• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Measuring the resilience of an airport network

    2020-01-09 01:04:40YanjunWANGJianmingZHANXinhuaXULishuaiLIPingCHENMarkHANSEN
    CHINESE JOURNAL OF AERONAUTICS 2019年12期

    Yanjun WANG, Jianming ZHAN, Xinhua XU, Lishuai LI,Ping CHEN, Mark HANSEN

    a College of Civil Aviation, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China

    b Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, Cambridge 02139, USA

    c Air Traffic Management Bureau, Beijing 100012, China

    d Xiamen Airlines, Xiamen 361006, China

    e Department of Systems Engineering and Engineering Management, City University of Hong Kong, Kowloon, Hong Kong, China

    f State Key Laboratory of Air Traffic Management System and Technology, Nanjing 210000, China

    g Department of Civil and Environmental Engineering, University of California, Berkeley, CA 94720, USA

    Abstract Resilience is the ability of a system to withstand and stay operational in the face of an unexpected disturbance or unpredicted changes. Recent studies on air transport system resilience focus on topology characteristics after the disturbance and measure the robustness of the network with respect to connectivity. The dynamic processes occurring at the node and link levels are often ignored. Here we analyze airport network resilience by considering both structural and dynamical aspects. We develop a simulation model to study the operational performance of the air transport system when airports operate at degraded capacity rather than completely shutting down.Our analyses show that the system deteriorates soon after disruptive events occur but returns to an acceptable level after a period of time. Static resilience of the airport network is captured by a phase transition in which a small change to airport capacity will result in a sharp change in system punctuality.After the phase transition point,decreasing airport capacity has little impact on system performance.Critical airports which have significant influence on the performance of whole system are identified,and we find that some of these cannot be detected based on the analysis of network structural indicators alone.Our work shows that air transport system’s resilience can be well understood by combining network science and operational dynamics.

    KEYWORDS Air transport;Complex networks;Network resilience;Resilience indicator;Transportation resilience

    1. Introduction

    Air transport has been an important mode of transportation for many decades.In 2017,a total of 4.1 billion passengers flew on commercial airlines according to preliminary statistics from the International Civil Aviation Organization (ICAO).1Continued traffic growth puts continuing pressure on the aviation industry to expand operations without compromising safety or service quality. Hence, significant attention has been given to the development of advanced systems/tools and operational concepts to improve the safety, predictability, and efficiency of air transport systems.For instance,Trajectory Based Operations (TBO) promises to enhance air traffic safety, predictability, and efficiency. As one of the core concepts of next generation air transportation system, TBO relies on the planning and prediction of four dimensional trajectories(4DT, the position of aircraft in 3D space plus time), along with algorithms that explicitly calculate conflict-free trajectories for thousands of flights.2-4

    While many dimensions of aviation system performance are being advanced through concepts such as TBO, resilience has not been so widely investigated in this domain.5As a complex socio-technical system, air transport system is composed of a huge number of interconnected subsystems with complex interactions between them.It may therefore be highly vulnerable to internal or external disruptive events. For example, a single airport closure due to serve weather or industrial action can have widespread systemic effects. The consequence caused by such disruptive events can range from a few delayed flights to system collapse.6As reported in Ref.7the average damage arising from US transportation system disruptions from extreme weather exceeds $1 billion annually. On April 3,2018, an inadvertent link between a computer being used to test new software and the live operational system triggered the Enhance Tactical Flow Management System (ETFMS)outage which caused almost half of European flights to be delayed.8,9To prevent such system failures and to make contingency plans, it is crucial to understand the resilience of the air transport system.

    Resilience,the ability of a system to remain operational and withstand the impact of unexpected disturbances or unpredicted changes by adjusting its structure or activity,has drawn a lot of attention in a wide range of disciplines, ranging from economics,10through ecology,11to engineering fields.12,13Despite various definitions of resilience given in these works,a general consensus is that resilience has four major properties,commonly referred to as the 4Rs: Robustness, Redundancy,Resourcefulness, and Rapidity.14Robustness is mainly used to describe the ability of a system to maintain its basic functions in the presence of disruptions/errors, while redundancy is the availability of substitutes that can be used when some parts of the system are damaged or malfunction. Resourcefulness refers to the ability of the system to mobilize needed resources, while rapidity is the speed at which the system can be restored to normal operation.

    A review of generic definitions and measures of resilience in the context of engineering systems can be found in Ref.15.The definitions of resilience can be further divided into three groups based on the phases of system response when disruptive events occur. In this regard, the terms absorptive capability,adaptive capability, and restorative capability have been suggested, and an integrated metric for quantifying system resilience is developed.16However, the definition and metrics for resilience in air transport have not yet been well explored.

