the second largest one-stop flow in the world, the city pair with the second most pax traveling via a hub, only SYD-LON is larger. I hope that answers your question.


Yes, that clarifies the data a bit.

I think my explanation for why that market is the second biggest one-stop market remains the same - it's because of the DCA perimeter rule.
 
Sokes
Posts: 459
Joined: Sat Mar 09, 2019 4:48 pm

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 3:59 pm

blockski wrote:
Sokes wrote:
Why would Los Angeles and San Francisco to Washington D.C connect via a hub if they are the second and third most popular city pairs in the world?


I'm guessing by 'Washington, D.C." the author means DCA, which has only a handful of direct flights to SFO and LAX because of the DCA perimeter rule.


That explains it. Thanks.
I had to read what the DCA perimeter rule is about. Let's see if I understood it right:
Reagan national airport is very comfortable located. Transcontinental traffic would be interested to use that airport. However there is not enough space to make it a proper hub. Therefore beside a few exemptions traffic beyond 1250 miles has to go via Washington Dulles, which is less comfortable located.
Is it dirty politics or is it a good thing for the sake of connectivity?
And why would one rather make a stop than fly directly to Dulles. Is Dulles so horribly connected?
Why can't the world be a little bit more autistic?
 
blockski
Posts: 602
Joined: Wed Jun 15, 2016 8:30 pm

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 5:03 pm

Sokes wrote:
blockski wrote:
Sokes wrote:
Why would Los Angeles and San Francisco to Washington D.C connect via a hub if they are the second and third most popular city pairs in the world?


I'm guessing by 'Washington, D.C." the author means DCA, which has only a handful of direct flights to SFO and LAX because of the DCA perimeter rule.


That explains it. Thanks.
I had to read what the DCA perimeter rule is about. Let's see if I understood it right:
Reagan national airport is very comfortable located. Transcontinental traffic would be interested to use that airport. However there is not enough space to make it a proper hub. Therefore beside a few exemptions traffic beyond 1250 miles has to go via Washington Dulles, which is less comfortable located.
Is it dirty politics or is it a good thing for the sake of connectivity?
And why would one rather make a stop than fly directly to Dulles. Is Dulles so horribly connected?


DCA is a small airport. It opened in 1941, well before the jet age. It is in a very central location, but due to noise concerns, restricted airspace around the Capitol and the White House, and the small site, there is simply no room to expand the airfield.

After the region decided to build a new airport (that would become Dulles), the perimeter rule was established to both control traffic at DCA but also push long-haul traffic to IAD. Whether it's 'dirty politics' or a good policy: It's really impossible to separate out the two. DCA was simply not sufficient as an airport for the region, it couldn't accommodate the international flights to Washington. Dulles was also planned with jet noise in mind - part of the reason it's so far away is to limit the impact of jet operations on the surrounding community.

As for why passengers choose to connect rather than fly direct - some of that might be a preference for DCA over IAD; some might be a preference for a particular airline for miles or other loyalty programs.

There are lots of flights between IAD and LAX; United usually has 7 flights a day, Alaska has 2 and American has 1.

There are another 6 flights between BWI and LAX; 3x on Southwest, 1x each for Spirit, Alaska, and United.

If you're loyal to Delta, for example, there's one daily DCA-LAX flight - otherwise you're going to have to connect.
 
atcpeter
Posts: 70
Joined: Thu Sep 05, 2013 7:47 pm

Re: Airline Hub Benchmarking

Tue Oct 29, 2019 9:20 pm

This is so interesting. I love your approach -- we can all differ on exact definitions, but I appreciate the time and effort you put into coming up with hub definitions, and following that the further categorization. There are so many ways to slice and dice this data. I'd love to dive into my personal favorite hubs, but it would also be interesting to look at hub-to-hub flying as a proportion of total/connecting traffic and to see where it falls in the course of the airline's banks at either end.

Do you have the data posted somewhere that we can play with it, too? Tableau is pretty expensive, but maybe a Jupyter notebook or some other tool?
 
WorldspotterPL
Topic Author
Posts: 256
Joined: Thu May 27, 2004 2:40 am

Re: Airline Hub Benchmarking

Wed Oct 30, 2019 7:11 am

Apparently it works better to simply put Part II in this post - so here goes.

Now it is time for Part II where I continue with the medium-sized and small hubs, still going down the 'hub curve' from largest to smallest hubs. In Part III I will describe the really exciting hubs, the special ones - Niche Hubs, Pure Connectors, and Mega-City hubs of different sizes.


