POEM Passenger-Oriented Enhanced Metrics
Partners: University of Westminster (coordinator), Innaxis Research Institute
POEM was prized in the 2014 SESAR Awards ceremony as Outstanding Project during the 2014 World ATM Congress in Madrid (link)
Interview to Andrew Cook, Project coordinator available in SESAR page (here)
Central to the POEM project is the design of new performance metrics and their evaluation through a European network simulation model under novel flight and passenger prioritisation scenarios. A normative day with full passenger itineraries is simulated. Trade-offs between the (new) flight-centric and passenger-centric metrics are explored. The propagation of delay through the network is characterised using classical and complexity science techniques. The importance of using passenger-centric metrics in fully assessing system performance is repeatedly observed, since key changes are not expressed through any of the flight metrics.
Social and political priorities in Europe are continuing to shift in further support of passenger rights in transportation, as evidenced by high-level position documents such as ‘Flightpath 2050’ and the European Commission’s ‘Roadmap to a Single European Transport Area’. To better measure progress in reaching such objectives, passenger-centric metrics are needed. These are largely absent from the metrics currently in place to measure air transport system performance. We also need better models to understand the implications of policy measures and the trade-offs between them. Such models and metrics thus need to reflect the progress of corresponding planned regulatory review, particularly with regard to the underpinning regulatory instrument, Regulation 261. At the core of POEM is the design of new performance metrics and their evaluation through a European network simulation model under novel flight and passenger prioritisation scenarios. Key objectives were to explore the trade-offs between the (new) flight-centric and passenger-centric metrics and to characterise the propagation of delay through the network.
For the modelling, a baseline traffic day in September 2010 was selected as a busy day in a busy month, without evidence of exceptional delays, strikes or adverse weather. The busiest 199 European Civil Aviation Conference (ECAC) airports are modelled, having identified that these airports accounted for 97% of passengers and 93% of movements in 2010. Routes between the main airports of the EU-27 states and airports outside the EU-27 have been used as a proxy for determining the major flows between the ECAC area and the rest of the world. This process allowed the selection of 50 non-ECAC airports for inclusion of their passenger data.
The two principal datasets used to prepare the input data for the POEM model were IATA’s PaxIS passenger itineraries and EUROCONTROL’s PRISME traffic data. Extensive data cleaning of the source traffic data was required. Departure and arrival times were converted to local times (in addition to UTC) in order to define local 0400-0359 operational days at the modelled ECAC airports and to enable schedule matching (published in local times only), taking into account daylight saving time adjustments. There are approximately 30 000 flights in each day’s traffic and around 2.5 million passengers distributed among 150 000 distinct passenger routings. The assignment of passengers to individual flights, with cost characteristics and full itineraries, was a fundamental component of POEM, since the project explores new passenger-centric metrics. All the allocated connections were viable with respect to airline schedules and published minimum connecting times (MCTs).
The model rules include realistic simulations for missed passenger connections due to delays and cancellations (such as dynamic passenger reaccommodation onto aircraft with free seats, using detailed fleet and load factor data) and tail-tracked, aircraft wait rules. Core cost estimations in the model are with respect to delay costs to the airline, since it is these which drive airline behaviour. The model represents a normative day and the simulation results thus reflect schedule robustness (e.g. with respect to passenger reaccommodation).
Two airline case studies, including on-site visits and workshops, have focused on developing and testing specific aspects of the model rules as examined in an operational context. Furthermore, key parameters of the model were calibrated against independent data sources, including flight departure, arrival and reactionary delay, and also European load factors.
To measure the effect of increased perturbation, two disrupted days were derived from the baseline traffic day through the application of internal rules in the model. This allowed like-for-like comparisons between the disrupted days and the baseline day. One disrupted day imposed 1 extra minute on the average departure delay (making a new average of 14.9 minutes across all flights). The other disrupted day imposed just under 1% of additional cancellations on morning operations. Comparing the model outputs for the disrupted days showed them to be well modelled in that changes to the core metrics were as expected and reflected operational experience (e.g. with regard to relatively low impacts on flight punctuality metrics during periods of high cancellations).
Flight and passenger prioritisation scenarios were applied to the baseline traffic day and the two disrupted days. The prioritisation scenarios provided primary inputs into the network simulation model and were designed in parallel with the metrics, through which they were assessed in terms of their impact on performance. ANSP and AO modelled scenarios involve decision-making based on reasonable information for that agent to possess in either the current information environment or a future one (e.g. in the context of System-Wide Information Management).
Flight prioritisation scenarios operating during arrival management based simply on the numbers either of inbound passengers or on those with connecting onward flights, were ineffective in improving performance. Such performance was even slightly worse under high delay or increased cancellation rates. These effects were only discernible through the use of passenger-centric metrics.