    A wide body of research on resilience can be found in transport studies, especially transportation networks facing natural hazards or inclement weather.17,18Resilience can be investigated either by modeling system evolution or by analyzing empirical data. For instance, Nogal et al.19investigate the dynamical aspects of a system when disruptive events take place. They assess the resilience of a transport network from the beginning of the disruption to total system recovery. To simulate network performance evolution, they develop a dynamic equilibrium-restricted assignment model. A second example is to quantify the resilience of transportation system using vehicles’GPS data.20Gao et al.21developed a set of analytical tools to predict system resilience.They found that three key structural factors affecting system’s resilience are density,heterogeneity and symmetry. However, their work is limited since the systems examined are those in which interactions between nodes are mutualistic.

    Research about the resilience of the air transport system has lagged until recently, when a few publications have appeared.22,5,6,23,24However, there has been considerable research focusing on the robustness of airport networks over the past two decades.25These works mainly focus on statistical regularities uncovered from analyzing airport or airline networks.Airport networks are normally defined based on scheduled flights, with airports as nodes and direct flights between them forming the edges. Network robustness is typically measured by assessing changes in topological metrics that result from removing certain nodes when disruptive events occur.For example, Lordan et al.26modeled the global air transport network and assessed robustness based on simulated attacks on certain airports chosen based on different selection criteria.They found that an adaptive selection strategy based on betweenness centrality can be used to detect the critical airports whose failure would cause a great loss of network connectivity. Du et al.27by employing a modeling approach based on multiple layer networks, found that Chinese airline network is less redundant than the worldwide airline network.Hossain and Alam28analyzed the robustness of Australia Airport Network(AAN)under different disruptive scenarios,and found that the AAN is robust when a few edges or air-routes fail, but is very sensitive to malfunctions of central nodes(those with high degree and betweenness) measured by topological sensitivity, reachability and re-routing cost. Ren and Li29compared the structure of the China and United States air traffic networks. They found that Chinese air traffic network is less well connected and less robust.

    Attention has recently shifted from discovering macroscopic properties of complex networks to understanding the role of individual nodes, links, and motifs played in the network,30,31using network science and big-data techniques.Cong et al.32found six airports which play important roles in delay propagation within the Chinese airport network. Du et al.33focused on identifying critical links that have the largest impact on network robustness as measured by the size of giant components. However, the complex networks approach has important limitations. One obvious one is that system dynamics are neglected since these works focus only on the static structure of the system. This means that the ability of the system to adapt to disruptions is neglected.

    As the importance of complexity as a fundamental feature of air transportation has become increasingly recognized, we need to rethink the resilience of the system. There are several important questions that still need to be answered. For example,how do we assess the resilience of airport networks?Where are the vulnerable parts of the system?How can we increase air transport resilience? In this paper, we present an experimental study to understand airport network resilience.Specifically,we want to explore(A)what is the degree of disruption the system can absorb while still being able to function; (B) what is the relationship between resilience and network dynamics (specifically how flight delays and punctuality change over time) in an air transport setting.

    The rest of this paper is structured as follows: Section 2 describes disruptive events in air transport system and their impacts. In Section 3, we propose four metrics to measure the resilience of airport network. Section 4 presents a simulation model and scenarios used to evaluate these metrics.Results are analyzed in Section 5. Section 6 gives concluding remarks and discusses future studies.

    2. Disruptive events in transport system and their impact

    Fig. 1 depicts how a disruption affects the performance of a system over time. As shown in Fig. 1, the system operates in an initial steady state before disruptive event(s) occur at time t0. After the disruptive event occurs, the performance of the system begins to decline at time td. Three possible paths that the system may follow are represented by the solid red line,blue dashed line, and grey dashed line in the figure. In the first scenario (red line), when the negative impact of disruptive events dissipates, the performance of the system will gradually recover without any outside intervention until it reaches the original steady state. In the second scenario (blue dashed line), the system first reaches a new steady state, but eventually returns to its original state. For example, temporary routes may be built quickly for the purpose of rescue and to meet immediate needs when transportation systems are damaged due to earthquake or other natural disasters.It may take weeks or months for the system to fully recover.The worst scenario is that the system is unable to recover(shown in grey dashed line). When severity of the disruption exceeds some limit, the system completely collapses. For any given path, the performance lost is the gap between the performance in the original steady state and the worst performance after the disruption; this is shown for the second scenario in Fig. 1.

    Fig.1 System performance evolution before and after disruption.

