Review of optimization problems, models and methods for airline disruption management from 2010 to 2024

Yuzhen Hu; Sirui Wang; Song Zhang; Zhisheng Li; Yuzhen Hu; Sirui Wang; Song Zhang; Zhisheng Li

doi:10.48130/dts-0024-0022

This paper conducts a thorough review of airline disruption management between 2010 and 2024. Unlike previous review papers, the present paper analyses the research on airline disruption management in three ways. One is to perform a statistical analysis of these papers based on the journal distribution, number of papers by year, and types of recovery resources. The second is to categorize integrated recovery methods based on the degree of integration of the resources during the recovery process: the aircraft and crew, the aircraft and passengers, and all three resources. The last way is to study the research findings based on statistical analysis and perform future research direction identification in the areas of problems, models, and solution approaches. Further, with the increasing complexity of actual demands, integrated flight disruption recovery considering multiple factors such as aircraft, crew, and passengers has become a research hotspot in recent years. For further research, we can delve deeper into issues from both practical circumstances and theoretical extensions. At the model level, more detailed characterizations are needed, along with more efficient solution methods to accommodate increasingly complex problems.

Number

Name

Formula

Species

Ref.

2',5-Dihydroxy-3',6,7,8-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[17]

2',6-Dihydroxyflavone

C₁₅H₁₀O₄

S. baicalensis

[16]

3',4',5,5',7-Pentamethoxyflavone

C₂₀H₂₀O₇

S. baicalensis

[16]

4',5-Dihydroxy-3',5',6,7-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[22,27]

4',5-Dihydroxy-7-methoxyflavanone

C₁₆H₁₄O₅

S. baicalensis

[16]

5,4′-Dihydroxy-6,7,8,3′-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[17]

5,2′-Dihydroxy-6,7,8-trimethoxyflavone

C₁₈H₁₆O₇

S. baicalensis

[17]

5,2′-Dihydroxy-6,7,8,3′-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[17]

5,2′-Dihydroxy-7,8-dimethoxyflavone

C₁₇H₁₄O₆

S. baicalensis

[17]

5,2′-Dihydroxy-7,8,6′-trimethoxyflavone

C₁₈H₁₆O₇

S. baicalensis

[17]

5,6,7,3',4'-Pentahydroxyflavone-7-O-glucuronide

C₂₁H₂₀O₁₃

S. baicalensis

[13]

5,6,7,4'-Tetrahydroxydihydroflavone

C₁₅H₁₂O₆

S. baicalensis

[17]

5,6,7-Trihydroxy-4'-methoxyflavone

C₁₆H₁₂O₆

S. baicalensis

[15]

5,7-Dihydroxy-6-methoxyflavanon

C₁₆H₁₄O₅

S. baicalensis

[17]

5,7,4′-Trihydroxy-6-methoxyflavanone

C₁₆H₁₄O₆

S. baicalensis

[22]

5,7,4'-Trihydroxyflavanone

C₁₅H₁₂O₆

S. baicalensis

[13]

(2S)-5,7,8,4'-Tetrahydroxyflavanone 7-O-β-D-glucuronopyranoside

C₂₁H₂₀O₁₂

S. baicalensis

[21]

(2S)-5,6,7,4'-Tetrahydroxyflavanone 7-O-β-D-glucuronopyranoside

C₂₁H₂₀O₁₂

S. baicalensis

[21]

6-Hydroxyluteolin-7-O-glucuronide

C₂₁H₁₈O₁₃

S. baicalensis

[13]

7-Methoxychrysin

C₁₆H₁₄O₄

S. baicalensis

[16]

Apigenin

C₁₅H₁₀O₅

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[17,21−25]

Apigenin-4'-glucopyranside

C₂₁H₂₀O₁₀

S. baicalensis

[17]

Apigenin-6-C-glucoside-8-C-arabinoside

C₂₆H₂₈O₁₄

S. baicalensis

[13]

Apigenin-7-O-β-D-glucopyranside

C₂₁H₂₀O₁₀

S. baicalensis

[17]

Apigenin-7-O-β-D-glucuronide

C₂₁H₁₈O₁₁

S. baicalensis, S. amoena,
S. scordifolia

[13,23,24]

Apigenin-7-O-methylglucuronide

C₂₂H₂₀O₁₁

S. baicalensis

[28]

Baicalein

C₁₅H₁₀O₅

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[22−25]

Baicalein-7-O-D-glucopyranside

C₂₁H₂₀O₁₀

S. baicalensis

[22]

Baicalin

C₂₁H₁₈O₁₁

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[16,23−25]

Carthamidin

C₁₅H₁₂O₆

S. baicalensis

[13,20]

Carthamidin-7-O-β-D-glucuronide

C₂₁H₂₀O₁₂

S. baicalensis

[22]

Chrysin

C₁₅H₁₀O₄

S. baicalensis, S. amoena,
S. scordifolia

[13,21−24]

Chrysin-7-O-β-D-glucuronide

C₂₁H₂₀O₉

S. baicalensis, S. amoena

[23,29]

Dihydrobaicalin

C₂₁H₂₀O₁₁

S. baicalensis

[22,28]

Dihydrooroxylin A

C₂₁H₂₀O₁₁

S. baicalensis

[13]

Genkwanin

C₁₆H₁₂O₅

S. baicalensis

[16]

Isocarthamidin

C₁₅H₁₀O₆

S. baicalensis

[20,28]

Isocarthamidin-7-O-β-D-glucuronide

C₂₁H₂₀O₁₂

S. baicalensis

[28]

Isoschaftside

C₂₆H₂₈O₁₄

S. baicalensis

[13,16]

Isoscutellarein

C₁₅H₁₀O₆

S. baicalensis

[21,30]

Isoscutellarein 8-O-β-D-glucuronide

C₂₁H₁₈O₁₂

S. baicalensis

[21]

Kaempferol 3-O-β-D-glucopyranoside

C₂₁H₂₀O₁₁

S. baicalensis

[13,28]

Luteolin

C₁₅H₁₀O₆

S. baicalensis

[22,30]

Norwogonin-7-O-glucuronide

C₂₁H₁₈O₁₂

S. baicalensis, S. amoena

[13,17,23]

Oroxylin A

C₁₆H₁₂O₅

S. baicalensis, S. amoena

[17,23]

Oroxylin A-7-O-D-glucopyranside

C₂₂H₂₂O₁₀

S. baicalensis

[22,28]

Oroxylin A-7-O-β-D-glucuronide

C₂₂H₂₀O₁₁

S. baicalensis, S. amoena

[13,23]

Pinocembrin

C₁₆H₁₂O₅

S. baicalensis

[13]

Pinocembrin-7-O-glucuronide

C₂₁H₂₀O₁₁

S. baicalensis

[13,16,22]

Salvigenin

C₁₈H₁₆O₆

S. baicalensis

[21]

Scutellarein

C₁₅H₁₀O₆

S. baicalensis

[28]

Scutellarin

C₂₁H₁₈O₁₂

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[23−25,29]

Wogonin

C₁₆H₁₂O₅

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[21−25]

Wogonoside

C₂₂H₂₀O₁₁

S. baicalensis, S. viscidula

[13,25]

HTML

Search scope and methodology

Definition of airline disruption management

Currently, there is no unified and clear definition of disruption management in the academic field. However, most scholars have reached a common view. According to Yu & Qi^[3], disruption management is a methodology that addresses disruptions in real-time. At the beginning of a business cycle, an optimal or near-optimal operational plan is obtained by using certain optimization models and solution schemes. When such an operational plan is executed, disruptions may occur from time to time that are caused by internal and external uncertain factors. As a result, the original operational plan may not remain optimal or even feasible. Consequently, we need to dynamically revise the original plan and obtain a new plan that reflects the constraints and objectives of the evolved environment while minimizing the negative impact of the disruption. Clausen et al.^[8] limit disruption management to the basic goal of returning to the original plan as soon as possible to minimize system disturbance.

Although disruption management is currently not applied to the airline industry alone, the definition of disruption management indeed began with airline civil aviation. According to Yu & Qi^[3], airline disruption management refers to disruption management for applications in airline operations. Airline disruption management is commonly implemented by an airline operation center (AOC). When a disruption occurs, the airline first needs to assess the disruption and then make revisions to their flight planning and the schedules of related resources.

In Yu & Qi^[3], the recovery resources primarily include the aircraft and crew. Clausen et al.^[8] report that passenger itineraries have also been included in recovery resources since Lettovsky^[13] and Bratu & Barnhart^[14]. Therefore, airline disruption management currently involves aircraft, crew, and integrated recovery of three resources, including passenger itineraries.

Aircraft recovery problem: When flight schedules are disrupted, the problem tries to minimize the division from the original schedules in terms of aircraft and flight operations. Generally, where some aircraft can be rerouted (aircraft swap), some disrupted flights can be given a new departure and arrival time (flight delay), and some disrupted flights can be cancelled (flight cancellation).

Crew recovery problem: when the flight schedules are recovered, the problem attempts to make a crew rescheduling decision to minimize the division related to crew schedules. Sometimes, flights can be delayed or cancelled as well.

Integrated recovery problem: During the recovery process, the problem aims to consider two or three recovery problems, such as aircraft recovery, crew recovery, and passenger itinerary recovery, in an integrated model.

Search methodology
After understanding the core concepts of airline disruption management, we focused on the task of finding published journal papers related to airline disruption management. To do this, a systematic approach was used, as described in what follows. The search began with the databases of ISI's Web of Science, ScienceDirect, and the Wiley Online Library. The keywords 'airline', 'aircraft', 'crew', and 'passenger' combined with 'perturbance', 'disruption', 'disrupted', 'irregular', 'recovery', 'rerouting', 'cancellation', 'delay', 'rescheduling', and 'reassignment' were searched in the titles of journal papers published in English. Book chapters and working papers were excluded. The period of the search was limited to 2008 and thereafter. To be more precise, if a journal paper related to the present topic was included in Clausen et al.^[8], it was excluded from the present study. Therefore, most of the papers were published between 2010 and 2024.

After searching for the defined keywords in the mentioned journals, 1384 papers were obtained as the initial results. We were certain that many of these papers would not fit within the project scope. Hence, we continued extracting the papers most related to the topic of airline disruption management.

The titles and abstracts of the papers were checked and papers eliminated with titles completely different from the present research domain. For example, papers with titles or abstracts related to disruption management in railway transportation, public transportation, vehicle routing problems, and aircraft manufacturing were eliminted. Some papers focusing on airline delay prediction, delay propagation, airline gate assignment, air traffic flow management, and airport terminal optimization phases were also filtered out. By applying the above filters, 69 papers were identified for the present survey.

Literature breakdown and analysis

According to the classification that was applied in "Statistics on various resources of airline operations" and "Trends in the number of studies by year and type of recovery resources", we will describe the studies published during the years 2010−2024 in detail. Aircraft recovery refers to the literature on aircraft recovery, Crew recovery presents the literature on crew recovery, and Integrated recovery describes the literature on integrated recovery. Therefore, the first two sections focus on the recovery of single resources. The last section concentrates on the integrated recovery of multiple resources.