‘Medium-sized Hubs’

- 30% - 50% connecting passengers share
- 10,000 - 25,000 daily connecting passengers

Members (20):

Americas (15):

- Air [email protected] (YUL)
- Air [email protected] (YVR)
- [email protected] (SEA)
- [email protected] Lake City (SLC)
- [email protected] de Chile (SCL)
- [email protected] Paolo (GRU)
- [email protected] (ATL)
- [email protected] (BWI)
- [email protected] (DAL)
- [email protected] (DEN)
- [email protected] (HOU)
- [email protected] Vegas (LAS)
- [email protected] (PHX)
- [email protected] Louis (STL)
- [email protected] D.C. (IAD)

Europe (2):
- [email protected] (VIE)
- [email protected] (ZRH)

Asia/Pacific (3):
- Air [email protected] (DEL)
- [email protected] (DXB)
- Malaysia [email protected] Lumpur (KUL)


Image

The medium-sized hubs are dominated by one Airline: Southwest. I have to apologise in advance, though, that I do not have any waves for Southwest hubs! Almost solely located within the continent, its eight hubs in this group average 2,500km weighted distance flown per passenger (ranging from DAL’s 2,140km to DEN at 3,024km). Together with Delta’s Utah hub at SLC (2,728km) and Alaska’s prime hub at Seattle (3,283km), they form a sub-group of short-haul medium-sized hubs. Notably, Lufthansa Group’s [email protected] hub – although sporting a backbone of North American and Asian as well as holiday long haul flights - is also part of this sub-group, connecting mostly Western to Eastern Europe at 3,203km flown per passenger. The other half of the medium-sized hubs are more substantial long haul hubs and average more than double the distance in passenger km flown (ca. 5,800km vs. ca. 2,600km), with Pacific Rim hubs [email protected] de Chile and Air [email protected] leading at around 7,000km.

Let’s take a look at some interesting comparisons. The two Air Canada hubs at the East coast (Montreal) and West coast (Vancouver) work very similarly but geography plays its obvious role. While both have around 70% connecting passengers from North America (domestic at 55% for both), Vancouver connects almost all of the rest (25%) to Asia Pacific. The Quebec hub has a 20% European share with its French roots showing (6.3%). Also, a comparison to nearby big brother Toronto shows Montreal’s historic ties and its specialisation on Francophile countries such as Morocco, the Lebanon, or Haiti.

This cluster also features the two main hubs of the recently merged LATAM – formally LAN of Chile (Santiago hub) and TAM of Brazil (Sao Paolo GRU hub). The Brazilian hub takes a few more connecting passengers from Latin America (GRU 75%, SCL 70%) and this five percentage points difference is mirrored in a larger US share for SCL (15%, GRU 10%) . SCL notably caters more to Argentina (23%) than to its own country (20%), while Sao Paolo has a major 46% domestic share. As is the case in Canada, hub geography explains long haul patterns. [email protected] focuses on Australia and New Zealand (9% of all connecting pasengers) while [email protected] has a 12% share on the large European Economies with Spain roughly twice the size of the UK, France, Germany, and Italy.

In this group, one can also observe the differences in business models between alliance hubs and low cost hubs. Hubs such as [email protected], ATL or PHX have many more ‘thick’ city pair passenger flows between 10 and 50 passengers connecting between two cities per day. For example, while [email protected] D.C. only has 3 city pairs with 15-25 passengers per day, [email protected] has 91 of these flows. Contrary, however, the D.C. hub has more than three times the number of smaller flows between with 0.3 and 3 passenger per day. Similar comparisons can be made between other Southwest hubs and [email protected] or [email protected] Interestingly, [email protected] and Air [email protected] have a very similar mix in city pair thicknesses to the low cost hubs, with many thick pairs and only a third of the thin pairs compared to Vienna, Zurich or D.C. This might be explained by fewer but larger cities in Asia compared to Europe or the US and thus by the differences in the depths of hub-networks.

Image
Air India's Delhi hub has - with a lot of goodwill - four waves. Only the noon-wave is somewhat of size, with Sydney, Paris, Birmingham, Frankfurt, Heathrow, and Milan departures following a stack of domestic arrivals (and vice versa). The rest of the day the operation is very scattered. The night-time long haul pattern is unique. There is no wave whatsoever, instead, one Europe or US-bound flight leaves every 30 minutes from 00:30 until 04:00 in the morning.

Image
Image
The two LATAM hubs in comparison: Sao Paolo's GRU hub (top) has a clear bank of European arrivals in the early morning before 6 a.m., with a bank of regional departures in the hours until 9. While the latter depart, a bank of North American flights arrive around 9 a.m. The other notable wave is in the evening, when the bulk of regional evening flights come in between 7 and 9 p.m. and the long hauls to Europe and North America leave between 9 p.m. and midnight. Most of these long haul flights have a second daily flight that leaves in the afternoon. Santiago de Chile (SCL) (bottom) has a similar looking, purple and very flat set of 'waves'. In fact, I should rather call it a rolling hub. Due to time-zones - similar to Johannesburg for example - Santiago has almost all long haul (only US except for Madrid) coming in from night flights in the morning and leaving in the evening between 9 p.m. and midnight. Note the nice little green dots - these represent Sydney and Auckland flights, a specialty of Santiago.