    In air transport systems, disruptive events occurring around airports include severe weather (e.g. dense fog, heavy rain, snow), equipment outages (e.g. approach lighting outages), terrorist attacks, and accidents. These events usually force the airport to operate in a degraded mode instead of shutting down. The capacity of the affected airports decreases and then returns to a normal level after recovery.In contrast to what is assumed in most studies on airport network robustness, the decrease in node capacity will not disconnect it from the rest of the network, but could result in a large number of delayed flights at both the airports directly affected by the disruptive events and other airports as a result of delay propagation. When disruptive events occur, the system will normally not collapse; rather it will absorb the perturbations, withstand the disruption, and thus remain functional. Normally, the system will bounce back to an acceptable stable state after a period of recovery if the external ‘‘pressure” is not too great. In the worst case, the system will not be able to regain its original state. In this setting,resilience is the ability of the system to return to its original state after a disruption.

    The understanding of the resilience of air transport network has fundamental meanings to both academic and engineering fields. Recent resilience studies in air transport have explored the resilience from a particular perspective, such as scheduled block time reliability (Hao and Hansen 2014), airport closures due to weather or natural disaster,6,23or the cost of resilience of an air traffic management system.23However, few efforts have been made to develop metrics of air transport network resilience that comprehensively account for the impact on air traffic operations. This paper focuses on understanding the operational resilience of an air transport network when perturbations occur in airport(s) via the analysis of several resilience metrics. The analysis is built upon previous work on the impact of flight schedule and ground turn-around times of aircraft on delay propagation,but is geared toward evaluating system wide behavior from a macroscopic point of view rather than delay propagation per se. Specifically, we want to investigate the relationship between the ‘‘pressure” on air transport system and the degree and dynamics of the operational ‘‘deformations” that result.

    3. Measurements of performance and network resilience

    To the best of our knowledge,there is no common metric of air transport network resilience. Delay and punctuality are the two commonly indicators of system performance in resilience studies of transportation system.34In the air transport system,large flight delays or low punctuality indicate poor performance of the system. Therefore, we introduce the following four metrics to quantify air transport system performance:airport departure delay,System-Wide Delay(SWD),punctuality,and General Resilience Index (GRI).

    3.1. Airport departure delay

    Normally,departure delay of a flight f is calculated by the time difference between its actual departure timeand its schedule departure time.However,there is no universally recognized method to measure departure delay in an airport in a given time period [t1,t2]. Here, we adopt the following definition.First, determine non-negative delay time of each single flight in [t1,t2] by

    3.2. System wide delay

    To directly represent the demand and supply of the airport over time, Liu et al.35,36proposed a method based on queuing diagram of airport flight departures (a similar figure could be constructed for arrivals). We use this method to propose a measure of delay for the whole air transport system.The queuing diagram for system departure (or arrival) flights is shown in Fig. 2. The blue line represents the scheduled cumulative demands,S(t ), with the slope indicating schedule demand rate of the system. It can be constructed by determining, for any time t, the number of flights that are scheduled to depart by time t across all the airports in the system. The red line shows the actual cumulative departures, A(t ), which is obtained by determining, for any time t, the number of departures that have actually taken place.Its slope reflects the throughput rate of the system, a measure of supply. At the beginning of the operating day, the two curves almost overlap until time T1,which suggests a balance between supply and demand. From time T1to T2,the actual supply rate decreases because of some disturbance, causing an imbalance between supply and demand.After time T2the system starts to recover,and supply rate exceeds the demand rate within a short time. Eventually,the actual cumulative departure curve again overlaps scheduled cumulative demand curve, which indicates the system has recovered.

    For a given cumulative value n the horizontal difference between the two curves measures the time difference between when the nth departure was scheduled to occur and when the nth departure actually occurred. (If the flight order of scheduled and actual departures match, it is also the delay for the nth flight.) The total SWD can be obtained by summing these differences across all values of n. This is the same as the area between scheduled cumulative demand curve and actual cumulative departure curve,which is highlighted by the striped area in Fig. 2. Thus SWD can be calculated as:

    Fig.2 Queuing diagram of the departure flights of entire system.

    where S( t ) is the scheduled cumulative demand, A(t ) is the actual cumulative supply, Tsis the initial time of the flight schedule and Teis the time when the last flight actually departs.

    3.3. Punctuality

    Here we use departure punctuality as another measure of system performance. Let Ft1,t2be the set of the flights which are scheduled to depart in the time period [t1,t2], Nschedule,t1,t2be the total number of flights in the set,Nschedule,t1,t2be the number of flights in the set whose departure delay is less than 15 minutes. Departure punctuality is then defined as

    3.4. General resilience index (GRI)

    In Ref.16,a general resilience indicator is proposed considering absorptive, adaptive, and restorative capability based on the evolution of system performance as shown in Fig.1.Similarly,here we define a General Resilience Index (GRI) for an air transport network as follows.