Key information related to the airline disruption management problem proposed in the present reviewed papers is summarized in Tables 1−6, with Tables 1 & 2 for aircraft recovery, Tables 3 & 4 for crew recovery, and Tables 5 & 6 for integrated recovery. The information in Tables 1, 3 & 5 includes the following common attributes: 'multiobjective', 'model'. The attributes of 'objective', 'single', and 'multi' mean the paper develops a mathematical model with single or multiple objectives, respectively. The attributes of 'model', 'LP' means linear programming, 'IP' means integer programming, 'CP' means constraint programming, and 'MICQ' means mixed-integer conic quadratic. The information in Tables 2, 4 & 6 includes the following common attributes: 'solution methodology', 'data', 'data dimension', and 'solution time (sec)'. The attributes of 'data', 'RL' means the paper uses real-world instances to test the performance of the proposed model and solution methodology. The attribute of 'data dimension', aims to display the largest instance scale in the computation experiments.

Table 1. The first part of the proposed method overview for aircraft recovery.

	Network	Cancel	Delay	Aircraft swap	Cruise speed	Fleet	Objective	Model	Objectives
Chen et al.^[29]	NA	NO	YES	YES	NO	Single	Multi	nonlinear	Flight non-connection, duty swap, delay time, delay number, delay number over 30 min
Liu et al.^[31]	NA	NO	YES	YES	NO	Multi	Single	Nonlinear	Operation, delay, passenger cost
Babić et al.^[32]	NA	YES	YES	YES	NO	Multi	Single	Nonlinear	Max revenue minus operational and disturbance costs
Liu et al.^[30]	NA	NO	YES	YES	NO	Single	Multi	Nonlinear	Delay time, duty swap, variance of flight delay time, number of delayed flight, number of long-delayed flight
Gao et al.^[36]	NA	NO	YES	NO	NO	Single	Multi	Nonlinear	Weighted flight delay time
Mou et al.^[37]	NA	YES	YES	YES	NO	Multi	Multi	Nonlinear	Delay minutes, delay, and cancellation cost
Aktürk et al.^[38]	CN	NO	YES	YES	YES	Multi	Single	Conic IP	Delay, deadhead, additional fuel and carbon emission, passengers spilled cost
Vos et al.^[17]	TN	YES	YES	YES	NO	Single	Single	LP	Operation, delay, cancellation, aircraft ground cost
Guimarans et al.^[33]	NA	NO	YES	YES	NO	Single	Single	CP	Delay time
Xu et al.^[22]	TBN	YES	YES	YES	NO	Single	Single	IP	Delay, cancellation cost
Hu et al.^[24]	CN	YES	YES	YES	NO	Multi	Multi	IP	Min delay and cancellation cost, maximal flight delay time, min the number of swapped aircraft
Wu et al.^[25]	CN	YES	YES	YES	NO	Single	Single	IP	Delay time
Wu et al.^[45]	CN	YES	YES	YES	NO	Multi	Single	IP	Delay, cancellation cost
Wu et al.^[46]	CN	YES	YES	YES	NO	Multi	Single	IP	Delay, cancellation cost
Zhang^[27]	CN	YES	YES	YES	NO	Single	Single	IP	Cancellation, aircraft assignment, terminal balance violation cost
Bouarfa et al.^[40]	NA	NA	NA	NA	NO	NA	NA	NA	NA
Khaled et al.^[41]	NA	YES	YES	YES	NO	Single	Multi	IP	Operation recovery cost, number of flight changed, number of impacted airports
Liang et al.^[28]	CN	YES	YES	YES	NO	Multi	Single	IP	Flight cancellation, route cost
Lin et al.^[34]	NA	NO	YES	YES	NO	Single	Single	Nonlinear	Delay time
Wang et al.^[42]	NA	YES	YES	YES	NO	Multi	NA	NA	NA
Şafak et al.^[39]	NA	YES	YES	YES	YES	Multi	Single	MICQ	Max revenue minus fuel burn, passenger spilled, flight arrival tardiness, crew service, aircraft swap cost
Vink et al.^[18]	TN	YES	YES	YES	NO	Multi	Single	Mixed IP	Operation and disruption cost
Pei et al.^[43]	NA	NO	YES	YES	NO	Multi	NA	NA	NA
Lee et al.^[19]	TN	YES	YES	YES	YES	Multi	Single	Nonlinear	Expected recovery cost
Ji et al.^[47]	NA	YES	YES	NO	NO	Multi	Single	Nonlinear	Delay time
Huang et al.^[20]	TN	YES	YES	YES	NO	Multi	Single	IP	Retimed, cancelled or assigned cost
Şi̇mşek et al.^[48]	NA	YES	YES	YES	YES	Multi	Single	Nonlinear	Fuel consumption and CO₂ emission cost
Zang et al.^[21]	TN	YES	YES	YES	NO	Multi	Single	Nonlinear	Delay, cancellation cost

Table 2. The second part of the proposed method overview for aircraft recovery.

	Delay cost	Aircraft maintenance	Airport congestion	Solution method	Data	Data dimension		Solution time (s)
	Delay cost	Aircraft maintenance	Airport congestion	Solution method	Data	AC	Flight	Solution time (s)
Chen et al.^[29]	NA	NO	NO	Hybrid multi-objective genetic algorithm	RL	7	70	600
Liu et al.^[31]	Linear	NO	NO	Hybrid particle swarm optimization heuristic	RL	NA	34	236
Babić et al.^[32]	Linear	YES	NO	Heuristic	JAT Airways	9	47	NA
Liu et al.^[30]	NA	NO	NO	Hybrid multi-objective genetic algorithm	RL	7	84	450
Gao et al.^[36]	NA	NO	YES	Polynomial algorithm	Generated	4	8	NA
Mou et al.^[37]	Linear	NO	NO	Polynomial algorithm	Generated	5	10	NA
Aktürk et al.^[38]	Nonlinear	YES	YES	CPLEX	An airline in the US	60	207	248.4
Vos et al.^[17]	Nonlinear	YES	NO	Selection algorithm	Kenya Airways	43	NA	600
Guimarans et al.^[33]	NA	NO	NO	Large neighbourhood search	RL	48	294	205.514
Xu et al.^[22]	Linear	NO	NO	CPLEX	RL	60	254	949.7
Hu et al.^[24]	Linear	NO	NO	Heuristic based on ε-constraints and neighbourhood search	A major Chinese airline	104	401	1200
Wu et al.^[45]	NA	NO	NO	CPLEX	RL	12	140	7.02
Wu et al.^[25]	Linear	NO	NO	CPLEX	RL	30	215	NA
Wu et al.^[46]	Linear	NO	NO	CPLEX	RL	27	162	286.6
Zhang^[27]	Linear	YES	NO	Heuristic + CPLEX	RL	44	638	150
Bouarfa et al.^[40]	NA	NA	NA	Multi-agent system approach	NA	NA	NA	NA
Khaled et al.^[41]	NA	YES	NO	ε-Constraints + CPLEX	RL	11	111	30
Liang et al.^[28]	Linear	YES	YES	Column generation + CPLEX	RL	44	638	356.13
Lin et al.^[34]	NA	NO	NO	Fast variable neighbourhood search	RL	12	70	0.3
Wang et al.^[42]	NA	YES	NO	Simulation	RL	628	5071	NA
Şafak et al.^[39]	nonlinear	NO	NO	CPLEX	United Airlines	81	300	8074
Vink et al.^[18]	Nonlinear	YES	NO	Selection algorithm	An airline in the US	100	600	44
Pei et al.^[43]	NA	YES	NO	AHP + algorithm	A Chinese airline	29	92	NA
Lee et al.^[19]	Nonlinear	NO	YES	Look-ahead approximation and sample average approximation	RL	NA	852	300
Ji et al.^[47]	NA	NO	NO	Build-in flight feasibility verification algorithm	RL	NA	300	14.69
Huang et al.^[20]	NA	NO	NO	Iterative copy generation algorithm	Nine RL scenarios	4-162	789	0.5-855
Şi̇mşek et al.^[48]	NA	NO	Yes	Aircraft Swapping and Search Algorithm	Bureau of Transportation Statistics (2021)	NA	NA	NA
Zang et al.^[21]	Linear	YES	YES	Decision-decomposition-based algorithm	Four Chinese airlines	733	2,877	23.82

Table 3. The first part of the overview of proposed methods for crew recovery.

	Network	Fixed f	Cancel f	Delay f	Crew swap	Objective	Objectives
AhmadBeygi et al.^[49]	SP	YES	NO	NO	YES	Single	Pairing cost minus flight dual contribution
Chang et al.^[50]	NA	NO	YES	NO	YES	Multi	Number of deadhead trip, unconnected flight, schedule changes and affected crews
Fang et al.^[51]	NA	YES	NO	NO	YES	Multi	Deviation cost of flight time and duty time
Liu et al.^[52]	SC	NO	YES	NO	YES	Single	Number of uncovered flights
Luo et al.^[59]	SP	YES	NO	NO	YES	Single	Pairing cost
Bayliss et al.^[55]	NA	NO	NO	YES	NO	Single	Expected crew delay
Chen et al.^[53]	NA	YES	NO	NO	YES	Multi	Number of crew changes, number of duty changes, maximal duty changes, largest changed flight time, derivation of the changed duties, derivation of changed flight time
Bayliss et al.^[56]	NA	NO	YES	NO	YES	Single	Cancellation
Bayliss et al.^[57]	NA	NO	YES	YES	YES	Single	Delay and cancellation
Wen et al.^[60]	DN	YES	NO	NO	NO	Single	Robustness-related cost
Herekoğlu et al.^[61]	NA	NO	YES	YES	YES	Multi	Assignment cost, swapping cost, deadheading costs, cancellation costs, delaying costs, penalties
Zhong et al.^[62]	NA	NO	YES	NO	YES	Multi	Deviation of duty time, total recovery cost

Table 4. The second part of the overview of proposed methods for crew recovery.

	Solution method	Data	Data dimension				Solution time (s)
	Solution method	Data	Aircraft	Crew	Flight	Recovery period (d)	Solution time (s)
AhmadBeygi et al.^[49]	CPELX	A major US hub-and-spoke carrier	NA	NA	329	1	2065
Chang et al.^[50]	Genetic algorithm	An international Taiwanese airline	NA	70	628	18	600
Fang et al.^[51]	Hybrid simulated annealing	Domestic airlines	NA	87	342	NA	195.04
Liu et al.^[52]	Simulated annealing	A major airline in the US	NA	482	1069		236
Luo et al.^[59]	Primal-dual sub-problem simplex method in a branch-and-price framework	Three airlines	NA	NA	NA	30	NA
Bayliss et al.^[55]	Greedy heuristic	Generated	37	120	300	1	1400
Chen et al.^[53]	Evolutionary algorithm	A short-haul airline in Taiwan		270	1048	14	1080
Bayliss et al.^[56]	Heuristic + CPLEX	Generated	37	148	243	3	3600
Bayliss et al.^[57]	heuristic + CPLEX	Generated	74	209	566	2	600
Wen et al.^[60]	Customized column generation based solution algorithm	An airline in Hong Kong	NA	NA	98	3	NA
Herekoğlu et al.^[61]	Column generation-based solution approach	A major European airline company	400	13500	1873	3	5438
Zhong et al.^[62]	Ad-hoc particle swarm optimization -based optimizer	Vari Flight Company	NA	NA	166	4	NA

Table 5. The first part of the overview of proposed methods for integrated recovery.

