Passenger Connection Management in scheduling – why Origin and Destination (OnD) matters (1/2)

The ability of offering passenger connections is one of the unique selling propositions of any network carrier. Enabling this USP requires a smart scheduling approach, in which the network goes through constant optimization and in which the individual itineraries (offered for purchase and booked already) are protected as good as possible. This blog post discusses reasons why minding passenger connections is so crucial, why modern scheduling systems often lack Origin and Destination (OnD) decision support and how smart system design can bridge this gap.

Why minding passenger connections in Scheduling?

The ability of offering passenger connections is one of the unique selling propositions of any network carrier. This sentence is more than a truism in the airline world; we can well confirm it by facts:

  • Lufthansa showed for example in its “Politikbrief 2018/5” (“Political letter 2018/5”), that a long-haul flight Frankfurt to Chicago carries 73% connection passengers coming mainly from all over Europe.[1]
  • Emirates, Qatar Airways, Etihad Airways and Turkish Airlines have established super-connecting hubs between Europe and Asia and increased indirect connectivity share between EU and Asia from 4.9% in 2007 to 19% in 2017. [2]
  • Also, the low cost carrier are ramping up: Eurowings, Norwegian and easyJet started to offer options for passengers to book connection flights. [3]

Additionally sudden changes in both the global and regional economies are a threat to the profitability of any carrier. This means that carriers face an increasingly volatile situation in their business environment. As a network carrier, another aspect is key to lasting success: an optimized route network. Combining these aspects, as a network carrier, you’d better be prepared. You have to identify and analyze market changes and develop possible solutions allowing you to switch your competitive role from being reactive to pro-active. You have to make it work within your complex network and you must never forget that your network is nothing more than a functional instrument allowing a passenger itinerary from origin A via your hub(s) to destination B (OnD).


Stake of scheduling departments

But, how does a carrier’s scheduling department come into the game now? One part of the answer lies in the central role of scheduling within the airline planning process. The schedule decides about if and how different flights are combinable to both feasible and attractive OnD connections. The other part lies in one aspect, which almost all modern airline schedules have in common: change.

Every schedule change has the potential to affect the feasibility and attractiveness of the passenger connections offered in the market, especially due to the common hub management principle of bundling inbound and outbound hub traffic into waves. Just imagine the intercontinental flight to Chicago gets time shifted to an earlier point in time and passengers from several continental flights can no longer connect. The utilization of both, the intercontinental and the feeder flights, will significantly drop.

So, do all network carriers have to fear schedule changes now? The answer is no, schedule changes are a very normal part of aviation life and most likely the first thing that was invented after schedule. The challenge is to enable an OnD based network management perspective, which allows the scheduling department to make the right decisions ahead of time.

Before we look at the question, how to enable an OnD based network management perspective, let me share with you some thoughts about network planning and scheduling, so that we are all on the same page.


Network planning, scheduling and their collaboration

OnD network planning within a network carrier aims to develop, optimize, and evaluate long-term and medium-term schedules. It tries to get answer to questions, such as:

  1. What is the demand?
  2. How can we create a schedule fitting the demands?
  3. Which routes are the right ones?
  4. How do changes made by competitors affect my network?
  5. How do we optimize our profit with the network?

Even though many carriers handle these questions with secret care, their processes and network planning tools work based on similar processes. They simulate new connections, create forecasts for passenger flows, costs and revenues, and identify the strengths and weaknesses of a carrier’s route network. In order to get a bigger picture including the supply of competitors, most network planning tools utilize global supply and demand datasets to create a market model. Carriers often have to purchase these datasets (e.g. Marketing Information Data Tapes, MIDT).

On the contrary, to the long-term perspective of network planning, scheduling is mid-term or short-term focused. The scheduling department aims to create a fully feasible schedule, to anticipate future developments, and to react quickly to ever-changing environments. This agility only works based on constant exchange of information with slot management, crew planning, revenue management and other departments and results in a plenty of schedule changes.