    where R is the capability of the system to resist disruptions,i.e.the minimum performance of the system during the period of[td,tns] in Fig. 1. RAPIDPrefers to the rate of performance decline when disruptions occur and can be approximated by the average slope of the line in the disruptive state,i.e.between the time when negative impacts appear (td) and the time of recovery (tr). Similarly, RAPIRPdescribes the rate of system recovery which can be obtained as the slope of the line between time of recovery ((tr).) and the time when the system reaches the new steady state (tns).TAPL is the Time Average Performance Loss, i.e.,

    where MOP stands for Measurement of Performance, which will be punctuality or airport flight delay in our study. Recalling that t0is the time when disruptive event(s)occur,Recovery Ability (RA) is defined as

    4. Model

    Building up on the previous work in Refs.37,38, we develop an agent-based model to investigate operational resilience of air transport network. The model takes daily flight schedules and airport capacity profiles as basic input. Traffic dynamics of the system is considered by incorporating aircraft rotation in the airports and the growth and dissipation of queues. In the following,we describe the general framework of the model and its key elements and mechanisms.

    The overall framework of the model is depicted in Fig.3.It should be mentioned that detailed model in each part can be further developed. For example, the turnaround time can be further modeled with taxiing, deboarding, boarding, etc.

    Each flight is carried out by an aircraft with a unique identification number (i.e. tail number). An aircraft usually has to fulfill several flights during a daily operation.The properties of a flight record include flight number,scheduled departure/arrival airports, scheduled departure/arrival times, estimated departure/arrival times,actual departure/arrival times,and tail number.Once the preceding flight has landed at its destination airport, the successor flight’s status is changed to turnaround.Once the turnaround procedure is completed, the flight will request to enter departure queue.

    In our model, we simplify an airport as a single/multiple runways system. An airport runway may have various operation modes.For example,Shanghai Pudong International Airport has three runways operating simultaneously, with one runway only used for departures, one used for arrivals, and the third one used for both.The main goal of the airport agent is to compute and determine takeoff and landing times. The airport agent consists of three subagents, namely departure runway agent,arrival runway agent,and mixed runway agent.When a flight completes its turnaround procedure and requests for departure, airport agent will assign it to a runway agent together with a departure slot through the interactions with three subagents.

    Flight delay propagates through the airport network. The flying timeof aircraft i between airport a and b not only depends on aircraft performance but also differs from airline to airline.Seasonal wind is a major factor affecting flying time as well. In our model, we generate the distributions of flying times of the same aircraft type of an airline between each given airport pair from an empirical dataset.

    The model was calibrated and validated using one month of historical data of China’s air traffic network. The data contains 358695 flights operated in 251 airports mainland China in May, 2017, providing a comprehensive picture of air transport in China. Each flight record reports the flight number,execute date, scheduled/actual departure (arrival) airport,schedule/actual departure (arrival) time, and the unique aircraft registration number (tail number). Such data allows us to easily reconstruct the path of each aircraft flying in the network.To evaluate the performance of our model,we compare the simulated airports’ operations and the actual operational data. Fig. 4 plots both actual and simulated flights operations and delay in the whole network as well as a single airport, for an arbitrarily chosen date.As can be seen from the figures,our model provides a good macroscopic approximation of the actual operation of the air transport system. In addition, our model is fast - the computational time of simulating 24 hours of operations at 251 airports is less than 2 minutes.It is therefore suitable for the analysis of air transport system resilience.

    Fig.3 Framework of simulation model.

    Fig.4 Comparison between simulation results and operational data.

    5. Results

    5.1. Impact of random disruptions in the network

    To investigate the impact of randomly occurring disruptions to the air transport network, disruptive events are simulated by decreasing airport capacities. In the simulations presented in this section, disruptive events randomly occur in a certain number of airports. We mainly test two factors related to the disruptions: the scale of disruption which is captured by the number of airports under disruption (Nd), and the intensity of disruptive event, which is captured by increased required headway between two consecutive aircraft that are departing/-landing on the runway (Ts). The time periods of disruptive events are set during daily peak hours to intensify the impact.According to the hourly traffic distribution,the departure peak hours are between 08:00 and 10:00, and the peak hours for departures and arrivals combined are between 14:00 and 16:00.In each scenario,we randomly choose a certain number of airports to be disrupted airports and decrease their capacities during the two time periods identified above. A Monte Carlo simulation method was used to simulate each scenario 100 times.The flight data of a typical day,05/28/2017 was used for the baseline flight schedule and itineraries.Elements of randomness in the Monte Carlo simulation include flight times,turn-around times, runway configurations, all of which are drawn from empirical distributions, and the set of airports to be disrupted.