	Aircraft	Crew	Passenger	Net-work	Cancel	Delay	Aircraft swap	Fleet	Cruise speed	Objective	Model	Objectives
Eggenberg et al.^[75]	YES	NO	YES	CN	YES	YES	YES	Multi	NO	Single	IP	Passengers delay, cancellation cost
Jafari et al.^[68]	YES	NO	YES	NA	YES	YES	YES	Multi	NO	Single	Nonlinear	Aircraft assignment, flight delay, flight cancellation, passenger disruption cost
Bisaillon et al.^[11]	YES	NO	YES	NA	YES	YES	YES	Multi	NO	Single	NA	Aircraft and flight operation, passenger disruption cost, constraints violation
Artigues et al.^[9]	YES	NO	YES	NA	NA	NA	NA	NA	NA	NA	NA	NA
Mansi et al.^[78]	YES	NO	YES	TBN	YES	YES	YES	Multi	NO	Single	NA	Aircraft and flight operation, passenger disruption cost, constraints violation
Petersen et al.^[86]	YES	YES	YES	CN	YES	YES	YES	Multi	NO	Single	IP	Flight delay and cancellation, aircraft assignment, crew pairing and deadheading, passenger delay and unassignment cost
Jozefowiez et al.^[76]	YES	NO	YES	NA	YES	YES	YES	Multi	NO	Single	NA	Aircraft and flight operation, passenger disruption cost, constraints violation
Brunner ^[85]	YES	YES	YES	NA	YES	YES	YES	NA	NO	Single	Mixed IP	Flight arrival and departure delay, flight cancellation, crews' and passengers' misconnection cost
Sinclair et al.^[80]	YES	NO	YES	TN	YES	YES	YES	Multi	NO	Single	NA	Aircraft and flight operation, passenger disruption cost, constraints violation
Hu et al.^[35]	YES	NO	YES	TBN	YES	YES	YES	Multi	NO	Single	IP	Flight delay, passenger transiting, passenger refunding cost
Maher^[87]	YES	YES	YES	CN	YES	YES	YES	Multi	NO	Single	IP	Flight delay and cancellation, crew deadheading, and passenger reassignment cost
Zhang et al.^[65]	YES	YES	NO	TN	YES	YES	YES	Multi	NO	Single	IP	Flight delay and cancellation, crew misconnection, aircraft and crew swap cost
Arıkan et al.^[88]	YES	NO	YES	CN	NO	YES	YES	Multi	YES	Single	Conic IP	Aircraft delay, passengers delay, spill cost, swap cost, and increased fuel cost
Hu et al.^[81]	YES	NO	YES	CN	YES	YES	YES	Multi	NO	Single	IP	Passengers' delay, reassignment, refund cost
Maher^[66]	YES	YES	NO	CN	YES	YES	YES	Single	NO	Single	IP	Flight delay and cancellation, reserve crew, crew duty and deadhead cost
Sinclair et al.^[79]	YES	NO	YES	TN	YES	YES	YES	Multi	NO	Single	NA	Aircraft and flight operation, passenger disruption cost, constraints violation
Zhang et al.^[77]	YES	NO	YES	TN	YES	YES	YES	Multi	NO	Single	NA	Aircraft and flight operation, passenger disruption cost, constraints violation
Arıkan et al.^[88]	YES	YES	YES	CN	YES	YES	YES	Multi	YES	Single	conic IP	Flight cancellation, aircraft ferrying, crew deadheading, passenger delay, reallocation and refund, additional fuel cost, constraints violation
Marla et al.^[72]	YES	NO	YES	TN	YES	YES	YES	Multi	YES	Single	IP	Flight delay and cancellation, aircraft swap, passenger disruption, additional fuel cost
Santos et al.^[73]	YES	NO	YES	CN	NO	YES	YES	Multi	NO	Single	Mixed IP	Additional operation, passenger disruption cost
McCarty et al.^[64]	NO	NO	YES	CN	NO	YES	YES	NA	NO	Single	Mixed IP	Passengers' expected delay cost
Yang et al.^[82]	YES	NO	YES	CN	YES	YES	YES	Multi	NO	Multi	IP	Airline recovery, passenger uitility cost
Yeti̇moğlu et al.^[74]	YES	NO	YES	CN	YES	YES	NO	NA	YES	Single	Nonlinear	Revenue - fuel and CO₂ emission cost - overnight passenger cost - spilled passenger cost.
Evler et al.^[89]	YES	YES	YES	NA	YES	YES	YES	Multi	NO	Single	Mixed IP	Operating and delay cost, cancel cost, connection cost, cost of assigning turnaround recovery options
Xu et al.^[90]	YES	YES	YES	CN	YES	YES	YES	Multi	NO	Single	Mixed IP	Cost of flight cancellation, delay, crew deadhead and unassigned passengers
Zhao et al.^[83]	YES	NO	YES	TN	YES	YES	YES	Single	NO	Single	Mixed IP	Disruption cost, passenger delay cost, curfew violation cost, cancel cost.
Cadarso et al.^[92]	YES	NO	YES	TN	YES	YES	YES	Multi	YES	Single	Mixed IP	Flight operating cost, extra fuel consumption cost, flight delay cost, passenger reaccommodation cost, passenger delay cost, crew cost, penalizes aircraft changes
Ding et al.^[91]	YES	YES	YES	TN	YES	YES	YES	NA	YES	Single	Mixed IP	Flight cancellation cost, passenger delay cost, external link cost, additional fuel cost and following schedule cost
Chen et al.^[84]	YES	NO	YES	NA	YES	YES	YES	NA	NO	Multi	IP	The total delay cost of each flight, the sum of the passenger delay time of each flight

Table 6. The second part of the overview of proposed methods for integrated recovery.

	Delay cost	Aircraft maintenance	Airport congestion	Solution method	Data	Data dimension					Solution time (s)
	Delay cost	Aircraft maintenance	Airport congestion	Solution method	Data	AC	Flight	Crew	Passenger	Passenger itinerary	Solution time (s)
Eggenberg et al.^[75]	Linear	YES	NO	Column generation	Thomas Cook Airlines	100	760	NA	30000	NA	3603
Jafari et al.^[28]	Linear	NO	NO	Lingo	Swedish domestic airline	13	100	NA	2236	8	NA
Bisaillon et al.^[11]	Linear	YES	YES	Large neighbourhood search	2009 ROADEF Challenge	618	2178	NA	NA	29151	600
Artigues et al.^[28]	Linear	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA
Mansi et al.^[78]	Linear	YES	YES	Two stage heuristic	2009 ROADEF Challenge	618	2178	NA	NA	29151	600
Petersen et al.^[86]	Linear	YES	NO	Bender decomposition, column generation	Hub-and-spoke airline in the US	NA	800	NA	NA	NA	2407
Jozefowiez et al.^[76]	Linear	YES	YES	Three stage	2009 ROADEF Challenge	618	2178	NA	NA	29151	600
Brunner^[85]	Nonlinear	NO	NO	CPLEX	American Airlines	NA	71	26	651	651	NA
Sinclair et al.^[80]	Linear	YES	YES	Large neighbourhood search	2009 ROADEF Challenge	618	2178	NA	NA	29151	600
Hu et al.^[35]	Nonlinear	NO	NO	CPLEX	A major airline in China	188	628	NA	NA	NA	106
Maher^[87]	Linear	NO	NO	Column and row generation	RL	48	262	79	28492	NA	1800
Zhang et al.^[65]	Linear	YES	NO	Iteration heuristic + CPLEX	Regional airline in the US	70	351	134	NA	NA	72.418
Arıkan et al.^[88]	Linear	NO	NO	CPLEX	Airline in US	NA	1429	NA	NA	NA	142
Hu et al.^[81]	Linear	NO	NO	GRASP	A major airline in China	87	340	NA	NA	NA	600
Maher^[66]	Linear	NO	NO	Column and row generation	RL	123	441	182	NA	NA	1200
Sinclair et al.^[79]	Linear	YES	YES	Column generation	2009 ROADEF Challenge	618	2178	NA	NA	29151	1385
Zhang et al.^[77]	Linear	YES	YES	Three stage	2009 ROADEF Challenge	618	2178	NA	NA	29151	600
Arıkan et al.^[88]	Nonlinear	NO	NO	CPLEX	A major U.S. airline	402	1254	NA	NA	NA	1212.4
Marla et al.^[72]	Linear	NO	NO	Xpress	A major European airline	NA	250	NA	NA	NA	120
Santos et al.^[73]	Nonlinear	NO	YES	CPLEX	Kenya airways	45	140	NA	10000	NA	3600
McCarty et al.^[64]	Linear	NO	NO	Benders Decomposition + CPELX	Delta Airlines	NA	NA	NA	200	15	93.9
Yang et al.^[82]	Linear	NO	NO	Genetic algorithm	A major airline in China	59	209	NA	24860	NA	11
Yeti̇moğlu et al.^[74]	NA	NO	NO	Novel math-heuristic algorithm	A major airline in America	53	208	NA	NA	2033	NA
Evler et al.^[89]	Linear	YES	NO	Rolling horizon algorithm	Frankfurt airport	17	85	NA	NA	NA	45
Xu et al.^[90]	Linear	NO	NO	Branch-and-cut solution method, large neighborhood search heuristic	One main legacy carrier in the US	NA	230	172	NA	NA	937.71
Zhao et al.^[83]	Linear	YES	NO	Two-stage algorithm, rolling horizon approach	GE Aviation	73	207	NA	594	NA	NA
Cadarso et al.^[92]	Linear	YES	YES	An original solution approach	A IBERIA airline	19	1074	NA	1204	32	600
Ding et al.^[91]	Linear	YES	YES	Variable neighborhood search algorithm	Generated	50	NA	NA	NA	NA	0.359
Chen et al.^[84]	Nonlinear	NO	NO	Genetic algorithm-II	A Chinese airline in Fuzhou airport	NA	NA	NA	1818	NA	14.57

In Table 1, 'network' refers to the network type for aircraft recovery model construction: connection network (CN), time space network (TN), and time band network (TBN). The detailed network representation has been issued in Clausen et al.^[8]. 'Cancel', 'delay', 'aircraft swap', 'fleet', and 'cruise speed' refer to whether the paper considers the recovery options of flight cancellation, flight delay, aircraft swapping, swapping between multiple fleet types, and cruise speed control, respectively. In Table 2, 'delay cost' means, in the proposed model, the linear or nonlinear penalty for the departure or arrival delay time of each flight. 'Aircraft maintenance' and 'airport congestion' represent whether the paper considers the constraints of aircraft maintenance and airport capacity in the proposed model.

In Table 3, 'network' represents the network applied for the crew recovery model: set partition network (SP), set covering network (SC), and duty-based network (DN). 'Fixed f', 'cancel f', and 'delay f' refers to whether the crew recovery model includes the decision variables of flight delay and flight cancellation. 'Crew swap' refers to whether the paper considers the recovery options of crew swaps. In Table 5, the first three attributes, 'aircraft', 'crew', and 'passenger', represent the combination of several resources in the recovery problems.

Aircraft recovery

The aircraft recovery problem (ARP) has been acknowledged as the earliest research branch compared to crew and integrated recovery in the field of airline disruption management. In addition to the smaller number and simpler rules of aircraft compared to other resources, such as crews and passengers^[8], the core reason may be that aircraft are equipment, while crews and passengers are human beings. We can seek only a profit maximum or cost minimum in the ARP and can ignore the complicated psychology of human beings. This type of problem is exactly what operations research and optimization methods are good at addressing. As a classical optimization problem, there are typically two considerations in the ARP: a model formulation that accurately defines the detailed boundary of the problem and a solution methodology that shows how to obtain the solution to the problem. For this reason, categorizing aircraft recovery papers based on the model formulation and solution methodology is meaningful. Formally, these classifications are as given below. Tables 1 & 2 show the key information of the ARP literature in chronological order.