A modelled policy-driven scenario represented the special case where we ran the model under putative conditions not driven by current airline or ATM objectives but which may nevertheless benefit the passenger. This scenario, rebooking disrupted passengers at airports based on minimising delays at their final destination, produced very weak effects when current airline interlining hierarchies were preserved. When these restrictions were relaxed, marked improvements in passenger arrival delay were observed, although at the expense of an increase in total delay costs per flight (due to passenger rebooking costs).
A prioritisation process assigning departure times based on cost minimisation (with due consideration of ATFM delays) markedly improved a number of passenger delay metrics and airline costs, the latter determined largely by reductions in passenger hard costs to the airline. One of the very few negative outcomes associated with this scenario was an increase of two percentage points in overall reactionary delay. Furthermore, in this trade-off, the additional reactionary delay was manifested through relatively few flights and was introduced purposefully by airlines through the cost model (i.e. waiting for late passengers) such that the overall cost to the airlines decreased. The addition of independent, cost-based arrival management apparently foiled these benefits due to lack of coordination between departures and arrivals.
Factor analysis was undertaken to ascertain the extent to which a derived (data reduction) technique would compare with a complexity science approach in analysing the results. Granger causality is held to be one of the only tests able to detect the presence of causal relationships between different time series. This method was used along with the eigenvector centralities of nodes (airports) to further explore delay propagation under the cost minimisation prioritisation scenario and to compare the passenger- and flight-centric network properties.
POEM final technical report: https://www.dropbox.com/s/9fi3cmydgokl90i/POEM%20Final%20Technical%20Report.pdf?dl=0
Cook A., Tanner G., Cristóbal S. and Zanin M., 2012. Passenger-Oriented Enhanced Metrics, in Schaefer, Dirk (ed) Proceedings of the 2nd SESAR Innovation Days (2012) EUROCONTROL. ISBN 978-2-87497-068-9. http://www.sesarinnovationdays.eu/2012/papers
Cook A., Tanner G. and Zanin M., 2013. Towards superior air transport performance metrics – imperatives and methods, Journal of Aerospace Operations, DOI 10.3233/AOP-130032. http://dx.doi.org/10.3233/AOP-130032 [URL only, as paid subscription required]
Cook A., Tanner G., Cristóbal S. and Zanin M., 2013. New perspectives for air transport performance, in Schaefer, Dirk (ed) Proceedings of the 3rd SESAR Innovation Days (2013) EUROCONTROL. ISBN 978-2-87497-074-0. http://www.sesarinnovationdays.eu/2013/papers
Cook A, Tanner G, Cristóbal S and Zanin M, 2015. Delay propagation – new metrics, new insights. Eleventh USA/Europe Air Traffic Management Research and Development Seminar (ATM2015), Lisbon. http://www.atmseminarus.org/11th-seminar/
Video of the Close-out Meeting
Impact, Future Research Lines and conclusions
The importance of using passenger-centric metrics in fully assessing system performance was repeatedly observed, since such changes were not expressed through any of the flight metrics at the common thresholds set. Most prioritisation scenarios performed similarly for the high cancellation and high delay days, demonstrating robustness in terms of their efficacy under increased disruption.
Remarkably, there was practically no relationship between the role of airports across the different flight and passenger layers, thus demonstrating the difference between the passenger- and flight-centric networks and the impact of the cost-based prioritisation process on these networks. This prioritisation scenario also produced a further trade-off: the propagation of delay was contained within smaller airport communities, but these communities were more susceptible to such propagation. A significant advance on earlier work has been the explicit estimation of reactionary costs (since each flight is individually modelled with its connectivity dependencies) and of the passenger costs of delay to the airline. In previous work these costs were estimated statistically. Passenger value of time has also been quantified as a function of delay at the final destination.
Smaller airports were significantly implicated in the propagation of delay through the network at a level which has hitherto not been commonly recognised. This is probably due to reduced delay recovery potential at such airports, for example through: flexible or expedited turnarounds; spare crew and aircraft resources; and, whether a given airport has sufficient connectivity and capacity to reaccommodate disrupted passengers.
Back-propagation was found to be an important characteristic of the persistence of delay propagation in the network. Paris Charles de Gaulle, Madrid Barajas, Frankfurt, London Heathrow, Zürich and Munich all demonstrated more than one hundred hours of back-propagated delay during the modelled (baseline) day. Most of the delay characterised in the network was indeed distributed between a relatively limited number of airports.
The full report (https://www.dropbox.com/s/9fi3cmydgokl90i/POEM%20Final%20Technical%20Report.pdf?dl=0) identifies further areas of research.