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

This blog post continues the discussion, why minding passenger connections is so crucial for any network carrier. We outline challenges, a roadmap and a passenger connection evolution ladder for decision support in scheduling.

Recap “Why minding passenger connections in Scheduling?”

The ability of offering passenger connections is one of the unique selling propositions of any network carrier. This was the opening sentence of my previous blog post, in which I gave a first overview, why minding passenger connections on an Origin and Destination (OnD) basis is so important for network carriers.

I introduced how network planning and scheduling processes usually work, how risky it can get, if passenger connections are “forgotten” during scheduling, and that even state-of-the-art scheduling systems only insufficiently enable a network management perspective.

I concluded that an OnD based decision support feature would enable this network management perspective, if it

  1. visualizes OnD connection in the scheduling system,
  2. alerts in case of OnD violations and
  3. estimates OnD network impacts.

Given that, it already became obvious that this vision would not be achievable in an easy manner. One would have to question, how to handle certain challenges but also whether a highly sophisticated but most likely expensive solution makes sense for every carrier.

Within this blog post, I will focus on these questions by giving you an overview about the challenges associated with our vision, followed by some thoughts about where to compromise. Derived from challenges and compromises I will share with you an evolution ladder of possible passenger connection decision support features for scheduling systems. This shall provide you with an assessment tool, on which step of the ladder different carriers stand.


The proposed ideal solution comes with challenges

When looking at the three key targets - visualizing, alerting and estimating - we can derive challenges. Please note the following list only represents a selection of challenges:


  • How to display a passenger connection?
    An OnD dependency exists whenever one or several passengers intend (or are forecasted) to connect at a hub airport between two flight segments. A scheduling system should therefore draw a symbolic link between the inbound and the outbound segment for visualization to the scheduler. If there are comparably few passengers on a connection the scheduling system should ignore or highlight it with a light visualization only, if there are more passengers the visualization should be more intense. That seems pretty straight forward, but the challenge arises when asking “What is the scope?”.
  • What is the scope?
    The number of possible passenger connections usually is very high as the “network” of passenger connection links is complex. Passengers may connect to flights, which are not managed by and often not visible for the individual scheduler (codeshare flights of partner carriers or flights managed by other schedulers, e.g. short-/long-haul flights).
    Another aspect of this question is the definition of monitored OnDs. A simple combination of all inbound with all outbound flights results in to many unreasonable passenger connections (e.g. a return flight A - B – A with short layover in B only). One could filter out unreasonable connections with simple algorithms (e.g. application of a maximum detour factor: setting distances of direct and connecting flights in relation), but this process is error-prone. Ultimately, a carrier will have to define and maintain its set of OnD connections (as base data or as a feed from other systems).
  • What triggers a violation?
    (Scheduling) systems usually distinguish between hard and soft constraints. Whilst the violation of hard constraints makes a passenger connection completely infeasible, soft constraints lower the quality of a passenger connection. The major hard constraint is the Minimum Connecting Time (MinCT: layover time between inbound and outbound flight). Soft constraints on the other hand can be diverse. At best, we can summarize them as “passenger connection quality”.
  • What impacts the “passenger connection quality”?
    There is a good and a bad response to this question:
    The good one is, that network planners have a sound understanding about passenger connection quality. They have derived reliable connection building models, respecting aspects of supply (flight times, airports, aircraft types, etc.) and demand (customer segmentation, competitor offerings, etc.).
    The bad one is, that connection building models require a broad range of data, which must be purchased by carriers at comparably high costs.


How to compromise and how to roadmap? Four levels of OnD decision support

Given the challenges above, we might have to compromise to build a solution, which makes economic sense. At Lufthansa Systems, we believe in roadmap approaches with incremental improvements starting at a minimum viable product. For OnD decision support features, we came up with the “ladder of evolution” of OnD decision support:



  • Level 1: Our minimum viable product
    On level 1, we foresee that carriers can define their (most important) passenger connections and add average passenger amounts. The passenger amounts can be global average values for a complete season, single weeks / traffic days or even single days. The carrier can simply utilize historic data, assuming that today’s traffic pattern has similarities with last year’s traffic pattern.
    The definition of the MinCT serves as the only (hard) constraint and a feasibility warning is in place, which warns the scheduler in case of violations.
  • Level 2: Improving data feeds and connection building:
    On level 2, the most crucial improvement is a data feed from the reservation and/or revenue management system providing a real passenger forecast on OnD basis.
    In addition to the MinCT monitoring, we introduce a simple connection building logic to determine the quality of passenger connections. The logic however is limited to factors available within the schedule, such as times of the day of departure and arrival times or the used aircraft type.
  • Level 3:
    On level 3, we feed global schedule data, holding all commercial flights offered (purchasable at providers like OAG or Innovata) into the scheduling system. These kind of schedule data feeds is usually more affordable than demand and supply datasets and holds the potential of further improved connection scoring. The logic of deriving the connection quality is now also capable of respecting external factors (e.g. flight times of competitors on an OnD). Additionally, monitoring the passenger connections linked to partner schedules (codeshare flights) becomes possible as carrier and partner schedules are synchronized.
  • Level 4:
    On the ultimate level 4, we foresee utilizing global demand and supply datasets for even improved connection building and scoring. If allowed by computation power, carriers could think about a “real-time” monitoring of competitor actions and their impact on the passenger connections offered. With such a comprehensive scoring logic in place, we could even consider integrated schedule and hub optimization.


Where do carriers and system providers stand on the ladder?

My colleagues’ and my conversations with schedulers from different airlines around the globe indicated, that most carriers and scheduling system providers stand somewhere between level zero and level 2. At Lufthansa Systems, we aim to provide carriers with systems and services, which help stepping up the ladder.

Thank you for reading this blog post. We are looking forward to a fruitful discussion about passenger connection management in scheduling, both online in the comments section and offline at conferences or during workshops on airlines’ premise.