(1) Solution methodology based on network. Studies in this group are similar to most of the papers introduced in Clausen et al.^[8]. The problem is formulated with networks (connection networks, time-line networks, or time-band networks) and then solved by an exact algorithm and the CPLEX solver.

(2) Meta-heuristic studies. Studies in this category seek more efficient meta-heuristics to obtain satisfactory solutions to the ARP. Note that papers in this category give less attention to model formulation. Some papers do not even include an accurate model.

(3) Polynomial studies. Studies in this category aim to analyze the optimization characteristics of the ARP in certain special situations and to design polynomial algorithms for optimal or approximately optimal solutions.

(4) Other studies. Unlike the above groups, studies in this group consider some specific approaches to analyzing the ARP.

Solution methodology based on network
To date, the majority of publications use integer programming solution methods to solve the aircraft recovery problem. Integer programming models are generally formulated based on various networks. Related studies by the classification of network representation will be introduced, which have been described in Clausen et al.^[8] on network graphs and networks, i.e., time-space network, time-band network, and connection network.

Time-space network
The time-space network was first introduced by Yan & Yang^[16] for flight rescheduling and aircraft recovery problems. The main advantages of the network lie in the clear graphical representation of the flight network by the time-space node and relative arcs. Moreover, the idea of discretizing flight delay time can ingeniously transform the aircraft recovery problem to a network optimization problem with boundary constraints. However, the disadvantage of the time-space network is the majority of the decision variables due to the copies of flight delay arcs. It will lead to lower efficiency in computation time. Therefore, some studies focus on the dynamic recovery methodology based on a time-space network. It can not only reduce the recovery scale but also fit the practical dynamic character of airline disruption. For example, Vos et al.^[17] develop a dynamic modeling framework based on a parallel time-space network. Vink et al.^[18] presented an exact mathematical model for the proposed dynamic aircraft recovery problem. Lee et al.^[19] presented a stochastic reactive and proactive disruption management model that combines a stochastic queuing model of airport congestion to minimize the expected recovery costs. Huang et al.^[20] proposed a copy evaluation method and develop a solution approach to the Aircraft Recovery Problem by incorporating the method within an iterative process of copy generation and filtration. Zang et al.^[21] examined aircraft recovery problem from the viewpoint of balancing supply and demand across the airport time-space network through aircraft rotations between airports.

Time-band network
Compared to the time-space network, the time-band network plays an important role in reducing the number of flight arcs and corresponding decision variables. It divides the recovery time periods into several discrete time intervals. Of course, the core drawback of the time-band network is that the flight delay cost will be overestimated or underestimated, and even sometimes the rescheduled flight network is not feasible due to time connection failure unless the optimization result is fine-tuned. Xu & Han^[22] extend the time band network for aircraft recovery in a hub-and-spoke network and use a simplex group cycle approach to control the fight disruption scope and depth.

Connection network
In contrast to the few studies that construct integer models based on time-space or time-band networks, more studies prefer to use connection networks, which are most popular for airline scheduling, to formulate the aircraft recovery problem. The major advantage of this network is that it is more suitable for solving large-scale problems by combining column generation and bender decomposition algorithms. One disadvantage is that it is often used for single-objective optimization in airline disruption management. Zhu et al.^[23] establish a two-stage stochastic programming model based on a connection network, where the aircraft recovery time is not fixed. Hu et al.^[24] describe multiobjective mathematical programming and solve the proposed problem using heuristics instead of exact algorithms. Wu et al.^[25] develop a distributed computation algorithm^[26] to generate feasible flight routes for solving the integer programming model based on a connection network. Zhang^[27] introduces a two-stage heuristic to design an aircraft recovery network before using a set partition model. Liang et al.^[28] add the constraints of aircraft planning maintenance and the airport slot capacity to the connection network and then solve it using a column generation heuristic.

Meta-heuristics
As an airline is more concerned with recovery time efficiency than with the optimality of the recovery solution, it is more attractive to obtain a satisfactory recovery solution in a short CPU time than to obtain a near-optimal solution over a long time for real applications in the airline industry. Therefore, heuristics have often been applied in recent years for solving aircraft recovery problems, such as the genetic algorithm in Chen et al.^[29] and Liu et al.^[30], hybrid particle swarm in Liu et al.^[31], and neighborhood search algorithm in Babić et al.^[32], Guimarans et al.^[33] and Lin & Wang^[34].

Polynomial algorithms
Although the aircraft recovery problem has been proven NP-hard^[35], some studies still prefer to analyze optimization characteristics and design polynomial algorithms for some special cases of the problem. Gao et al.^[36] focus on flight rescheduling under large-scale flight delays considering flight delays and flight cancellations rather than flight swaps between different aircraft routings. Then, a polynomial algorithm is designed to obtain the optimal solution for the flight rescheduling problem. Mou & Zhou^[37] developed an uncertain programming model with chance constraints, where the uncertainty distribution of the aircraft delay time is given by experts. A recovery solution method is designed based on a stepwise-delay algorithm and the Hungarian algorithm. Hu et al.^[24] solved the aircraft recovery problem with the hierarchical-objectives programming model by a polynomial algorithm.

Other studies on aircraft recovery
In addition to common concerns of aircraft recovery, some studies focus on special solution approaches to analyze or solve the proposed problem. For example, Aktürk et al.^[38] and Şafak et al.^[39] propose a mathematical model, especially for considering cruise speed control in the aircraft recovery problem. Bouarfa et al.^[40] evaluated the performance of a multiagent system for disruption management in an airline operation control (AOC) department, and Khaled et al.^[41] established a multicriteria recovery framework based on a tail assignment model. Wang et al.^[42] applied a simulation method to analyze the performance of the aircraft recovery process. Pei et al.^[43] presented a data-driven method to solve the flight rescheduling problem based on AHP and heuristics.

Research findings
(1) More studies have started to focus on multiple-fleet aircraft recovery. Approximately 15 of 25 papers (more than 50%) studied multiple fleet aircraft recovery operations after 2010, while the rate was less than 23% before 2010 according to Clausen et al.^[8]. This makes the recovery model more complex due to aircraft swapping between different fleets.

(2) More studies have started to focus on multiple-objective programming to formulate the aircraft recovery problem. The model formulation in approximately six of 25 papers was developed with multiple objectives, while the number was no more than three of 31 papers before 2010. The reason is apparent. Before 2010, most solution methodologies for aircraft recovery were derived from the airline planning phase. Most model formulations were constructed for airline planning with a single objective. It can easily be solved with corresponding optimization techniques. However, due to the complexities of the real-time environment, it is difficult to use a single-objective model to accurately describe many needs of airlines. Therefore, some scholars have aimed to construct multiple-objective formulations to describe aircraft recovery problems.

(3) Almost all studies propose the common consumption that the aircraft recovery model does not consider the recovery measures of using surplus aircraft and ferrying aircraft. The assumption is realistic to some extent. Because aircraft are expensive equipment and rented instead of being bought by airlines, it is a waste of aircraft resources for airlines to leave aircraft idle or ferry spare aircraft. This is the reason why using surplus aircraft and ferrying aircraft are the last choices in actual airline recovery operations.

(4) Another common assumption in the airline recovery literature is that the flight delay cost is a linear function of delay time. This is slightly inconsistent with airline recovery practice since flight delay costs tend to be a nonlinear function of delay time^[44]. Some papers, such as
Aktürk et al.^[38], Vos et al.^[17], Şafak et al.^[39], and Vink et al.^[18], attempt to make changes to solve the actual airline operation problem.

(5) Airport capacity and aircraft dynamic fuel costs have begun to be considered in recovery models. This promotes disruption management more practically.

(6) More studies have started to solve aircraft recovery problems using heuristics. These heuristics include meta-heuristics and polynomial algorithms. Understandably, airline disruption management is an NP-hard problem^[35], and it pursues efficient computational time rather than optimization accuracy alone in real-time. Therefore, they have to use heuristics for practical efficient application.

(7) Aircraft recovery has begun to be considered in combinatorial optimization theory. Before 2010, more studies focused on the similarity of airline disruption management and airline scheduling. However, there are clear differences between them. The airline planning stage includes the scheduling of all resources, while airline disruption management is focused on minor changes, which reduces the problem scale and provides the possibility of optimization through theoretical analysis. Moreover, aircraft recovery is the most classical optimization problem in the field of airline disruption management. Optimization analysis of airline disruption management commonly starts from the aircraft recovery problem.

Crew recovery
Most papers continue previous research topics such as crew recovery after the occurrence of an airline disruption. These papers include AhmadBeygi et al.^[49], Chang^[50], Fang & Xia^[51],
Liu et al.^[52], and Chen & Chou^[53]. Other studies by Bayliss et al.^[54−57] focus on reserve crew scheduling to reduce flight delays and cancellations under various disruption situations. Tables 3 & 4 show the key information of the crew recovery literature in chronological order.

Classic crew recovery
AhmadBeygi et al.^[49] aimed to develop a crew pairing generator to support integrated airline planning, robust planning and automated recovery in addition to modeling nonlinear constraints and cost functions in crew scheduling. Multiple solutions for crew recovery are usually obtained from the genetic algorithm used by Chang^[50] and Chen & Chou^[53]. Another meta-heuristic for generating multiple solutions is the hybrid simulated annealing heuristic described in Fang & Xia^[51].

Reserve crew scheduling
Faced with airline disruptions, especially crew absence or delays, the application of reserve crews can promote the rapid recovery of airline schedules. To the best of our knowledge, airline reserve crew scheduling was first studied by Dillon & Kontogiorgis^[58]. In recent decades, reserve crew scheduling for airline recovery has mainly been studied in Bayliss et al.^[54−57]. Bayliss et al.^[54] focus on assigning the duty start time of reserve crews in case of the disruption of crew absence. Bayliss et al.^[55] extend the model for crew delay recovery. Bayliss et al.^[56] focus on reserve crew scheduling for both crew absence and delay. Bayliss et al.^[57] extend the influence of reserve crew scheduling on the probability of fight cancellation. Luo et al.^[59] introduce a development and improvement of the proposed models and solution approaches of Sabre for crew augmentation, initially for airline safety reasons during the crew pairing process. Wen et al.^[60] proposed a customized column generation based solution algorithm and on this basis, Herekoğlu & Kabak^[61] utilized a customized deep learning model to provide recovery actions as inputs. Zhong et al.^[62] designed an ad-hoc particle swarm optimization-based optimizer to solve Integrated Aircraft and Crew Recovery with Multi-objective and Priority efficiently.

Research findings
(1) In the field of airline disruption management, the crew is the other type of resource parallel to the aircraft. Crew recovery is also important for airline disruption management. Crew recovery research includes 13 papers published 14 years before 2010 and 12 papers published in the last 14 years. However, there have been research achievements in terms of problem extension and solution methodology.

(2) Although the scheduling of spare aircraft is uncommon due to the higher cost of calling a spare aircraft, studies of crew augmentation have begun to appear. After all, it is not as expensive to hire a spare crew as to hire an aircraft.

(3) In the existing literature for crew recovery, one common assumption is that the flight schedules have been recovered before the crew recovery process. Therefore, flights are generally fixed during the crew recovery process. The assumption is consistent with reality. Once the flight has been rescheduled in actual operation, only a few flights can be cancelled or delayed along with crew rescheduling. These assumptions are also considered by some papers.

(4) Although the number of papers focusing on crew recovery has not increased, more studies try to apply crew recovery problems with heuristics such as simulated annealing and genetic algorithms. There is only one paper^[63] that uses a combination method of column generation and a genetic algorithm according to Clausen et al.^[8], while there are no fewer than five such papers (nearly 50%) after 2010 according to our statistics.