Image
Image
Lufthansa Group's hubs #3 and #4 - [email protected] (top) and [email protected] (below) have those typical 6 (ZRH) to 7 (VIE) Euro-hub waves. These are not as clinical and thin as seen in many US-hubs - which could be due to lack of runway capacity in Europe as compared to 6 parallels at ATL for example. Like Munich, which has 6 waves (see part I), waves are broader but very defined and visible. Swiss has a much more colourful - hence intercontinental - network, while the bright blue indicates the Eastern European specialisation of the Austian Capital.

Image
Image
Secondary to its main hub Toronto, Star Alliance carrier Air Canada has two smaller hub operations with clear roles in its network. For North American standards, both are more rolling than with defined waves, and their daily patterns differ. Montreal (top) peaks in the morning and the evening with nearly all long hauls being blue for Europe - especially French speaking Brussels, Paris, Geneva, and Nice, most of which are not on the top of the list for trans-atlantic flights at other North American airports. Due to time-zones, YVR (bottom) peaks around noon with - again by North American standards - a lot of yellow Asia-bound flights in the midday bank. Also, like southern Pacific Rim colleague Santiago, it has both Auckland and Sydney flights.



‘Small Hubs’

- 10% - 30% connecting passengers share
- 3,000 - 10,000 daily connecting passengers

Members (38):

Americas (19):
- Air [email protected] Calgary (YYC)
- [email protected] (PDX)
- [email protected] (DCA)
- [email protected] Paolo (VCP)
- [email protected] York (LGA)
- [email protected] (DEN)
- [email protected] Paolo (GRU)
- [email protected] (HNL)
- [email protected] Lauderdale (FLL)
- [email protected] York (JFK)
- [email protected] (BSB)
- [email protected] (BNA)
- [email protected] Angeles (LAX)
- [email protected] City (MCI)
- [email protected] (MCO)
- [email protected] (OAK)
- [email protected] Diego (SAN)
- [email protected] (YYZ)
- [email protected] (YYC)

Europe (10):
- Aer [email protected] (DUB)
- Air [email protected] (ORY)
- [email protected]üsseldorf (DUS)
- [email protected] (TXL)
- [email protected] (IST)
- British [email protected] (LGW)
- [email protected] (OSL)
- [email protected] (OSL)
- [email protected] (ARN)
- [email protected] (BCN)

Asia/Pacific (19):
- [email protected] (BNE)
- Hainan [email protected] (PEK)
- Xiamen [email protected] (XMN)
- Air [email protected] (PVG)
- Hong Kong [email protected] Kong (HKG)
- Malindo [email protected] Lumpur (KUL)
- Air New [email protected] (AKL)
- Vietnam [email protected] (HAN)
- Vietnam [email protected] Chi Minh City (SGN)

Image

The small hubs are the largest group by members and it is a group of interesting pairs of hubs to compare. I cannot do all of them, I am afraid. There are the two private (!!) Chinese sister airlines Hainan and Hong Kong Airlines and their hubs in Beijing and Hong Kong as well as the two secondary-to-Copenhagen sister hubs of SAS – Stockholm and Oslo. What is more, British Airways and Air France both have a very similar set-up of secondary leisure hubs to complement their major Heathrow and CDG hubs in London and Paris respectively. I could go on. There are the two former airberlin hubs at Berlin and Dusseldorf, the oddly similar sized hubs of Vietnam Airlines in Ho Chi Minh and Hanoi, the two West-Jet low cost hubs at Calgary and Toronto, or the quite recently established hubs of JetBlue at JFK and Fort Lauderdale. Apart from these pairs, there is again a vast amount of smaller Southwest Airlines hubs in this category and a few odd but interesting hubs such as the island connector [email protected], recently founded [email protected] Paolo (VCP), or Frontier’s Denver hub that would all deserve a closer look.

Before I do some pairs, I will attempts to make some overall statements about the group even though for the shear size of the group that I quite challenging. In terms of distance per passenger, the group has a few very short haul oriented hubs. These include [email protected] Washington D.C. as well as both very domestic secondary SAS hubs at Stockholm and Oslo. These three peak at 1,000 - 2,000km per pax. Frontier's DEN hub is quite unique in its distance per pax graph, peaking at a plateau between 2,000 and 4,000km. One off is also Hawaiian's HNL hub which has two clear niches - as intra-island connector and islands to mainland USA specialist it peaks once below 1,000km (!!) and once at 5,000km. The group has two clear long haul hubs with Air New [email protected] and [email protected] peaking at 13,000km flown per pax.