    The impact of disruptive events on the air transport system is summarized in Fig. 5. We present the results for two sets of experiments.In the first set,we fix the intensity of disruption at a constant level - the minimum headway of affected airports during periods of disruption is set to 10 minutes, and we change the number of disrupted airports. The results are plotted in Figs. 5(a) and (b). Different colors represent different numbers of disrupted airports. In the second set of experiments,we set the scale of disruption as a constant-30 airports are disrupted-and vary the intensity of disruption. The results are presented in Figs.5(c)and(d),where different colors stand for different intensities of disruption.

    Fig.5 Average delay and punctuality (disruption period 08:00-10:00).

    It is observed that both the scale and the intensity of disruptions affect the severity of performance degradation and the time to recover.As shown in Figs.5(a)and(b),when the number of airports under disruption becomes larger,the departure delay increases, the punctuality drops, and both slowly return to the normal condition.A similar trend is observed in Figs.5(c) and (d) when minimum headway is increased. Note that in the latter case, additional departure delay persists through midnight,although it will eventually clear sometime after that.

    Another observation is that the performance degradation is more sensitive when the scale and the intensity of disruption are relatively small.After the scale and the intensity of disruption reach some threshold,the marginal performance degradation is very small. We observe that (A) there is not much difference either in flight delay or punctuality when the number of airports under disruption exceeds 30(see Figs.5(a)and b));(B) departure delay and punctuality reach theirs limits when runway separation is bigger than 30 minutes, as shown in Figs. 5(c) and (d).

    As can be seen from Figs. 5(a) and (c), departure delay increases rapidly when the disruption starts at 08:00, and reaches its peak in the hour between 09:00 and 10:00. Afterwards, delay decreases as the disruption dissipates. Closer inspection reveals the second peak between 12:00 and 13:00.We checked the flying times between most Chinese cities,and they can be described as lognormal distribution:

    With μ=4.87,σ=0.32, which means that typical flying times between Chinese cities are around 130 minutes. Adding the required ground turnaround time for an aircraft between its arrival and next departure, the scheduled departure time for a flight of an airframe whose first departure is around 08:00 should be around about 12:00. However, these flights were delayed. That is the reason for the second peak between 12:00 and 13:00 in the figure.

    Figs.5(b)and(d)show the punctuality of air transport system. In the normal operation scenario, the punctuality approaches 1.0,while in the disruptive scenarios,the punctuality declines quickly as the number of disrupted airports increases.In the recovery stages, the rate at which punctuality recovers increases with the number of disrupted airports.However, as the number of disrupted airports increases, it takes longer time to return to any level of punctuality that might be deemed acceptable(for example 0.9).Under the more severe disruptions, the punctuality level has still not recovered by midnight.

    We compute the General Resilience Indicator (GRI) to jointly compare the effect of disruption scale (number of airports under disruption) and disruption intensity (minimum headway between flights).In each disruption scenario,theoretical runway capacity is calculated based on the minimum headway between flights. An average GRI is then obtained. As shown in Fig. 6, GRI decreases with an increasing number of airports under disruption or a decreasing runway capacity(i.e. increasing minimum headway). The change is more sensitive to the changes in runway separation than the number of airports under disruption.

    Fig.6 General resilience indicator of air transport network.

    One of our aims is to understand the degree of disruption that an air transport network can stand. The GRI graph(Fig. 6) helps to answer this. Clearly, there are three different states that air transport system undergoes when increasing the disruption intensity or scale. During State I, the resilience of the system drops quickly when either runway capacity decreases or the number of disturbed airports increases.When averaged runway capacity is down to 11-12 aircraft per hour,or the number of affected airports is over 35,the whole system reaches to State II. Interestingly, we find that lower bound of air transport system resilience (State III in the Fig. 6). The dark blue area (runway separation >20 mins and number of airports under disruption >30) in the GRI graph is where the system reaches its resilience limit. Any disruption larger or more severe than that will make it difficult for the system recover.

    5.2. Network measures and operational indicators for resilience

    The second objective of this study is to identify airports whose operation has significant impact on the performance of the whole network. We rank the airports by network topological indicators, compute their operational indicators for resilience,and then compare these two indicators to see if airports that are important by topological indicators have a large impact on network resilience.

    In network studies, topological indicators such as degree,Betweenness Centrality (BC), closeness, and H-index are used to measure the importance of nodes.Nodes that are ranked on the top of the list measured by these topological indicators are considered important nodes, and networks can be severely damaged when these nodes fail or are removed. The degree of a node is defined as the number of the neighbor nodes(i.e. directly connected nodes) that it has. It is the most intuitive and basic measure of interactivity of a node with other nodes in the network.The BC of a node is defined as the number node pairs whose shortest paths pass through this node.Closeness is obtained as the sum of the length of the shortest paths between the node and all other nodes in the network.H-index is proposed in complex networks by Hirsch in 2005 to evaluate the scholars’ academic achievements,39and has recently been found to be another important structural index to measure the importance of a node.30The H-index of a node in a network is defined has the largest value h such that thenode has at least h adjacent nodes of degree not less than h.Table 1 lists topological indicators of 20 airports in mainland China.