Integrated recovery

Due to the complexity of airline disruption management in real situations, it is necessary to focus on the integrated recovery of multiple resources, which generally include aircraft, crew, and passengers. The integrated recovery models that are most constructed based on connection networks, aircraft routing and crew pairing are given in Clausen et al.^[8]. As another important recovery resource, passenger itinerary recovery is generally formulated as follows. Let F be the set of scheduled flights and I be the set of scheduled passenger itineraries. For each flight f∈F, the number of scheduled passengers NPass_f in flight f and the set of aircraft routes R(f) covering flight f are given. For each passenger itinerary i∈I, the ticket refunding cost refundc_i and the number of scheduled passengers NumP_i in itinerary i are given. We define the integer variable r_i as the number of passengers who refund their tickets and the binary variable z_i as 1 if itinerary i is disrupted and 0 otherwise. R(f) denotes the set of aircraft routes covering flight i. For each r∈R(f), Cap_r denotes the capacity of aircraft covering routing r, and the flight delay cost of flight f in route r is denoted by delayc_fr. We define the variable x_r, which is equal to 1 if aircraft route r is implemented and 0 otherwise. The objective (see Eqn (1)) aims to minimize the passenger delay and ticket refunding cost. The constraints promise that passengers can refund tickets only if their itineraries are disrupted (see Eqn (2)), and the reassignment for passengers cannot exceed the seat capacities of the aircraft (see Eqn (3)).

$\min\sum\limits_{f\in F}^{ }\sum\limits_{r\in R(f)}^{ }N Pass_f\cdot delayc_{fr}\cdot x_r+\sum\limits_{i\in I}^{ }re fundc_i\cdot r_i$

(1)

${\rm{S.t.}}\; {r_i} \leqslant Num{P_i} \cdot {z_i}\begin{array}{*{20}{c}} {}&{} \end{array}\forall i \in I$

(2)

$\sum\limits_{i \in I(f)} {{r_i}} \geqslant N Pas{s_f} - \sum\limits_{r \in R(f)} {Ca{p_r} \cdot {x_r}} \begin{array}{*{20}{c}} {}&{} \end{array}\forall f \in F$

(3)

${x_r} \in \{ 0,1\} \begin{array}{*{20}{c}} {}&{} \end{array}\forall r \in R(f),f \in F$

(4)

${z_i} \in \{ 0,1\} \begin{array}{*{20}{c}} {}&{} \end{array}\forall i \in I$

(5)

${r_i} \in \{ 0,1,...,N um{P_i}\} \begin{array}{*{20}{c}} {}&{} \end{array}\forall i \in I$

(6)

The model is commonly part of the integrated recovery model, instead of being applied alone for airline recovery. Additionally, there will be slight differences in various articles for recovery variants. For example, multifleet aircraft with different seat capacities and the cost of passengers transiting between different itineraries will be considered in some studies. It is noted that passenger itinerary recovery is seldom considered separately and is often one component of integrated recovery. Only McCarty & Cohn^[64] focus on the problem of performing passenger reallocation before itinerary misconnections occur. This is a proactive approach for passenger rerouting when facing uncertain itinerary delays rather than for post-disruption recovery.

According to the integration degree of the resources during the recovery process, integrated recovery publications can be analyzed in the following three categories: the integrated recovery of aircraft and crews, the integrated recovery of aircraft and passengers, and the integrated recovery of all three resources. Tables 5 & 6 show the key information of the integrated recovery literature in chronological order.

Integrated recovery of aircraft and crew

Zhang & Lau^[65] is one of the few studies that focus on the integrated recovery of aircraft and crew. The authors first generate routes of aircraft and crews based on a time-space network and then propose a novel two-stage heuristic to solve the integrated recovery problem. Maher^[66] represents a column-and-row generation algorithm framework to solve the integrated recovery of aircraft and crew problems based on Muter et al.^[67].

Integrated recovery of aircraft and passengers
The majority of publications formulate the integrated recovery of aircraft and passengers following the multicommodity network flow model. However, there are various solution methodologies for the problem, such as directly using commercial optimizers, applying large-scale optimization methods, and designing meta-heuristics. We classify the publications according to the solution methods used in these papers.

Direct method using a commercial optimizer
Jafari & Zegordi^[68] was the first attempt to establish a single objective model to represent and solve the integrated recovery of aircraft and passengers simultaneously based on aircraft rotations and passenger itineraries as an extension of the aircraft recovery formulation in Abdelghany et al.^[69]. Based on the time-band network in Bard et al.^[70], Hu et al.^[35] designed a reduced time-band network to study the integrated recovery of aircraft and passengers with itineraries along a single flight. Both Arıkan et al.^[71] and Marla et al.^[72] add the action of speed change in addition to common recovery actions in the integrated recovery of aircraft and passengers. Another study on the passenger recovery problem under airline delay management was reported by Santos et al.^[73] in daily flight operation at a hub airport under capacity limitations of the runway, taxiway and bay. Yeti̇moğlu & Aktürk^[74] worked on integrated networks at which aircraft routings and passenger itineraries are superimposed, and calculated the actual profit and cancellation cost by evaluating each passenger itinerary while considering the seat capacity limitations.

Large-scale optimization method
Due to the large number of decision variables and constraints in the integer programming formulation of integrated recovery problems for aircraft and passengers, most publications solve them in several stages.

Some studies obtain aircraft routings in the first stage and then a passenger reassignment solution in the second stage. In Eggenberg et al.^[75], the first stage focuses on the aircraft recovery problem with maintenance scheduling using column generation, and the second stage concentrates on the passenger recovery problem by computing a minimum cost flow problem with a seat capacity alternatively. Jozefowiez et al.^[76] and Zhang et al.^[77] both introduce a three-stage solution method for the integrated recovery problem of aircraft and passengers. In Jozefowiez et al.^[76], the first two stages focus on aircraft recovery and passenger recovery successively, and the third stage attempts to transport more passengers to their destinations by creating and inserting new flight sequences into available aircraft routings. In Zhang et al.^[77], however, the three stages mainly deal with aircraft recovery, flight rescheduling, and passenger recovery successively.

Other studies solve such problems by following a heuristics process; i.e., an initial solution is obtained in the first stage, and an improved optimal solution is derived in the second stage. In the first stage of Mansi et al.^[78], an initial feasible solution is first obtained using mixed-integer programming models and a repairing heuristic. In the second stage, an algorithm is developed based on an oscillation strategy of alternating between constructive and destructive phases to improve the integrated recovery solution. In Sinclair et al.^[79], the initial feasible solution is obtained in the first stage by a large neighborhood search heuristic. The second stage focuses on applying mixed-integer programming for integrated recovery based on a connection network.

Meta-heuristics
The large neighborhood search heuristic is commonly used for airline recovery, especially for aircraft and passenger recovery. Bisaillon et al.^[11] represent the integrated recovery problem of aircraft and passengers with an oval description and then introduce the large neighborhood search heuristic to solve the problem. Sinclair et al.^[80] improve the heuristic introduced in Bisaillon et al.^[11] by adding some steps to the proposed three phases. Then, Sinclair et al.^[79] use the solution result of the large neighborhood search (LNS) heuristic in Sinclair et al.^[80] as the initial variables of the column generation algorithm.

Based on a neighborhood heuristic, Hu et al.^[81] propose the greedy randomized adaptive search procedure (GRASP) heuristic by combining a semi greedy heuristic and the neighborhood search heuristic. Another optimization method based on a multiobjective genetic algorithm was created by Yang & Hu^[82] for a novel integrated recovery problem considering passenger preferences. Zhao et al.^[83] produced a two-stage algorithm compared with a rolling horizon approach. Chen et al.^[84] developed an adaptive non-dominated sorting genetic algorithm-II based on dominant strengths (ANSGA2-DS).

Integrated recovery of all three resources
Due to the complexity of practical irregular operations in the airline industry, there are fewer studies on the fully integrated recovery of aircraft, crews, and passengers. Most papers solve the integrated problem using various methods to limit recovery scopes. For example, Brunner^[85] limits recovery in a terminal airport when a ground delay program (GDP) is issued.
Petersen et al.^[86] and Maher^[87] solve the integrated recovery problem in several stages by column and row generation methods. Arıkan et al.^[88] represent a novel flight network to limit the size of the recovery, which has been commonly used in
Aktürk et al.^[38] and Arıkan et al.^[71]. Few studies use heuristic algorithms like the rolling horizon algorithm^[89], large neighborhood search algorithm^[90], and variable neighborhood searches^[91].

Research findings
(1) Competition has greatly promoted the development of the research field. According to statistical analysis, it was found that most papers (approximately 20 publications of a total of 30 papers) focus on the integrated recovery of aircraft and passengers. The large number of studies is a benefit of the 2009 ROADEF Challenge. Approximately six papers tested their algorithms using instances from the challenge.

(2) In integrated recovery, including passenger reassignment, it is generally assumed that passengers obey the arrangement of airlines and that passengers can only stay in the original itineraries unless passengers cannot arrive at their destinations through the original itineraries. This is somewhat inconsistent with the actual recovery operation for passengers. If their original itineraries are delayed in reality, the passengers have the right to choose their itineraries: endorsing other itineraries or refunding tickets. However, it is difficult to obtain data about passengers' choices under itinerary disruption. We can still make some attempts in the field of operations research. This will promote the development of behavioral operations research in practical airline operations.

(3) Most studies assume that the passenger delay cost is a linear function of delay time, which does not fit reality. Some papers, such as Brunner^[85], Hu et al.^[35], Arıkan et al.^[88], and Santos et al.^[73], think that the passenger delay cost is based on the number of passengers or passenger classes.

(4) Methods that promote passenger recovery have begun to be proposed by combining various traveling modes. Although there are only a few publications, it is a positive attempt.

(5) In terms of the solution algorithm, heuristics have been dominant. Even if the model is directly solved by an optimizer solver, heuristic algorithms are often used. The reasons are twofold. First, due to the large scale of integrated recovery, it is difficult to achieve the required timeliness by relying directly on a commercial optimizer. Second, the goal of disruption management is to pursue the recovery of resources in a short period of time.

Future research directions

Through a qualitative and quantitative analysis of the above topics, the trends of airline disruption management are summarized below. This can provide some direction for future research, especially for new scholars in the field of airline disruption management.

(1) From the perspective of problems, two extremes may occur. One is closer to reality, and the other is related to combinatorial optimization theory.

① Airline disruption management is used to solve practical problems. Some research trends for practical application may occur in the following areas as well as in aircraft recovery problems from single to multiple fleets: i) The disruption period may be unknown and undetermined. ii) Airline disruption management should not only make flight schedules return to the original plan as soon as possible but also use fewer resources, such as swapping aircraft and crews, and consider the environmental effects of aircraft fuel. iii) Airline disruption management is commonly related to multiple agents, such as airlines and passengers. Both of their interests should be considered. iv) Future research may consider more limitations during the recovery process, such as air traffic flow constraints and airport congestion. v) Some recovery measures can also be frequently considered in the near future, such as cruise speed control, combining various traveling modes, and starting the recovery process before a disruption occurs. Although the measures are only involved in a few papers, they are commonly used in airline practical operations.

② Although airline disruption management is an NP-hard problem, under assumptions based on practical operations, some simplified problems can also have good optimal properties in terms of combinatorial optimization. Of course, this is often the task of scholars in the fields of applied mathematics and computation theory. However, anything is possible, especially in the case of interdisciplinary work becoming increasingly common.