As for the aforementioned pairs, let’s start with my favourite hub pair: London Gatwick and Paris Orly. Both complement the airlines’ major hubs at Heathrow and CDG, where all the major business destinations short and long haul are serviced from. Gatwick has a long haul sub-fleet that connects nearly exclusively holiday destinations in Florida, the Caribbean, and the Indian Ocean. On the short haul side, it flies to Greek Islands and other Mediterranean resorts. Most of its major one-stop city pair flows have one clear holiday destination at one end, eg. Glasgow - Orlando, New York – Venice, Dublin - Cancun, or Bridgetown (Barbados) - Edinburgh. Paris Orly by Air France appears similar at first but is a very special case. Orly has two strengths it combines nicely as a hub: First, being closer to downtown Paris than CDG, it has the 'La Navette' shuttle services to all the major secondary centres of France (hourly flights to and from Bordeaux, Marseille, Nice, and Toulouse) mainly for origin and destination (O&D) passengers. What is more, it is home to a high-density fleet of 777s that fly a large O&D - mainly visiting friends and relatives (VFR) as well as sun-seekers, to French overseas departments. As a result, as a (coincidental? Only 10% connecting pax!) hub, it mainly connects secondary French cities to long haul VFR/holiday destinations. Major pax flows include Pointe-a-Pitre (Guadeloupe) to Lyon, Toulouse to La Reunion, or Bordeaux to Fort de France (Martinique). As we have seen in the chapter on ‘Large hubs’, Air France’s Orly hub is a staggering 75% domestic when counting their long haul operation to oversea departments as domestic flights.

Next: Vietnam. It is rare for an airline to have two – and only two - very similarly sized hubs – especially when they are this small (economies of scale, network effects etc.). SGN is the larger and more regional hub. Due to its superior location in South East Asia compared to Hanoi, its South East Asian share of connecting passengers is more than 57% while sister hub Hanoi has 43%. The latter, on the other hand, is more focused on Japan and South Korea (32% vs. 23% at SGN) and true long haul destinations make up 15% of its connecting passengers (vs. 10% at SGN). Hanoi’s role as Capital of Vietnam, despite being the smaller hub, becomes obvious from the destinations of connecting passengers. Ties to former colonial leaders France are double the volume compared to Ho Chi Minh City and there are several links to African countries from Hanoi that SGN lacks altogether. These ties to Nigeria, Kenya, and Congo etc. can probably be explained by government traffic that help execute Vietnamese infrastructural investments in Africa (similar to those of China).

Another pair I choose to analyse in more detail are the two JetBlue hubs. In one sentence, both are prominently US to the Caribbean hubs - although keep in mind they are actually prominently not hubs but O&D focused operations (at 10-15% connecting pax only). O&D or connecting - they have surely played their part in eliminating the island hub of American Eagle at San Juan/Puerto Rico. Yet, there are huge differences between the two B6 hubs. The majority (54%) of flows via JFK still are domestic, with some of the thickest flows being coast to coast flows such as San Francisco or Los Angeles to Burlington or Syracuse. I say ‘thick’ – yet, they only carry five passengers per day each way. Remember Eva Air’s 150 passengers between Ho Chi Minh and Los Angeles? Nearly all other flows are between US-cities and islands in the Caribbean. FLL, on the other hand, only has 13% domestic passenger flows. The rest is mainly US the Caribbean (33% of [email protected] pax go to the Caribbean), South America has only 8% of originating pax via FLL, yet it has the thickest flows. New York to Lima, Medellin, Bogota, and Quito are at the very top of the list with between 9 and 17 PPDEW. While both hubs have 1.1m connecting passengers and almost identical numbers of city pairs below 5 pax per day each way (PPDEW), the Florida hub has significantly more city pairs - and passengers on city pairs - on thicker flows over 5 PPDEW.

Image

Image

Image
Three pretty random hub waves to represent this very heterogeneous group: American's DCA hub (top) is one of the very few hubs in my dataset that has destinations to only one colour. Also - by North American standards - its waves are almost European - not very high and quite broad. Vietnam Airlines' Saigon hub (middle) is very regional and must be defined as a rolling hub as opposed to one with clear waves. [email protected] (bottom) is very regional, too, and has, with some goodwill, 5 or 6 wave-ish structures with a handful of Asia and North America-bound long hauls in the evening that arrive again in the early morning.



Next up in Part III: Special hubs off the ‘hub-curve’: ‘Mega City Hubs’, ‘Niche Hubs’, and ‘Pure Connectors’

Image

Off the ‘hub-curve are five clusters of special hubs. Crawling up the Y-Axis, on the left hand side of the diagram, are the ‘Niche Hubs’ and the ‘Pure Connectors’. Both groups share relatively small numbers of total connecting passengers and high or very high percentages of connecting passengers. In other words: they are the truest hubs, the airline/airport operations exist because they connect passengers - hubbing in these cases is no by-product of an airport that is large anyway because it caters to a large metropolitan area.

Additionally, to the (bottom) right of the three main hub groups on the ‘hub-curve’ (small, medium and large hubs) are what I defined as the ‘Mega City Hubs” of different sizes. For they have a larger number of absolute connecting passengers while having a similar or lower share of connecting passengers than non-mega city peers. This equals a higher number of O&D passengers which in turn points to large catchment areas. Fittingly, nearly all of these hubs are located in or in proximity to so-called mega cities – metropolitan areas (‘metros’) with around ten million inhabitants or more. Throughout my analysis I only use this 10 million mark as a vague reference to distinguish between the many average metros between two and five million inhabitants and the larger ones. A metropolis such as Hong Kong, with officially 7.4 million inhabitants, that lies in the urban area of the Pearl River Delta can be counted as a ‘Mega City’. Yet, there are mega city airports in the regular hub groups (‘Small Hubs’ for example) and non-mega-cities in the Mega City Hub groups since the categories connecting passengers and its share are the primary variables of grouping the hubs. Let’s start with the three groups of ‘Mega City Hubs’...

to be continued.