    Table 1 Topological measures of top 20 airports in China mainland.

    To examine the importance of an airport, we selectively decrease the airport’s capacity by increasing minimum headway to [10, 20, 30, 40, 50, 60] minutes. System Wide Delay(SWD) is calculated based on the description in Section 3. In order to intuitively show the correlation between node topological and operational indicators, we sort SWD according to the node’s network topological value (see Fig. 7).

    Since the abscissas in Fig.7 are ordered by topological indicators of airport importance,a given plot should be monotonically decreasing if its associated topological indicator is perfectly correlated with the impact of a disruption on network performance. As shown in Fig. 7(a), for the five airports with the highest degree(ZBAA,ZGGG,ZSSS,ZUUU,ZSPD),the SWD increases as the degree increases.The curves show a general trend of ‘‘low left and high right”, but there exist many‘‘jump points”whose SWD is bigger than that of the next airport to its right, which has a higher topological indicator value. If an airport is associated with a ‘‘jump point,” it is more important based on the operational performance metric than it is with respect to degree.More‘‘jump points”appear in Fig.7(b)which plots SWD in order of airport betweenness.In Fig.7(d),we sort airports by H-index first then by SWD if the airports have same H-index. As shown in Fig. 7(d), jump points are different from the ones ranked by degree,closeness,and betweenness, but still exist. Degree, closeness, betweenness, and H-index all show similar patterns. Although the nodes with high values of topological indicators are critical to network connectivity, none of the indicators can be used to reliably assess airport importance in terms of operational impact of a disruption.

    Fig.7 Performance of airport network when an airport is disrupted.

    In addition to the topological indicators,gross measures of airport activity, such as the number of flight operations, may capture the importance of an airport in terms of operational disruption.Fig.7(e)shows SWD sorted by the number of daily flights. From the perspective of SWD, Fig. 7(e) shows fewer and smaller jump points suggesting that the number of flights is an important driver operational impact of airports. This is further confirmed by the Spearman correlation analysis between structural indicators and operation performance at the end of this section(see Fig.8).However,some jump points remain in Fig. 7(e), implying that that the number of flights is also not a perfect predictor of operational impact of an airport disruption.Other factors,including topological indicators considered above as well as metrics related to aircraft connectivity,may also play a role.

    Fig.7 (continued)

    Fig.8 Spearman’s rank correlations between structural indicators and operational measures.

    We take a closer look at the ‘‘jump” airports in Fig. 7.Fig. 9 shows the locations of jumps in the whole air transport network. The colors of the circle indicate the frequency of jumps. ZYTX (Shenyang Taoxian International Airport) and ZSFZ(Fuzhou Changle International Airport)are the two airports that appear as jumps irrespective of which metric is used for sorting. This suggests that no readily available topological or activity measure can completely account for the importance of ZYTX and ZSFZ airports in terms of the operational impact of a capacity reduction: these airports are more operationally significant than any of these indicators suggest. ZPPP(Kunming Changshui International Airport), ZSNJ (Nanjing Lukou International Airport), ZSHC (Hangzhou Xiaoshan International Airport) appeared four times as jump points.Among these, the results for ZPPP are the most readily explainable: this airport has the second highest betweenness of the airports in the Chinese network, that is, the number of shortest paths through ZPPP in the network is relatively large. Therefore, the average travel time of the network will be greatly increased if ZPPP is disrupted,but the system delay impact is not necessarily large. System wide operation managers should take precautions if the capacity of these‘‘jumping airports” drops.

    Fig.9 Frequency of jumps in the five sorting strategies.

    We further analyze the correlation between topological and activity indicators and operational resilience measures through Spearman’s rank correlation coefficient. The ranking of airports by topological indicators is compared with the ranking of airports by SWD. The Spearman’s rank correlation coefficient assesses how similar two rankings are - a high value means the rankings are similar,while a low value indicates dissimilarity. The results are shown in Fig. 8. Apart from traffic volume rank, degree and H-index are the two indicators that have the highest correlations with the operational impact measures. BC is the least predictive indicator of the operational importance of nodes. None of the topological indicators has a high a rank correlation with SWD impact as traffic volume.This reinforces the finding that the operational importance of airports is best predicted by traffic volume.Fig.8 also suggests that, in general, SWD rank correlations with operational impact are highest when the disruption is most severe.