(2) The construction of the model tends to become more complex. Combined with the analysis of practical operations of airline disruption management, models can be extended in the following directions.

① As airline disruption management aims at returning to the original plan as soon as possible, it is the goal of the recovery model to be easy to solve and to obtain a solution efficiently. It is still popular to construct an effective and simple model to exactly describe the airline disruption management process. For example, a multistage model can be established according to the peak period of flight operations in hub airports.

② For a problem with an uncertain disruption period, stochastic programming must be developed. This may exceed the research scope of traditional deterministic optimization models for airline disruption.

③ Whether aiming to complete flight recovery as soon as possible using few resources or to balance multiple agents' interests, multiple objectives should be employed. Traditional models of airline disruption management are only concerned about the recovery cost. Even aircraft rerouting must be converted into costs to simplify the model. However, the current trend is to acknowledge the necessity of the diversity of the model objectives. This is closely related to the essence of airline disruption management, i.e., minimizing the negative impacts of disruption for flights, aircraft, crew, passengers, and even hub airports.

④ Airline disruption management is closely related to multiple agents. Aircraft and crews are internal resources of the airline, while passengers are the subject of the airlines' service. Therefore, airline disruption management mainly involves the interests of the airline and passengers. However, airlines are mainly concerned with short-term economic interests, while passengers pay more attention to their psychological feelings under itinerary disruptions. Therefore, studies of passengers' feelings should be added to airline disruption management solution methodologies. A multi-agent approach, which has been proven valid for managing airline disruption management by De Castro^[93] and Bouarfa et al.^[40], could be applied to solve the model.

⑤ Although airline disruption management is concerned with post-disruption recovery, subsequent disruptions may continue to occur. Therefore, the robust optimization of the recovery model should be given more attention. Even the combination of robust planning before the occurrence of disturbance and disruption management after the occurrence of the disturbance should also be considered in the future.

(3) Due to the increasing complexity of the problem and model of current airline disruption management, the efficiency and effectiveness of the solution approach in both computation time and recovery period are important. After all, returning to the original plan as soon as possible is the main goal of airline disruption management.

① Heuristics have to be used more often than before in solving the problem. Multiple objective models and nonlinear models are currently often used to describe airline disruption management. Perhaps some nonlinear solvers can be capable of solving nonlinear models. However, in the short term, researchers have to resort to the combination of heuristic and commercial optimization solvers for solving the practical multiple objective model. Moreover, it is still difficult but important to promote solving multiple objective models and nonlinear models efficiently by using commercial optimization solvers.

② The actual recovery rules can be combined into the airline disruption management solution approach in future research. This will improve the efficiency. For example, changes to the flight peak period in hub airports and cruise speed have been added to the solution methodology design.

③ Machine learning can also be used for airline disruption management in several ways. First, as airline disruption management is commonly solved based on networks and in several stages, the traditional solution approach can be replaced with machine learning, in particular, deep learning based on neural networks and reinforcement learning based on a multistage recovery process. Second, new solution approaches can be generated by combining traditional combinatorial optimization algorithms and machine learning algorithms. For example, in integrated recovery, passengers' behavior and feelings can be analyzed by machine learning and data analysis methods, and then the analysis results can be used in the subsequent recovery process. Additionally, a database of historical disruption scenarios and corresponding recovery solutions can be constructed for current efficient disruption management, where the management of historical scenarios and response to current disruption requires machine learning technology. In the future, we may see some corresponding research achievements.

Number

Name

Formula

Species

Ref.

2',5-Dihydroxy-3',6,7,8-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[17]

2',6-Dihydroxyflavone

C₁₅H₁₀O₄

S. baicalensis

[16]

3',4',5,5',7-Pentamethoxyflavone

C₂₀H₂₀O₇

S. baicalensis

[16]

4',5-Dihydroxy-3',5',6,7-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[22,27]

4',5-Dihydroxy-7-methoxyflavanone

C₁₆H₁₄O₅

S. baicalensis

[16]

5,4′-Dihydroxy-6,7,8,3′-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[17]

5,2′-Dihydroxy-6,7,8-trimethoxyflavone

C₁₈H₁₆O₇

S. baicalensis

[17]

5,2′-Dihydroxy-6,7,8,3′-tetramethoxyflavone

C₁₉H₁₈O₈

S. baicalensis

[17]

5,2′-Dihydroxy-7,8-dimethoxyflavone

C₁₇H₁₄O₆

S. baicalensis

[17]

5,2′-Dihydroxy-7,8,6′-trimethoxyflavone

C₁₈H₁₆O₇

S. baicalensis

[17]

5,6,7,3',4'-Pentahydroxyflavone-7-O-glucuronide

C₂₁H₂₀O₁₃

S. baicalensis

[13]

5,6,7,4'-Tetrahydroxydihydroflavone

C₁₅H₁₂O₆

S. baicalensis

[17]

5,6,7-Trihydroxy-4'-methoxyflavone

C₁₆H₁₂O₆

S. baicalensis

[15]

5,7-Dihydroxy-6-methoxyflavanon

C₁₆H₁₄O₅

S. baicalensis

[17]

5,7,4′-Trihydroxy-6-methoxyflavanone

C₁₆H₁₄O₆

S. baicalensis

[22]

5,7,4'-Trihydroxyflavanone

C₁₅H₁₂O₆

S. baicalensis

[13]

(2S)-5,7,8,4'-Tetrahydroxyflavanone 7-O-β-D-glucuronopyranoside

C₂₁H₂₀O₁₂

S. baicalensis

[21]

(2S)-5,6,7,4'-Tetrahydroxyflavanone 7-O-β-D-glucuronopyranoside

C₂₁H₂₀O₁₂

S. baicalensis

[21]

6-Hydroxyluteolin-7-O-glucuronide

C₂₁H₁₈O₁₃

S. baicalensis

[13]

7-Methoxychrysin

C₁₆H₁₄O₄

S. baicalensis

[16]

Apigenin

C₁₅H₁₀O₅

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[17,21−25]

Apigenin-4'-glucopyranside

C₂₁H₂₀O₁₀

S. baicalensis

[17]

Apigenin-6-C-glucoside-8-C-arabinoside

C₂₆H₂₈O₁₄

S. baicalensis

[13]

Apigenin-7-O-β-D-glucopyranside

C₂₁H₂₀O₁₀

S. baicalensis

[17]

Apigenin-7-O-β-D-glucuronide

C₂₁H₁₈O₁₁

S. baicalensis, S. amoena,
S. scordifolia

[13,23,24]

Apigenin-7-O-methylglucuronide

C₂₂H₂₀O₁₁

S. baicalensis

[28]

Baicalein

C₁₅H₁₀O₅

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[22−25]

Baicalein-7-O-D-glucopyranside

C₂₁H₂₀O₁₀

S. baicalensis

[22]

Baicalin

C₂₁H₁₈O₁₁

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[16,23−25]

Carthamidin

C₁₅H₁₂O₆

S. baicalensis

[13,20]

Carthamidin-7-O-β-D-glucuronide

C₂₁H₂₀O₁₂

S. baicalensis

[22]

Chrysin

C₁₅H₁₀O₄

S. baicalensis, S. amoena,
S. scordifolia

[13,21−24]

Chrysin-7-O-β-D-glucuronide

C₂₁H₂₀O₉

S. baicalensis, S. amoena

[23,29]

Dihydrobaicalin

C₂₁H₂₀O₁₁

S. baicalensis

[22,28]

Dihydrooroxylin A

C₂₁H₂₀O₁₁

S. baicalensis

[13]

Genkwanin

C₁₆H₁₂O₅

S. baicalensis

[16]

Isocarthamidin

C₁₅H₁₀O₆

S. baicalensis

[20,28]

Isocarthamidin-7-O-β-D-glucuronide

C₂₁H₂₀O₁₂

S. baicalensis

[28]

Isoschaftside

C₂₆H₂₈O₁₄

S. baicalensis

[13,16]

Isoscutellarein

C₁₅H₁₀O₆

S. baicalensis

[21,30]

Isoscutellarein 8-O-β-D-glucuronide

C₂₁H₁₈O₁₂

S. baicalensis

[21]

Kaempferol 3-O-β-D-glucopyranoside

C₂₁H₂₀O₁₁

S. baicalensis

[13,28]

Luteolin

C₁₅H₁₀O₆

S. baicalensis

[22,30]

Norwogonin-7-O-glucuronide

C₂₁H₁₈O₁₂

S. baicalensis, S. amoena

[13,17,23]

Oroxylin A

C₁₆H₁₂O₅

S. baicalensis, S. amoena

[17,23]

Oroxylin A-7-O-D-glucopyranside

C₂₂H₂₂O₁₀

S. baicalensis

[22,28]

Oroxylin A-7-O-β-D-glucuronide

C₂₂H₂₀O₁₁

S. baicalensis, S. amoena

[13,23]

Pinocembrin

C₁₆H₁₂O₅

S. baicalensis

[13]

Pinocembrin-7-O-glucuronide

C₂₁H₂₀O₁₁

S. baicalensis

[13,16,22]

Salvigenin

C₁₈H₁₆O₆

S. baicalensis

[21]

Scutellarein

C₁₅H₁₀O₆

S. baicalensis

[28]

Scutellarin

C₂₁H₁₈O₁₂

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[23−25,29]

Wogonin

C₁₆H₁₂O₅

S. baicalensis, S. amoena,
S. scordifolia, S. viscidula

[21−25]

Wogonoside

C₂₂H₂₀O₁₁

S. baicalensis, S. viscidula

[13,25]