See you soon! Feedback and discussions are welcomed, or should I say demanded? :)
Best regards,
Paul
 
WorldspotterPL
Topic Author
Posts: 256
Joined: Thu May 27, 2004 2:40 am

Re: Airline Hub Benchmarking

Wed Oct 30, 2019 3:02 pm

Part III:


Special hubs off the ‘hub-curve’ I: ‘Mega City Hubs’

Off the ‘hub-curve are five clusters of special hubs. Crawling up the Y-Axis, on the left hand side of the diagram, are the ‘Niche Hubs’ and the ‘Pure Connectors’. Both groups share relatively small numbers of total connecting passengers and high or very high percentages of connecting passengers. In other words: they are the truest hubs, the airline/airport operations exist because they connect passengers - hubbing in these cases is no by-product of an airport that is large anyway because it caters to a large metropolitan area.
Additionally, to the (bottom) right of the three main hub groups on the ‘hub-curve’ (small, medium and large hubs) are what I defined as the ‘Mega City Hubs” of different sizes. For they have a larger number of absolute connecting passengers while accommodating a similar or lower share of connecting passengers than non-mega city peers. This equals a higher number of O&D passengers which in turn points to large catchment areas. Fittingly, nearly all of these hubs are located in or in proximity to so-called mega cities – metropolitan areas (‘metros’) with around ten million inhabitants or more. Throughout my analysis I only use this 10 million mark as a vague reference to distinguish between the many average metros between two and five million inhabitants and the larger ones. A metropolis such as Hong Kong, with officially 7.4 million inhabitants, that lies in the urban area of the Pearl River Delta can be counted as a ‘Mega City’. Yet, there are mega city airports in the regular hub groups (‘Small Hubs’ for example) and non-mega-cities in the Mega City Hub groups since the categories of connecting passengers and its share are the primary variables of grouping the hubs, metros are only secondary. Let’s start with the three groups of ‘Mega City Hubs’.


Large ‘Mega City Hubs’

- 40% - 55% connecting passengers share
- 45,000 - 65,000 daily connecting passengers

Members (4):

Europe (3):
- [email protected] (SVO)
- Air [email protected] (CDG)
- British [email protected] (LHR)

Asia/Pacific (1):
- Cathay [email protected] Kong (HKG)


Image

The ‘Mega City Hubs’ of large size range from the bespoke Hong Kong (Cathay Pacific) to the European mega-cities Paris (Air France), Moscow (Aeroflot), and London (British Airways) and have an average population of 11.4 million. In comparison, the regular large and medium sized non Mega city hubs consist of many much smaller cities such as Munich (Lufthansa), Phoenix (American & Southwest), Salt Lake City (Delta), Minneapolis (Delta), Vienna (Austrian), Zurich (Swiss), or Brussels (Brussels Airlines) that in total have an average population of 5.6 million.

The four Large ‘Mega City Hubs’ consist of three very long-haul driven operations and Moscow Sheremetyevo (SVO), which has a very high domestic share - which, by the way, in Russia's case - does not necessarily exclude long haul flights! Moscow to Vladivostok is 9h on a Boeing 77W and that is over landmass only, no New York to Hawaii or Paris to Tahiti! Aeroflot's SVO hub averages only 5,500km per passenger, the other three hubs around 8,000km. The two European hubs CDG and LHR have their by far largest passenger peak at long haul to short haul city pairs between 9,000 and 11,000km. This is very interesting in comparison to [email protected] and [email protected] which both peak at around 2000km distance flown, with smaller peaks at the long haul end. [email protected] and especially [email protected] have much smaller short haul peaks. [email protected] only has a third of the passengers at around 2,000km compared to [email protected], [email protected] only a fifth. The British Airways hub has an extreme focus on US cities with more than a third of all passengers going transatlantic (including Canada at 3% and South America at 2%). Air France’s hub at CDG has a much more variable long haul operation with the US dominating at only 12% and many more, especially African, countries in its network (baby-sister hub Orly takes nearly all of the VFR routes to French departments in the Caribbean and the Indian Ocean with a sub-fleet of high-density 777-300/ER aircraft).