Network planning and scheduling usually have to collaborate in order to find an initial desired schedule, which on one hand allows rotation building but on the other hand is respecting major operational constraints. In practice, scheduling usually gets its input from network planning and rolls out a sample week to the whole flight schedule under consideration of all necessary schedule and operational constraints. From that moment on, the scheduling departments often starts working independently of network planning and starts to exchange information with the afore mentioned departments.


The risk of “forgetting” the OnD aspect

The risky part is that most of before mentioned departments naturally think in leg (one single flight event), block times (taxi-out, flight, taxi-in) ground times. Slot management needs to know the exact ground times at the airports and the connection of legs in the rotation. The block times are crucial for crew pairing feasibility and even many revenue management systems still work on a leg basis only.

For the scheduling department it seems to be tricky already to accommodate all requirements and change requests of the other departments to single legs and to create an optimized and resilient schedule at the same time. The OnD perspective can therefore easily move to the background, due to several reasons:

  • The plenty of OnD combinations and their feasibility and attractiveness
  • The lack of availability of OnD based forecast and booking data from revenue management systems
  • The insufficient display of the afore mentioned information in the scheduling systems

Unless the scheduler has the OnD and network perspective in his guts, it can easily happen that schedules are managed on a leg basis only and negative network impacts are unavoidable.

OnD restrictions of state-of-the-art scheduling systems

Smart integrations to other systems already aim to make a scheduler’s life easier and to assist him in his decision-making. Airport slot times or crew pairing designators can enrich the display of legs in scheduling systems. Even leg-based passenger forecasts can already be displays on a leg and indicate the scheduler if the planned equipment will meet the passenger demand.

However, in most cases there is no display of quantified passenger itineraries through the carrier’s schedule and beyond, for example to codeshare flights, which are operated by partner carriers. The same applies for a metric, how schedule changes affect the attractiveness of OnD connections.In some carrier setups “feedback loops” to the network management perspective though do exist: hub optimizers try to lower the risk of harmful network impacts, but unfortunately, they operate in a drastically limited solution space if airport slots or crew pairing must be considered as static.

In conclusion: network planning systems have the capabilities of network optimization and OnD connection attractiveness metrics. Scheduling systems on the other side insufficiently enable the network management perspective in decision support.


Necessity of OnD reflection in scheduling decisions – our vision for OnD based decision support

Having identified this gap the following aspect is important within scheduling: It really matters for scheduling decisions, what the carrier intends to sell and what has been sold already. Decision making in scheduling has to consider an OnD framework, if commercial success is the goal. In fact, a continuous OnD based picture in scheduling systems will therefore improve the quality of scheduling decisions.

How can such an OnD framework look like and how can it be brought into a scheduling system? To my point of view three main features should be covered:


illustration of 3 main features of OnD decision support in scheduling systems

1. Visualize the OnD connection:

A smartly integrated scheduling system visualizes OnD dependencies within the carrier network and beyond. It quantifies the importance of single connections based on the number of expected passengers and their assignment to high-yield or low-yield customer segments. Additionally a model derives the relative quality of connections by comparison to hard and soft connection model constraints and external factors.

2. Alert at OnD violations

The system alerts if OnD dependencies are violated. Hard constraints like falling short of the minimum connection time as well as soft constraints like extensions of connection times to unattractive durations generally lead to alerts. Threshold definitions filter the list of alerts to the most harmful ones.

3. Estimate network impacts

The system continuously monitors if all planned and already sold OnD connections are feasible. The system estimates network effects of scheduler actions and external factors.


How to accomplish the vision?

The before-mentioned vision does not yet cover how to implement it into a decision support feature of a scheduling system. There are plenty of challenges and the ideal OnD decision support feature might not be the best fit for every business case.

Find out more in my second blog post concerning this topic or meet my colleague Zuehal Aydin and me at this year’s AGIFORS SSP conference. Your feedback and even better your ideas towards this topic are highly appreciated. Please message me as well, if you would like to exchange more information.