[1]	Eltoukhy AEE, Chan FTS, Chung SH. 2017. Airline schedule planning: a review and future directions. Industrial Management & Data Systems 117:1201−43 doi: 10.1108/imds-09-2016-0358 CrossRef Google Scholar
[2]	Barnhart C, Belobaba P, Odoni AR. 2003. Applications of operations research in the air transport industry. Transportation Science 37:368−91 doi: 10.1287/trsc.37.4.368.23276 CrossRef Google Scholar
[3]	Yu G, Qi X. 2004. Disruption management: framework, models and applications. Singapore: World Scientific. doi: 10.1142/5632
[4]	Teodorović D, Guberinić S. 1984. Optimal dispatching strategy on an airline network after a schedule perturbation. European Journal of Operational Research 15:178−82 doi: 10.1016/0377-2217(84)90207-8 CrossRef Google Scholar
[5]	Etschmaier MM, Mathaisel DFX. 1985. Airline scheduling: An overview. Transportation Science 19:127−38 doi: 10.1287/trsc.19.2.127 CrossRef Google Scholar
[6]	Ball M, Barnhart C, Nemhauser G, Odoni A. 2007. Air transportation: Irregular operations and control. Handbooks in operations research and management science, eds. Barnhart C, Laporte G. vol. 14. Netherlands: Elsevier. pp. 1−67. doi: 10.1016/S0927-0507(06)14001-3
[7]	Kohl N, Larsen A, Larsen J, Ross A, Tiourine S. 2007. Airline disruption management—perspectives, experiences and outlook. Journal of Air Transport Management 13:149−62 doi: 10.1016/j.jairtraman.2007.01.001 CrossRef Google Scholar
[8]	Clausen J, Larsen A, Larsen J, Rezanova NJ. 2010. Disruption management in the airline industry—concepts, models and methods. Computers & Operations Research 37:809−21 doi: 10.1016/j.cor.2009.03.027 CrossRef Google Scholar
[9]	Artigues C, Bourreau E, Afsar HM, Briant O, Boudia M. 2012. Disruption management for commercial airlines: methods and results for the ROADEF 2009 Challenge. European Journal of Industrial Engineering 6:669−89 doi: 10.1504/EJIE.2012.051072 CrossRef Google Scholar
[10]	Visentini MS, Borenstein D, Li JQ, Mirchandani PB. 2014. Review of real-time vehicle schedule recovery methods in transportation services. Journal of Scheduling 17:541−67 doi: 10.1007/s10951-013-0339-8 CrossRef Google Scholar
[11]	Bisaillon S, Cordeau JF, Laporte G, Pasin F. 2011. A large neighbourhood search heuristic for the aircraft and passenger recovery problem. 4OR 9:139−57 doi: 10.1007/s10288-010-0145-5 CrossRef Google Scholar
[12]	Hassan LK, Santos BF, Vink J. 2021. Airline disruption management: A literature review and practical challenges. Computers & Operations Research 127:105137 doi: 10.1016/j.cor.2020.105137 CrossRef Google Scholar
[13]	Lettovsky L. 1997. Airline operations recovery: an optimization approach. Thesis. Georgia Institute of Technology, USA.
[14]	Bratu S, Barnhart C. 2006. Flight operations recovery: New approaches considering passenger recovery. Journal of Scheduling 9:279−98 doi: 10.1007/s10951-006-6781-0 CrossRef Google Scholar
[15]	InCites. 2018. Incites journal citations reports. https://jcr.incites.thomsonreuters.com/JCRJournalHomeAction
[16]	Yan S, Yang DH. 1996. A decision support framework for handling schedule perturbation. Transportation Research Part B: Methodological 30:405−19 doi: 10.1016/0191-2615(96)00013-6 CrossRef Google Scholar
[17]	Vos HWM, Santos BF, Omondi T. 2015. Aircraft schedule recovery problem–a dynamic modeling framework for daily operations. Transportation Research Procedia 10:931−40 doi: 10.1016/j.trpro.2015.09.047 CrossRef Google Scholar
[18]	Vink J, Santos BF, Verhagen WJC, Medeiros I, Filho R. 2020. Dynamic aircraft recovery problem-an operational decision support framework. Computers & Operations Research 117:104892 doi: 10.1016/j.cor.2020.104892 CrossRef Google Scholar
[19]	Lee J, Marla L, Jacquillat A. 2020. Dynamic disruption management in airline networks under airport operating uncertainty. Transportation Science 54:973−97 doi: 10.1287/trsc.2020.0983 CrossRef Google Scholar
[20]	Huang Z, Luo X, Jin X, Karichery S. 2022. An iterative cost-driven copy generation approach for aircraft recovery problem. European Journal of Operational Research 301:334−48 doi: 10.1016/j.ejor.2021.10.055 CrossRef Google Scholar
[21]	Zang H, Zhu J, Zhu Q, Gao Q. 2024. A proactive aircraft recovery approach based on airport spatiotemporal network supply and demand coordination. Computers & Operations Research 165:106599 doi: 10.1016/j.cor.2024.106599 CrossRef Google Scholar
[22]	Xu H, Han S. 2016. Weighted time-band approximation model for flight operations recovery considering simplex group cycle approaches in China. Mathematical Problems in Engineering 2016:3201490 doi: 10.1155/2016/3201490 CrossRef Google Scholar
[23]	Zhu B, Zhu JF, Gao Q. 2015. A stochastic programming approach on aircraft recovery problem. Mathematical Problems in Engineering 2015:680609 doi: 10.1155/2015/680609 CrossRef Google Scholar
[24]	Hu Y, Liao H, Zhang S, Song Y. 2017. Multiple objective solution approaches for aircraft rerouting under the disruption of multi-aircraft. Expert Systems with Applications 83:283−99 doi: 10.1016/j.eswa.2017.04.031 CrossRef Google Scholar
[25]	Wu Z, Li B, Dang C. 2017. Solving multiple fleet airline disruption problems using a distributed-computation approach to integer programming. IEEE Access 5:19116−31 doi: 10.1109/ACCESS.2017.2747155 CrossRef Google Scholar
[26]	Dang C, Ye Y. 2015. A fixed point iterative approach to integer programming and its distributed computation. Fixed Point Theory and Applications 2015:182 doi: 10.1186/s13663-015-0429-8 CrossRef Google Scholar
[27]	Zhang C. 2017. Two-stage heuristic algorithm for aircraft recovery problem. Discrete Dynamics in Nature and Society 2017:9575719 doi: 10.1155/2017/9575719 CrossRef Google Scholar
[28]	Liang Z, Xiao F, Qian X, Zhou L, Jin X, et al. 2018. A column generation-based heuristic for aircraft recovery problem with airport capacity constraints and maintenance flexibility. Transportation Research Part B: Methodological 113:70−90 doi: 10.1016/j.trb.2018.05.007 CrossRef Google Scholar
[29]	Chen CH, Liu TK, Chou JH, Tsai JT, Ho WH. 2008. A hybrid multiobjective genetic algorithm on optimizing aircraft schedule recovery problems under short-time response. 2008 IEEE International Conference on Systems, Man and Cybernetics, Singapore, 12-15 October 2008. USA: IEEE. pp. 915-20. doi: 10.1109/ICSMC.2008.4811397
[30]	Liu T-K, Chen C-H, Chou J-H. 2010. Optimization of short-haul aircraft schedule recovery problems using a hybrid multiobjective genetic algorithm. Expert Systems with Applications 37:2307−15 doi: 10.1016/j.eswa.2009.07.068 CrossRef Google Scholar
[31]	Liu W, Zhu M, Wang X. 2008. Hybrid particle swarm optimization applied to recovery scheduling of large-scale flight delays. 2008 Fourth International Conference on Natural Computation, Jinan, China, 18−20 October 2008. USA: IEEE. pp. 634−39. doi: 10.1109/ICNC.2008.793
[32]	Babić O, Kalić M, Pavković G, Dožić S, Čangalović M. 2010. Heuristic approach to the airline schedule disturbances problem. Transportation Planning and Technology 33:257−80 doi: 10.1080/03081061003732318 CrossRef Google Scholar
[33]	Guimarans D, Arias P, Mota MM. 2015. Large neighbourhood search and simulation for disruption management in the airline industry. In Applied simulation and optimization: in logistics, industrial and aeronautical practice, eds. Mujica Mota M, De La Mota I, Guimarans Serrano D. Cham: Springer. pp. 169-201. doi: 10.1007/978-3-319-15033-8_6
[34]	Lin H, Wang Z. 2018. Fast variable neighborhood search for flight rescheduling after airport closure. IEEE Access 6:50901−9 doi: 10.1109/ACCESS.2018.2869842 CrossRef Google Scholar
[35]	Hu Y, Xu B, Bard JF, Chi H. 2015. Optimization of multi-fleet aircraft routing considering passenger transiting under airline disruption. Computers & Industrial Engineering 80:132−44 doi: 10.1016/j.cie.2014.11.026 CrossRef Google Scholar
[36]	Gao M, Chi H, Xu B, Ding R. 2012. Flight rescheduling responding to large-area flight delays. Kybernetes 41:1483−96 doi: 10.1108/03684921211276693 CrossRef Google Scholar
[37]	Mou D, Zhao W. 2013. An irregular flight scheduling model and algorithm under the uncertainty theory. Journal of Applied Mathematics 2013:361926 doi: 10.1155/2013/361926 CrossRef Google Scholar
[38]	Aktürk MS, Atamtürk A, Gürel S. 2014. Aircraft rescheduling with cruise speed control. Operations Research 62:829−45 doi: 10.1287/opre.2014.1279 CrossRef Google Scholar
[39]	Şafak Ö, Atamtürk A, Selim Aktürk M. 2019. Accommodating new flights into an existing airline flight schedule. Transportation Research Part C: Emerging Technologies 104:265−86 doi: 10.1016/j.trc.2019.05.010 CrossRef Google Scholar
[40]	Bouarfa S, Müller J, Blom H. 2018. Evaluation of a Multi-Agent System approach to airline disruption management. Journal of Air Transport Management 71:108−18 doi: 10.1016/j.jairtraman.2018.05.009 CrossRef Google Scholar
[41]	Khaled O, Minoux M, Mousseau V, Michel S, Ceugniet X. 2018. A multi-criteria repair/recovery framework for the tail assignment problem in airlines. Journal of Air Transport Management 68:137−51 doi: 10.1016/j.jairtraman.2017.10.002 CrossRef Google Scholar
[42]	Wang D, Wu Y, Hu J-Q, Liu M, Yu P, et al. 2019. Flight schedule recovery: a simulation-based approach. Asia-Pacific Journal of Operational Research 36:1940010 doi: 10.1142/S0217595919400104 CrossRef Google Scholar
[43]	Pei S, He Y, Fan Z, Zhang B. 2021. Decision support system for the irregular flight recovery problem. Research in Transportation Business & Management 38:100501 doi: 10.1016/j.rtbm.2020.100501 CrossRef Google Scholar
[44]	Hansen MM, Gillen D, Djafarian-Tehrani R. 2001. Aviation infrastructure performance and airline cost: a statistical cost estimation approach. Transportation Research Part E: Logistics and Transportation Review 37:1−23 doi: 10.1016/S1366-5545(00)00008-9 CrossRef Google Scholar
[45]	Wu Z, Cao Q, Li B, Dang C, Hu F. 2017. A rapid solving method to large airline disruption problems caused by airports closure. IEEE Access 5:26545−55 doi: 10.1109/ACCESS.2017.2773534 CrossRef Google Scholar
[46]	Wu Z, Li B, Dang C, Hu F, Zhu Q, et al. 2017. Solving long haul airline disruption problem caused by groundings using a distributed fixed-point computational approach to integer programming. Neurocomputing 269:232−55 doi: 10.1016/j.neucom.2017.02.091 CrossRef Google Scholar
[47]	Ji C, Gao M, Zhang X, Li J. 2021. A novel rescheduling algorithm for the airline recovery with flight priorities and airport capacity constraints. Asia-Pacific Journal of Operational Research 38:2140025 doi: 10.1142/S021759592140025X CrossRef Google Scholar
[48]	Şi̇mşek D, Aktürk MS. 2022. Resilient airline scheduling to minimize delay risks. Transportation Research Part C: Emerging Technologies 141:103734 doi: 10.1016/j.trc.2022.103734 CrossRef Google Scholar
[49]	AhmadBeygi S, Cohn A, Weir M. 2009. An integer programming approach to generating airline crew pairings. Computers & Operations Research 36:1284−98 doi: 10.1016/j.cor.2008.02.001 CrossRef Google Scholar
[50]	Chang SC. 2012. A duty based approach in solving the aircrew recovery problem. Journal of Air Transport Management 19:16−20 doi: 10.1016/j.jairtraman.2011.12.001 CrossRef Google Scholar