Cathay’s HKG hub gets its 8,300km passenger average from a different curve of distances flown. It leaves out the Euro-typical peak at 2,000km to peak at 3,000-4,000km distance, representing its strategy of focusing on connecting the metropolises of China and Taiwan (31% of connecting passengers) to those in South and South East Asia (26%). What is more, [email protected] has the historical role of connecting the politically problematic relationship between Mainland China and Taiwan, between which after 60 years only in 2008 the first direct flights were reestablished. At 1,600 daily connecting passengers between the two, [email protected] carries as many people on this country pair as [email protected] does between non-London UK airports and the US. [email protected] has its second peak at very long distances between 10,000km and 15,000km. This peak is flatter but longer compared to the Euro-hubs’ long haul peaks. This shows Cathay’s stake in the kangaroo routes from Europe to Australia as well as its major role as a connector between North America and the up-and-coming economies of the so-called ‘VIP countries’ (Vietnam, Indonesia, and the Philippines).

In terms of thickness of city pairs - as seen in other groups – the Asian hub HKG sports by far the most pairs with 25 to 100 passengers per day each way (114) while Air France’s CDG hub, for example, only has 4 of these. Yet, Paris has three times the number of city pairs with between 0,3 and one passenger per day compared to HKG – although ‘Very Large Hubs’ [email protected] and [email protected] still easily top [email protected] in this category.

Image
Air France's CDG operation is very impressive, especially colour-wise. It is almost a piece of art how it connect all the continents. Check out the African arrivals in the early morning! People make a big deal about Brussels being the Africa-specialist - and of course it is, considering its size and the percentage of African flights compared to other long hauls. But in fact, by absolute numbers, CDG is the real gateway to Africa. Also, check out the South America and Asia wave just before midnight! By its wave-pattern, CDG is similar to the Euro-hub like [email protected], [email protected] etc., but with 8 waves. As seen in the previous chapters, [email protected] works a bit differently with an epic wave throughout the morning and more defined waves during the rest of the day.

Image
Cathay's Hong Kong hub is a rolling hub. The only real waves centre around the early morning bank of European and North American arrivals and their departures around midnight. In both cases, the O&D one-day business-trip flights coincide nicely with the long haul waves. Remember that all four hubs in this group are of very similar size! Just looking at the much smaller magnitude of HKG's waves you can tell they fly a wide-body-only fleet (yet with partner Cathay Dragon operating A320 family aircraft)

Image
Of course, Heathrow is special. It is as an airport in general, and it is as a hub. [email protected] is the definition of a rolling hub, and it is very colourful. It is hard to make out any sort of pattern, the all-dominating North American flights arrive and depart throughout the entire day until time-zones make flights impossible.

Image
Aeroflot's SVO hub has three or four very gentle waves, depending on how you look at it. The light blue colours indicate its focus on domestic and Eastern European destinations. Interestingly, many European flights arrive in the middle of the night, a fact that must have to do with aircraft utilisation more than passenger comfort, but please correct me if I am wrong.


Medium-sized ‘Mega City Hubs’

- 35% - 50% connecting passengers share
- 20,000 – 35,000 daily connecting passengers

Members (11):

Americas (8):
- [email protected] City (MEX)
- Air [email protected] (YYZ)
- [email protected] (ORD)
- [email protected] (MIA)
- [email protected] (PHL)
- [email protected] (BOG)
- [email protected] (MDW)
- [email protected] Francisco (SFO)

Europe (1):
- [email protected] (MAD)

Asia/Pacific (2):
- Singapore [email protected] (SIN)
- [email protected] (BKK)


Image

The home metros of the medium-sized ‘Mega City Hubs’ have an average population of 9.3 million. The group is very America-heavy with only Madrid (Iberia) and the two Asian hubs Singapore (Singapore Airlines) and Bangkok (Thai Airways) being exceptions. The latter two Asian hubs are clear-cut long haul hubs at average city pair distances of 9,200km and 7,200km respectively. They also feature by far the thickest city pair flows with each over 60 pairs over 25 passengers per day each way. Other hubs in this group such as [email protected] or [email protected] not even have ten flows of this size. The two South East Asian Star Alliance power houses SIN and BKK both have their thickest country flow to and from India (India-Australia at SIN and India-Thailand at BKK). In absolute terms, Australia is by far the most important connecting passenger market for Singapore Airlines in terms of volume (17%) followed by Indonesia (11%) and India (9%). In Bangkok, 21% of Thai’s connecting passengers have their origin or destination in Thailand, followed by China (9%), Japan (8%), India (7%), and Australia (7%).

Before I go into the bulk of North American hubs in this cluster, let me say a few words about the other odd-one’s-out. Iberia’s Madrid hub famously specialises in connecting Spain and the rest of Europe to South America. Not surprisingly, city pairs such as Barcelona to Buenos Aires, Mexico City, or Santiago dominate the list of flows around a – by European standards – very thick 50 passengers per day each way. Avianca’s Bogota hub has one of the earliest peaks (below 1,000km flown) in the graph of connecting passengers by km flown compared to all global hubs. This shows its strong focus on domestic connections (37%) in the small but mountainous country. It then has two smaller peaks at long-haul connections between South American destinations and the US (4,000-5,000km) and South America to Europe (8,000-9,000km). Aeromexico’s MEX hub in comparison has a unique long flat plateau-style peak (1,000-5,000km) between its major domestic operation (32%) and its US and South American flows.

