Modeling resource consumption in the US air transportation system via minimum-cost percolation

Minsuk Kim, C. Tyler Diggans, Filippo Radicchi · 2025

Read Modeling resource consumption in the US air transportation system via minimum-cost percolation by Minsuk Kim, C. Tyler Diggans, Filippo Radicchi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-04245.txt.

Introduction

Airlines burn fuel, crews, and slots as passengers flow through the flight network, and the system can jam when those resources run thin. Minsuk Kim and co-authors model that wearing-down process in this 2025 paper with a percolation process that always eats the cheapest path first.

Built on real United States flight schedules, the model shows a tipping point where the network stops serving demand. If rival airlines cooperated fully on carrying passengers, the system could serve demand about thirty percent more efficiently and shrug off cancellations better. The catch is that majors would share real market share, so the authors frame codesharing as the lever.

Catalog record

CREATORMinsuk Kim, C. Tyler Diggans, Filippo Radicchi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE68 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-04245

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Modeling resource consumption in the US air transportation system via minimum-cost percolation Minsuk Kim, C. Tyler Diggans, Filippo Radicchi Abstract We introduce a dynamic percolation model aimed at describing the consumption, and eventual exhaustion, of resources in transportation networks. In the model, rational agents progressively consume the edges of a network along demanded minimum-cost paths. As a result, the network undergoes a transition between a percolating phase where it can properly serve demand to a non-percolating phase where demand can no longer be supplied. We apply the model to a weighted, directed, temporal, multi-layer network representation of the air transportation system that can be generated using real schedules of commercial flights operated by US carriers. We study how cooperation among different carriers could improve the ability of the overall air transportation system in serving the demand of passengers, finding that unrestricted cooperation could lead to a 30 % 30\% efficiency increase compared to the non-cooperative scenario. Cooperation would require major airlines to share a significant portion of their market, but it would allow also for an increased robustness of the system against perturbations causing flight cancellations. Our findings underscore some key benefits that could emerge by simply promoting code-share arrangements among US airlines without altering their current cost of operation. Introduction The air transportation industry represents one of the major sectors of the US economy. In 2018 , civil aviation accounted for \$1.8 trillion in total activity, contributed 5.2\% to the gross domestic product, and supported more than 10.9 million jobs [ 1 ] . Still according to 2018 data, the industry enabled the transport of nearly 900 million passengers on flights covering more than 6 billion vehicle miles [ 2 ] . The impact of the air transport industry, like so many other areas of the national economy, was adversely affected by the COVID-19 pandemic, however, recent reports indicate that the industry went back to its pre-pandemic levels [ 3 ] . The infrastructure that sustains air transportation is composed of a large number of different elements, interacting and working together in a complex manner [ 4 ] . Individual components of the system have been studied for a long time, including flight scheduling [ 5 ] , fleet assignment [ 6 , 7 ] , aircraft maintenance routing [ 6 ] , crew scheduling [ 8 ] , management (e.g., revenue [ 9 , 10 ] , irregular operations [ 11 ] , airside operations [ 12 ] , and air traffic flow [ 13 ] ). The most studied of these components seems to be the air transportation network (ATN), sometimes referred to as the airline network. In the ATN, airports are nodes, and an edge between two nodes exists whenever there is a direct flight between the two airports. Many of the pioneering theoretical works on the structure of the ATN focused on comparisons between point-to-point and hub-and-spoke network configurations [ 14 , 15 , 16 ] . More recent works have provided network analysis of real-world data-generated ATNs, e.g., [ 17 , 18 ] . Real ATN structures share common attributes with many other types of real networks: they exhibit a broad degree distribution [ 19 ] , short average pair-wise distance [ 20 ] , large values of the clustering coefficient [ 20 ] , and a strongly modular structure [ 21 ] . The basic binary representation of edges in ATNs can be enriched in several ways. For example, it is possible to associate weights to the links depending on the number of passengers traveling between two airports [ 18 ] . Also, one can adopt a multiplex representation of the network [ 22 , 23 ] , with each layer representing a different air carrier serving different routes among a common set of airports [ 24 , 25 ] . Finally, the dynamic nature of flight schedules can be used to perform various types of temporal analyses [ 26 ] . For example, several studies have focused on the evolution of the structure of ATNs over time [ 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 ] . Dynamical changes of ATNs over short time scales have been considered in the definitions of time-dependent path lengths, correlations, and centrality metrics [ 37 , 38 ] . Further, time-series analyses of flight data can be used to study the propagation of flight delays in an ATN [ 39 ] . Data about the flow of people on ATNs are also useful in many applications, for instance, since the seminal work by Colizza et al. [ 40 ] , passenger data are systematically included in metapopulation models for epidemic spreading [ 41 , 42 , 43 , 44 ] . Further, passenger flow data on ATNs have been studied in papers focusing on Markovian processes with memory on networks, leading to suitably adapted centrality metrics and community detection algorithms [ 45 , 46 , 47 ] . A large body of literature focuses on the application of methods from percolation theory to ATNs, aiming at characterizing their robustness, where the robustness of a network is defined as its ability to remain connected as its nodes or edges are removed from the system; this being the straight forward adaptation of percolation to graphs [ 48 , 49 ] . In the context of ATNs, node removal is used to mimic failure or shutdown (e.g., due to weather) of airports, and the relative size of the largest connected component in the graph is used as a proxy of infrastructure function. These type of percolation models can be studied under either the isolated or the interdependent network frameworks, and network robustness is measured by looking at how the giant component of the network shrinks as a function of the fraction of nodes removed from the system. Different strategies for the removal of nodes can be used to model disruptive situations, but the most studied model is certainly the one of ordinary percolation where removed nodes are randomly selected at uniform [ 50 , 51 , 52 , 49 , 53 , 25 ] . In contrast, scenarios of maximal stress are simulated by optimizing the percolation process, thus selecting the nodes that lead to the quickest dismantling of the network [ 54 , 55 , 56 , 57 , 58 , 59 ] . In this paper, we introduce a percolation-based framework to study a rather different aspect of ATNs: their effectiveness in meeting the demand of passengers. As a pillar of the framework, we develop the minimum-cost-percolation (MCP) model, which is the natural extension of the shortest-path percolation model, recently considered in Ref. [ 60 ] , to weighted, directed, temporal, and multi-layer networks. The MCP model mimics how the resources of an infrastructural network are consumed, and eventually depleted, by demanding agents [ 48 ] . We apply the MCP model to data-generated representations of the US ATN. The framework enables analysis of how the US air transportation system would adapt and perform under various scenarios. For example, two hypothetical scenarios are considered: (i) a scenario of full cooperation where airlines cooperate by supplying

Read the complete text