[51]	Fang J, Xia HS. 2013. Hybrid simulate annealing algorithm based dynamic crew scheduling research. Applied Mechanics and Materials 411−414:2359−63 doi: 10.4028/www.scientific.net/amm.411-414.2359 CrossRef Google Scholar
[52]	Liu Q, Zhang X, Chen X, Chen X. 2013. Interfleet and intrafleet models for crew recovery problems. Transportation Research Record 2336:75−82 doi: 10.3141/2336-09 CrossRef Google Scholar
[53]	Chen CH, Chou JH. 2017. Multiobjective optimization of airline crew roster recovery problems under disruption conditions. IEEE Transactions on Systems, Man, and Cybernetics: Systems 47:133−44 doi: 10.1109/TSMC.2016.2560130 CrossRef Google Scholar
[54]	Bayliss C, De Maere G, Atkin J, Paelinck M. Probabilistic airline reserve crew scheduling model. Proc. 12th Workshop on Algorithmic Approaches for Transportation Modelling, Optimization, and Systems (2012), 2012: Schloss-Dagstuhl-Leibniz Zentrum für Informatik. https://doi.org/10.4230/OASIcs.ATMOS.2012.132
[55]	Bayliss C, De Maere G, Atkin J, Paelinck M. 2013. Scheduling airline reserve crew to minimise crew related delay using simulated airline recovery and a probabilistic optimisation model. 2013 IEEE International Conference on Systems, Man, and Cybernetics, Manchester, UK, 13−16 October 2013. USA: IEEE. pp. 1944−50. doi: 10.1109/SMC.2013.334
[56]	Bayliss C, De Maere G, Atkin JAD, Paelinck M. 2017. A simulation scenario based mixed integer programming approach to airline reserve crew scheduling under uncertainty. Annals of Operations Research 252:335−63 doi: 10.1007/s10479-016-2174-8 CrossRef Google Scholar
[57]	Bayliss C, De Maere G, Atkin JAD, Paelinck M. 2020. Scheduling airline reserve crew using a probabilistic crew absence and recovery model. Journal of the Operational Research Society 71:543−65 doi: 10.1080/01605682.2019.1567649 CrossRef Google Scholar
[58]	Dillon JE, Kontogiorgis S. 1999. US airways optimizes the scheduling of reserve flight crews. Interfaces 29:123−31 doi: 10.1287/inte.29.5.123 CrossRef Google Scholar
[59]	Luo X, Dashora Y, Shaw T. 2015. Airline crew augmentation: decades of improvements from sabre. Interfaces 45:409−24 doi: 10.1287/inte.2015.0803 CrossRef Google Scholar
[60]	Wen X, Ma HL, Chung SH, Khan WA. 2020. Robust airline crew scheduling with flight flying time variability. Transportation Research Part E: Logistics and Transportation Review 144:102132 doi: 10.1016/j.tre.2020.102132 CrossRef Google Scholar
[61]	Herekoğlu A, Kabak Ö. 2024. Crew recovery optimization with deep learning and column generation for sustainable airline operation management. Annals of Operations Research 342:399−427 doi: 10.1007/s10479-023-05738-z CrossRef Google Scholar
[62]	Zhong H, Lian Z, Zhou T, Niu B. 2024. A time-varying competitive swarm optimizer for integrated flight recovery with multi-objective and priority considerations. Computers & Industrial Engineering 190:110019 doi: 10.1016/j.cie.2024.110019 CrossRef Google Scholar
[63]	Guo Y. 2005. A decision support framework for the airline crew schedule disruption management with strategy mapping. Proc. Operations Research Proceedings 2004: Selected Papers of the Annual International Conference of the German Operations Research Society (GOR). Jointly Organized with the Netherlands Society for Operations Research (NGB) Tilburg, September 1–3, 2004. Berlin, Heidelberg: Springer. pp. 158−65. doi: 10.1007/3-540-27679-3_20
[64]	McCarty LA, Cohn AEM. 2018. Preemptive rerouting of airline passengers under uncertain delays. Computers & Operations Research 90:1−11 doi: 10.1016/j.cor.2017.09.001 CrossRef Google Scholar
[65]	Zhang D, Henry Lau H, Yu C. 2015. A two stage heuristic algorithm for the integrated aircraft and crew schedule recovery problems. Computers & Industrial Engineering 87:436−53 doi: 10.1016/j.cie.2015.05.033 CrossRef Google Scholar
[66]	Maher SJ. 2016. Solving the integrated airline recovery problem using column-and-row generation. Transportation Science 50:216−39 doi: 10.1287/trsc.2014.0552 CrossRef Google Scholar
[67]	Muter İ, Birbil Şİ, Bülbül K. 2013. Simultaneous column-and-row generation for large-scale linear programs with column-dependent-rows. Mathematical Programming 142:47−82 doi: 10.1007/s10107-012-0561-8 CrossRef Google Scholar
[68]	Jafari N, Zegordi SH. 2010. The airline perturbation problem: considering disrupted passengers. Transportation Planning and Technology 33:203−20 doi: 10.1080/03081061003643788 CrossRef Google Scholar
[69]	Abdelghany KF, Abdelghany AF, Ekollu G. 2008. An integrated decision support tool for airlines schedule recovery during irregular operations. European Journal of Operational Research 185:825−48 doi: 10.1016/j.ejor.2006.12.045 CrossRef Google Scholar
[70]	Bard JF, Yu G, Argüello MF. 2001. Optimizing aircraft routings in response to groundings and delays. Iie Transactions 33:931−47 doi: 10.1023/A:1010987008497 CrossRef Google Scholar
[71]	Arıkan U, Gürel S, Aktürk MS. 2016. Integrated aircraft and passenger recovery with cruise time controllability. Annals of Operations Research 236:295−317 doi: 10.1007/s10479-013-1424-2 CrossRef Google Scholar
[72]	Marla L, Vaaben B, Barnhart C. 2017. Integrated disruption management and flight planning to trade off delays and fuel burn. Transportation Science 51:88−111 doi: 10.1287/trsc.2015.0609 CrossRef Google Scholar
[73]	Santos BF, Wormer MMEC, Achola TAO, Curran R. 2017. Airline delay management problem with airport capacity constraints and priority decisions. Journal of Air Transport Management 63:34−44 doi: 10.1016/j.jairtraman.2017.05.003 CrossRef Google Scholar
[74]	Naz Yeti̇moğlu Y, Selim Aktürk M. 2021. Aircraft and passenger recovery during an aircraft's unexpected unavailability. Journal of Air Transport Management 91:101991 doi: 10.1016/j.jairtraman.2020.101991 CrossRef Google Scholar
[75]	Eggenberg N, Salani M, Bierlaire M. 2010. Constraint-specific recovery network for solving airline recovery problems. Computers & Operations Research 37:1014−26 doi: 10.1016/j.cor.2009.08.006 CrossRef Google Scholar
[76]	Jozefowiez N, Mancel C, Mora-Camino F. 2013. A heuristic approach based on shortest path problems for integrated flight, aircraft, and passenger rescheduling under disruptions. Journal of the Operational Research Society 64:384−95 doi: 10.1057/jors.2012.20 CrossRef Google Scholar
[77]	Zhang D, Yu C, Desai J, Lau HH. 2016. A math-heuristic algorithm for the integrated air service recovery. Transportation Research Part B: Methodological 84:211−36 doi: 10.1016/j.trb.2015.11.016 CrossRef Google Scholar
[78]	Mansi R, Hanafi S, Wilbaut C, Clautiaux F. 2012. Disruptions in the airline industry: math-heuristics for re-assigning aircraft and passengers simultaneously. European Journal of Industrial Engineering 106:690−712 doi: 10.1504/EJIE.2012.051074 CrossRef Google Scholar
[79]	Sinclair K, Cordeau JF, Laporte G. 2016. A column generation post-optimization heuristic for the integrated aircraft and passenger recovery problem. Computers & Operations Research 65:42−52 doi: 10.1016/j.cor.2015.06.014 CrossRef Google Scholar
[80]	Sinclair K, Cordeau J-F, Laporte G. 2014. Improvements to a large neighborhood search heuristic for an integrated aircraft and passenger recovery problem. European journal of operational research 233:234−45 doi: 10.1016/j.ejor.2013.08.034 CrossRef Google Scholar
[81]	Hu Y, Song Y, Zhao K, Xu B. 2016. Integrated recovery of aircraft and passengers after airline operation disruption based on a GRASP algorithm. Transportation Research Part E: Logistics and Transportation Review 87:97−112 doi: 10.1016/j.tre.2016.01.002 CrossRef Google Scholar
[82]	Yang T, Hu Y. 2019. Considering passenger preferences in integrated postdisruption recoveries of aircraft and passengers. Mathematical Problems in Engineering 2019:9523610 doi: 10.1155/2019/9523610 CrossRef Google Scholar
[83]	Zhao A, Bard JF, Bickel JE. 2023. A two-stage approach to aircraft recovery under uncertainty. Journal of Air Transport Management 111:102421 doi: 10.1016/j.jairtraman.2023.102421 CrossRef Google Scholar
[84]	Chen K, Wu J, Yang L. 2024. Reassigning departure slots with preferences of the airline and passengers. Transportation Research Record 2678:786−804 doi: 10.1177/03611981231182714 CrossRef Google Scholar
[85]	Brunner JO. 2014. Rescheduling of flights during ground delay programs with consideration of passenger and crew connections. Transportation Research Part E: Logistics and Transportation Review 72:236−52 doi: 10.1016/j.tre.2014.10.004 CrossRef Google Scholar
[86]	Petersen JD, Sölveling G, Clarke JP, Johnson EL, Shebalov S. 2012. An optimization approach to airline integrated recovery. Transportation Science 46:482−500 doi: 10.1287/trsc.1120.0414 CrossRef Google Scholar
[87]	Maher SJ. 2015. A novel passenger recovery approach for the integrated airline recovery problem. Computers & Operations Research 57:123−37 doi: 10.1016/j.cor.2014.11.005 CrossRef Google Scholar
[88]	Arıkan U, Gürel S, Aktürk MS. 2017. Flight network-based approach for integrated airline recovery with cruise speed control. Transportation Science 51:1259−87 doi: 10.1287/trsc.2016.0716 CrossRef Google Scholar
[89]	Evler J, Lindner M, Fricke H, Schultz M. 2022. Integration of turnaround and aircraft recovery to mitigate delay propagation in airline networks. Computers & Operations Research 138:105602 doi: 10.1016/j.cor.2021.105602 CrossRef Google Scholar
[90]	Xu Y, Wandelt S, Sun X. 2023. A distributionally robust optimization approach for airline integrated recovery under in-flight pandemic transmission risks. Transportation Research Part C: Emerging Technologies 152:104188 doi: 10.1016/j.trc.2023.104188 CrossRef Google Scholar
[91]	Ding Y, Wandelt S, Wu G, Xu Y, Sun X. 2023. Towards efficient airline disruption recovery with reinforcement learning. Transportation Research Part E: Logistics and Transportation Review 179:103295 doi: 10.1016/j.tre.2023.103295 CrossRef Google Scholar
[92]	Cadarso L, Vaze V. 2023. Passenger-centric integrated airline schedule and aircraft recovery. Transportation Science 57:813−37 doi: 10.1287/trsc.2022.1174 CrossRef Google Scholar
[93]	De Castro AJM. 2013. A distributed approach to integrated and dynamic disruption management in airline operations control. Thesis. Universidade do Porto, Portugal

Contents

Policies

Services

Partnerships

Review of optimization problems, models and methods for airline disruption management from 2010 to 2024

Abstract

Introduction

Chemical components of Huangqin Tea

Flavonoids

Essential oils

Others

Functional properties

Anti-inflammatory activity

Anti-bacterial and antiviral activity

Antipyretic and analgesic effect

Neuroprotective effect

Effects on cardiovascular and cerebrovascular diseases

Anti-aging effects

Others

Safety issues

Conclusions and future perspectives

Author contributions

Data availability

Acknowledgments

Conflict of interest

Rights and permissions

References

About this article

Cite this article

Article Metrics

Access History

Other Articles By Authors