The six North American hubs in this cluster consist of two long-haul-ish hubs [email protected] Francisco and Air [email protected] at around 6,000km per pax. Further, there are three short-haul hubs at Chicago (American and Southwest) and Philadelphia (American) at around 3,000km. Then there is an odd-one out – American Airlines’ Miami hub (4,500km). The latter has its by far largest peak at 3,000-4,000km per passenger, revealing its strong focus on US to South American and Caribbean destinations. The comparison to United SFO hub is interesting. Both located at the edge of the US land mass, [email protected] only caters for 6% domestic US to US traffic. [email protected] on the other hand has 58% US to US connecting passengers, with city pairs such as Portland to Los Angeles or San Diego to Seattle leading the list. Its long-haul hub status comes from its US to Asia flows that make up more than 20%. Last but not least, comparing the two Chicago hubs shows great similarities. Both have double peaks at 1,000-2,000km and 3,000-4,000km per passenger, both drop slightly in between. American has around 25% more volume on the first peak and equal volumes on the second. Of course, as opposed to Southwest, American has a long haul operation out of ORD, yet, this is very small in volumes compared to other hubs in this cluster. SFO, Singapore, Bangkok, and Madrid easily top [email protected]’s long haul passengers at all stage lengths, sometimes by a factor of five to six.

Out of interest I have added United’s ORD hub from the Large Hubs cluster for some analyses. It, too, shows the same double peak structure at roughly double the volume of the other two hubs. This double peak that all three hubs show clearly represents the geography of the location Chicago as a hub for the Mid-West. The first peak is Mid-West to the East Coast; the second is Mid-West to West Coast. Comparing both long haul operations of UA and AA, United’s transatlantic (7,000-8,000km) connecting services peak at three times the volume of American’s while both share a similar amount of connecting passengers at the Asian end of the curve, between 10,000 and 15,000km per passenger. I find this analysis of the multi-hub city Chicago so interesting that I might add a chapter at the end where I will explore more hubs that share the same geography but are in different hub clusters.

Image
Image
The two main South East Asian hubs that share the same alliance: [email protected] (top) and [email protected] (below). In my opinion both are surprisingly flar, especially BKK has no proper waves except for the early morning and the late evening. Due to time-zones and a very similar business model, colours in both images are almost identical. European and Indian destinations coming in in the morning and both going out again in the late evening and afternoon respectively. Its almost as if SIN is simply a double BKK with more secondary destinations and more frequencies to the same regions.

Image
As you would expect, Iberia's hub at MAD is a typical European hub with broad but defined waves, but with purple and pink dots for Latin America instead of yellow or green ones for Asian destinations. It has close to no Eastern European destinations.

Image
Miami is nowhere like all the other US-hubs we have seen so far. [email protected] is much more colourful and a true gateway between all three Americas and Europe!

Image
[email protected], too, is not a very typical US-hub. It lacks the clinical waves. It isn't as colourful as MIA, but compared to the hubs within the continent it is very long haul oriented.



Small ‘Mega City Hubs’

- 10% - 30% connecting passengers share
- 5,000 - 25,000 daily connecting passengers

Members (32):

Americas (11):
- Aerolinas [email protected] Aires (AEP)
- [email protected] Angeles (LAX)
- [email protected] York (JFK)
- [email protected] Paolo (GRU)
- [email protected] Angeles (LAX)
- [email protected] York (JFK)
- [email protected] (BOG)
- [email protected] Paolo (CGH)
- [email protected] Angeles (LAX)
- [email protected] York (EWR)

Europe (4):
- [email protected] (IST)
- [email protected] (SAW)
- [email protected] (DME)
- [email protected] (VKO)

Africa (1):
- [email protected] (CAI)

Asia/Pacific (17):
- Air [email protected] (PEK)
- Air [email protected] (BOM)
- All [email protected] (NRT)
- [email protected] (ICN)
- China [email protected] (TPE)
- China [email protected] (PVG)
- China [email protected] (CAN)
- [email protected] (NRT)
- Eva [email protected] (TPE)
- Japan [email protected] (NRT)
- Jet [email protected] (DEL)
- Jet [email protected] (BOM)
- Korean [email protected] (ICN)
- Philippine [email protected] (MNL)
- [email protected] (SYD)
- [email protected] (JED)
- [email protected] (RUH)

Image

At the bottom end of the ‘hub-curve’ is the biggest group, the cluster of ‘Small Hubs’ and its neighbour to the right, the ‘Small Mega City Hubs’. The former cater to an average metro of only seven million while the latter average 19 million. Note that there are several mega cities in the ‘Small Hubs’ group such as London or Paris. However, the priority lies within the grouping by connecting passenger and its share, the sub-grouping by mega cities is secondary and there is not always a clear cut.