    6. Conclusions

    The resilience of the air traffic network should be an important consideration in the design and management of air transport systems. Limited studies have been carried out to assess the resilience of the air traffic network considering the impact of disruptive events on air traffic operations. In this paper, we developed a number of operational resilience measures, and used them to assess the resilience of the air traffic network in China via simulation studies. Critical airports which have significant influence on the performance of whole system are identified. We also compared the operational resilience measures with the network topological indicators which are commonly used in existing studies of air traffic network robustness and reliability.

    Our simulation study captures key aspects of the process of how temporary capacity reductions disrupt the air transport system - from the original steady state to the disrupted state,the recovery state,and finally stabilizing at a new steady state.In addition, we explored the impact of two disruption factors on system resilience: the disruption scale (measured by the number of airports under disruption)and the disruption intensity (measured by minimum headway between flight operations). Both have a direct impact on the severity of system performance degradation and the time to recover. The marginal effect of these two factors becomes small when the system is approaching the limit of disruption tolerance. This could mean that as the two factors approach these limits,the system is driven to a state from which it cannot recover,or from which recovery would take a very long time. However, further work is required to develop a model that faithfully captures the mechanisms by which the air traffic network reaches this state.

    Our simulation model captures how the disruption at a given airport affects operations throughout the network.There is considerable variation among airports in the severity of this systemic impact. While topological measures such as degree and H-index are somewhat predictive of the impact of a disruption, the single best predictor and is also the simplest: the flight traffic volume at the airport. This challenges the use of complex network theory as a basis for assessing resilience in the context of air transport networks. However, there are a few airports, such as ZYTX and ZSFZ, at which operational disruptions have large systemic impacts that are not explained by any of the topological or activity measures that have been considered. Future research should explore whether there are other measures which, either individually or in combination,could account for these impacts.

    Although our study provides insights into the operational resilience of air traffic network, it has several limitations.The simulation scenarios could be enriched to consider more factors, such as baseline condition of the system, time of disruptive events,particular clusters of airports under disruption,etc. Also, the findings are not verified by real-world operational data.If one could collect empirical data of air traffic network undergoing disruptive events, and analyze the response of the air transport system in these natural experiments, our findings could be further validated.

    Future work will investigate strategies to improve the resilience of air traffic network.Optimization tools can be built to design the network, flight schedules, and mitigation strategies,such as flight cancellations,aircraft swaps,and traffic management initiatives, in order to better contain the operational impact of airport disruptions, and thus improve the resilience of the air transport networks.

    Acknowledgments

    This research was supported by the National Natural Science Foundation of China (Nos. 61773203, U1833126, 61304190),the Open Funds of Graduate Innovation Base (Lab) of Nanjing University of Aeronautics and Astronautics of China(No. kfjj20180703), the State Key Laboratory of Air Traffic Management System and Technology of China (No.SKLATM201707), and the Hong Kong Research Grant Council General Research Fund of China (No. 11209717).

