This project aims to address passengers' arrival patterns at airports through analyzing accessible datasets in order to provide an informed guidance for the resource planning of public airports in the U.S. The project will also assess if a new "baseline" arrival pattern should be established for airports to handle passenger traffic during the recover from the pandemic and post-pandemic.
Understanding passengers' arrival patterns could provide invaluable insights for effective resource planning of airports. While airlines have access to individual travelers' booking and historical check-in records to analyze and predict when travelers are likely to arrive, airports are presented with only obscure information to make such estimations for various key facilities, e.g. parking lots, check-in counters, security check lanes, retail spaces, airport lounges and passenger boarding areas. There have been some sparse studies on this topic, but most of them only focus on one airport, rely on one known probability distribution (Poisson or Gaussian) to fit historical data, and deal with an extremely brief period (a week or several days at most). To the best of our knowledge, there is no systematic study that addresses passengers' arrival patterns in a spatial-temporal manner, which is to analyze multiple airports for an extended period (a year to several years). Therefore, it is often impractical for airports to have a good grasp of factors that may be crucial to determine passengers' arrivals, i.e., spatial similarities or temporal trend. Failure to recognize the contribution of these latent conditions may mislead airports to make costly, inefficient, and possibly erroneous resource allocation decisions.
While the aviation sector is endeavoring hardly to bounce back from the crisis caused by the unprecedented COVID-19 pandemic, a practical challenge faced by airlines, airports and other service providers is: Are the data accumulated over the years before the pandemic still relevant to guide production during the recovery or post the pandemic? Through assessing and comparing existing datasets provided by TSA and DOT, this proposed project would be able to provide answers to this question if the pandemic has caused any structural change to how travelers plan their trips from home to airport.
This proposed study aims to address this knowledge gap for the airport industry by providing a systematic spatial-temporal analysis of passengers' arrival patterns at U.S. public airports. Using accessible datasets such as TSA security check throughput numbers, flight schedules and ticketing records, this study attempts to provide substantiated answers to the following questions:
1. What is the appropriate theoretical framework that can be used to describe, categorize, and compare passengers' arrival patterns at U.S. public airports?
2. What are the optimal probability distributions and associated parameters to fit passengers' arrivals of flights of different kinds departing in a 24-hour period?
3. Are passengers' arrival patterns consistent or variable across different periods in a year and across different airports in the U.S.?
4. Have travelers' changed how they plan their trips from home to airport due to the pandemic?
5. What tools and models can be used to project future passenger arrivals with accessible information by airports to achieve effective resource planning?
To achieve the aforementioned objectives, we anticipate the following steps will be taken within an estimated time frame of 18 months:
1. Pool and examine accessible datasets, including TSA passenger throughout, BTS T-100 & DB1B databanks;
2. Conduct descriptive analysis to build a theoretical framework to describe, categorize, and compare passengers' arrival patterns using measures such as skewness, kurtosis, standard deviation, mean, median, 'global' and 'local' arrival peaks;
3. Use different probability distributions, such as Poisson, Gaussian, Gamma, or Gumbel distribution, and flight schedule information to fit historical passenger data and to estimate distribution parameters;
4. Analyze and compare airports of different regions for spatial effects, i.e. spatial heterogeneity, spatial auto-correlation and spatial dependence;
5. Compare arrival patterns before and during the pandemic to assess if the pandemic has caused structural changes to how travelers plan their trips from home to airport;
6. Simulate passengers' arrivals with the model developed by this study and test the validity of the forecasting model using actual operational data; and
7. Draft the project report and present the final deliverables to the industry panel and ACRP.
We anticipate the completion of the project requires approximately 18 months including 3 months for review and revision of a draft final report. Considering the scale and complexity of the project, a budget of approximately $400,000 would be needed to cover the required direct, indirect, and other ancillary costs and expenses.
We have thoroughly reviewed existing literature on or related to the proposed topic. Within the ACRP community, the most relevant study is ACRP Report 23 - Airport Passenger-Related Processing Rates Guidebook (Cassidy & Navarrete, 2009), which primarily focuses on passenger processing rates at various airport facilities. Passenger processing is a subsequent step to passenger arrival in the passenger analysis flow chart, and both steps should be thoroughly investigated to provide accurate inputs for airport resource planning. Outside of ACRP publications, Robertson, Pendergraft, Johnson, and Silhert (2002) use check-in curves for passengers departing at different times in a day. Though they do not mention the distribution type, curves presented in their study resemble normal distribution. Another relevant study was conducted by Postorino, Mantecchini, Malandri, and Paganelli (2019). They estimate the arrival patterns of passengers traveling from Bologna Marconi Airport, a regional airport located in Northern Italy, on the flight level, and their model considers carriers' business types (full service & low cost), flight destinations (domestic, Schengen region and international), and departing time (morning and afternoon). The model selects Weibull distribution to describe passengers' arrival process. In a study specifically addressing Baltimore–Washington International (BWI) Airport, Reed (2003) fitted the passenger arrivals with different combinations and types of probability distributions at the airport facility level. His study confirms the significance of knowing the patterns of passengers' arrivals, and the distribution variation at different facilities in an airport. In a working paper by Mehri, Djemel, & Kammoun (2006), they use Poisson distribution to simulate the arrival of passengers at Monastir Habib Bourguiba International Airport in Tunisia. And in other modes of transportation, the arrival of passengers are estimated with Gumbel minimum distribution (Janosikova & Slavik, 2015).
From the brief review of aforementioned literature, it can be seen that existing studies on passenger arrivals focus mostly on one airport for a very brief period of time and use different types of probability distributions. There is no systematic study that addresses passengers' arrival patterns in a spatial-temporal manner, which is to analyze multiple airports for an extended period, let alone studies analyzing the impact of the pandemic to passengers' arrival patterns.
Cassidy, M., & Navarrete, J. (2009). Airport Passenger-Related Processing Rates Guidebook. In ACRP 03-02. https://doi.org/10.17226/22990
Janosikova, L., & Slavik, M. (2015). Modelling passengers' arrivals at public transport stops. Communications - Scientific Letters of the University of Zilina, 17(2), 30–34.
Mehri, H., Djemel, T., & Kammoun, H. (2006). Solving of waiting lines models in the airport using queuing theory model and linear programming the practice case: A.I.M.H.B.
Postorino, M. N., Mantecchini, L., Malandri, C., & Paganelli, F. (2019). Airport Passenger Arrival Process: Estimation of Earliness Arrival Functions. Transportation Research Procedia, 37(September 2018), 338–345. https://doi.org/10.1016/j.trpro.2018.12.201
Reed, R. (2003). Projecting future landside congestion delays at the BWI airport Baltimore: Final report. Retrieved from http://www.morgan.edu/Documents/ACADEMICS/CENTERS/NTC/Projecting_Reed_0304.pdf
Robertson, C. V, Pendergraft, D. R., Johnson, L. M., & Silhert, K. S. (2002). The role of modeling demand in process re-engineering. Proceedings of the Winter Simulation Conference, 1454–1458. https://doi.org/10.1109/WSC.2002.1166418