The ‘Small Mega City Hubs’ are a very interesting group in many ways. First, the population of their home metros is more than double the size of the medium and large Mega City Hubs. Second, half (16 of 32) of these hubs are located on the Pacific Rim, most importantly in the multi-hub cities Los Angeles (American, Delta, United), Taipei (Eva Air & China Airlines), Tokyo (ANA, JAL & Delta), and Seoul (Korean Air & Asiana). Not surprisingly, with an average weighted passenger-distance of around 8,600km (vs. 5,000km for all hubs in total), they cater for a large proportion of Trans-Pacific traffic long-haul traffic, mostly along with a significant amount of very regional hubbing.

The vast number of multiple hubs in one metropolitan area or even at one airport is in itself notable in this group of ‘Small Mega City Hubs’. New York has three of its five hubs in this cluster (at two different airports), Los Angeles three of its four (all at LAX). Sao Paolo features two hubs (of six in total) at two airports while Tokyo has three hubs at Narita. Many more double hub cities – Seoul, Mumbai, Moscow, Istanbul, and Taipei - complete this observation.

The ‘Small Mega City Hubs’ cater for only 10% - 30% connecting passengers and, especially at the lower end, hubbing appears to be merely a by-product of a large airport operation. Take Asiana’s Incheon hub at only 16% or Pegasus’ Istanbul hub at 12% for example. On the other hand, United’s Newark hub (29%) is by far the strongest of all five New York City hubs and its 8.7 million annual connecting passengers are in absolute terms as many as those of Delta’s Salt Lake City or Swiss’ Zurich clear-cut hubs.

In such a large group it is difficult to compare all hubs to one another. So in terms of thickness of city pairs and distribution of average distances flown, I will only go into a few significant examples. First, thickness of city pairs: Eva Air’s Taipei hub stands out with 35 flows with over 25 passengers per day each way, three of which even have more than 100 (Ho Chi Minh City – Los Angeles and vice versa as well as Ho Chi Minh City – San Francisco). On the other side of the chart the Taiwanese hub only sports a mere 600 city pairs with between 0,3 and one passenger per day each way – compare this to [email protected]’s 15,000! United’s Newark hub has by far the most diverse network in the group and has around 5,000 of these flows. Second, distribution of average distances flown per passenger: Most ‘Small Mega City Hubs’ peak between 2,000kms and 4,000kms. Eva [email protected] is again an interesting case with nearly no connecting passengers at all except between 10,000 and 16,000km distance. This resembles the airline’s strategy of connecting – in competition foremost to Cathay [email protected] – South East Asia to North America. Vietnam, the Philippines, and Thailand to the US alone make up more than half of the entire traffic via TPE on Eva Air.  

Image
Image
Image
Chinese state-run hubs from North to South: Air China's PEK hub (top) has an interesting pattern without arrivals in the morning. All planes appear to be arriving the evening before. Other than that it is a rolling hub. China Eastern's Shanghai operation (centre) is somewhat similar although with some more arrivals in the morning, so is China Southern's main hub at Guangzhou (bottom). Can anyone explain this phenomenon? Is it a curfew issue?

Image
Last but not least here is Qantas' Sydney hub. After a (curfew-driven?) very sudden start to the day with no hub-typical before-6-a.m.-arrivals, [email protected] is a rolling hub with regular peaks.


Ok guys, I am done for the day, Niche and Pure Connector hubs will follow the coming days.
Best, Paul
 
KeithMason
Posts: 2
Joined: Wed Oct 30, 2019 9:39 am

Re: Airline Hub Benchmarking

Wed Oct 30, 2019 4:20 pm

Hi there. There is some interesting analysis here. I would like to use some of these charts in my lectures. Who should I say is the source? Thank you.
 
WorldspotterPL
Topic Author
Posts: 256
Joined: Thu May 27, 2004 2:40 am

Re: Airline Hub Benchmarking

Wed Oct 30, 2019 7:23 pm

Keith, that source would be your Cranfield ex-student Paul from Germany (2012/13) ;) thanks for the compliments. Let's PM.

Popular Searches On Airliners.net

Top Photos of Last:   24 Hours  •  48 Hours  •  7 Days  •  30 Days  •  180 Days  •  365 Days  •  All Time

Military Aircraft Every type from fighters to helicopters from air forces around the globe

Classic Airliners Props and jets from the good old days

Flight Decks Views from inside the cockpit

Aircraft Cabins Passenger cabin shots showing seat arrangements as well as cargo aircraft interior

Cargo Aircraft Pictures of great freighter aircraft

Government Aircraft Aircraft flying government officials

Helicopters Our large helicopter section. Both military and civil versions

Blimps / Airships Everything from the Goodyear blimp to the Zeppelin

Night Photos Beautiful shots taken while the sun is below the horizon

Accidents Accident, incident and crash related photos

Air to Air Photos taken by airborne photographers of airborne aircraft

Special Paint Schemes Aircraft painted in beautiful and original liveries

Airport Overviews Airport overviews from the air or ground

Tails and Winglets Tail and Winglet closeups with beautiful airline logos