    中国三级夫妇交换| 国产在线视频一区二区| 久久久精品94久久精品| 免费黄频网站在线观看国产| 国产精品国产三级国产av玫瑰| 伊人久久国产一区二区| 亚洲精品,欧美精品| 国产黄色视频一区二区在线观看| 成人毛片a级毛片在线播放| 在现免费观看毛片| 欧美日韩综合久久久久久| 国模一区二区三区四区视频| 免费看光身美女| 啦啦啦视频在线资源免费观看| 亚洲久久久国产精品| 在线观看美女被高潮喷水网站| 国产成人午夜福利电影在线观看| 97超视频在线观看视频| 日本欧美视频一区| 日韩av免费高清视频| 亚洲av中文字字幕乱码综合| 97在线视频观看| 欧美日本视频| 日本av手机在线免费观看| 色综合色国产| 国产免费一区二区三区四区乱码| 夜夜看夜夜爽夜夜摸| 精品视频人人做人人爽| 国产伦理片在线播放av一区| 国产乱人偷精品视频| 精品一区二区免费观看| 极品少妇高潮喷水抽搐| 97超视频在线观看视频| 国产欧美日韩精品一区二区| 日韩欧美 国产精品| 男人爽女人下面视频在线观看| av网站免费在线观看视频| 久久精品国产亚洲网站| 亚洲成人中文字幕在线播放| 内地一区二区视频在线| 18禁裸乳无遮挡动漫免费视频| 各种免费的搞黄视频| 国产亚洲午夜精品一区二区久久| 亚洲国产精品专区欧美| 亚洲精品视频女| 久久久久久久久久久免费av| 精品视频人人做人人爽| 午夜视频国产福利| 91精品伊人久久大香线蕉| 日韩欧美 国产精品| 18禁动态无遮挡网站| 最近最新中文字幕免费大全7| 80岁老熟妇乱子伦牲交| 国产av国产精品国产| 免费观看在线日韩| a级毛色黄片| 精品久久久久久电影网| 午夜福利网站1000一区二区三区| 亚洲精品亚洲一区二区| 少妇高潮的动态图| a级一级毛片免费在线观看| 久久久久久久大尺度免费视频| 在现免费观看毛片| 精品久久久久久久末码| 国产精品偷伦视频观看了| 少妇精品久久久久久久| 超碰av人人做人人爽久久| 一级毛片黄色毛片免费观看视频| 久久人人爽人人片av| 久久久成人免费电影| 80岁老熟妇乱子伦牲交| 亚洲婷婷狠狠爱综合网| 中文字幕av成人在线电影| 久久久久久久国产电影| 一边亲一边摸免费视频| 国产在线视频一区二区| 国产淫片久久久久久久久| av免费观看日本| 中文字幕久久专区| 久久精品久久精品一区二区三区| 制服丝袜香蕉在线| 91精品国产九色| 啦啦啦在线观看免费高清www| 精品少妇黑人巨大在线播放| 婷婷色综合www| 国产白丝娇喘喷水9色精品| 99九九线精品视频在线观看视频| kizo精华| 99久国产av精品国产电影| 深夜a级毛片| h视频一区二区三区| 毛片一级片免费看久久久久| 亚洲精品亚洲一区二区| 老师上课跳d突然被开到最大视频| 久久ye,这里只有精品| 亚洲电影在线观看av| 干丝袜人妻中文字幕| 国产亚洲欧美精品永久| 亚洲无线观看免费| 热99国产精品久久久久久7| 亚洲天堂av无毛| av专区在线播放| videossex国产| 欧美日韩视频精品一区| 大又大粗又爽又黄少妇毛片口| 午夜福利视频精品| 欧美亚洲 丝袜 人妻 在线| 久久久久久久亚洲中文字幕| 最近中文字幕高清免费大全6| 欧美精品一区二区大全| 精品亚洲乱码少妇综合久久| 日韩欧美 国产精品| 久久久久久久久久久丰满| 免费大片18禁| 日韩av不卡免费在线播放| 直男gayav资源| 日本爱情动作片www.在线观看| 午夜福利高清视频| av在线app专区| 国产黄片视频在线免费观看| 亚洲激情五月婷婷啪啪| 国产免费又黄又爽又色| 国产免费视频播放在线视频| 国产精品偷伦视频观看了| 在线观看免费视频网站a站| 18禁裸乳无遮挡免费网站照片| 日本色播在线视频| av视频免费观看在线观看| 国产精品嫩草影院av在线观看| 久久精品国产自在天天线| 尾随美女入室| 亚洲精品第二区| 又粗又硬又长又爽又黄的视频| 成人亚洲欧美一区二区av| 女的被弄到高潮叫床怎么办| 激情 狠狠 欧美| 在线观看av片永久免费下载| tube8黄色片| 亚洲综合精品二区| 国产精品秋霞免费鲁丝片| 岛国毛片在线播放| 婷婷色av中文字幕| 久久影院123| 亚洲激情五月婷婷啪啪| 久久久精品免费免费高清| 熟妇人妻不卡中文字幕| 国产日韩欧美在线精品| 国产一区有黄有色的免费视频| 国产乱人偷精品视频| 国产深夜福利视频在线观看| 久久久久久久国产电影| 性色avwww在线观看| 18禁裸乳无遮挡动漫免费视频| 亚洲国产高清在线一区二区三| 国产一区有黄有色的免费视频| 婷婷色综合www| 中文精品一卡2卡3卡4更新| 永久网站在线| 成年女人在线观看亚洲视频| 国产淫片久久久久久久久| 少妇的逼好多水| 边亲边吃奶的免费视频| 亚洲欧美精品专区久久| 亚洲天堂av无毛| 一区在线观看完整版| 日本色播在线视频| 日韩中字成人| 最黄视频免费看| 国产一区二区在线观看日韩| .国产精品久久| 少妇精品久久久久久久| 国产人妻一区二区三区在| 成人特级av手机在线观看| 91aial.com中文字幕在线观看| 国产探花极品一区二区| 18禁动态无遮挡网站| 亚洲成人手机| 18禁裸乳无遮挡动漫免费视频| 高清黄色对白视频在线免费看 | 精品久久久久久久末码| 99久久中文字幕三级久久日本| 精品国产一区二区三区久久久樱花 | 久久精品国产自在天天线| 久久女婷五月综合色啪小说| 最近最新中文字幕大全电影3| 男人和女人高潮做爰伦理| 国产一区亚洲一区在线观看| 我的老师免费观看完整版| 成人一区二区视频在线观看| 97精品久久久久久久久久精品| 久久影院123| 最新中文字幕久久久久| 欧美人与善性xxx| 老熟女久久久| 成年人午夜在线观看视频| 精品一区在线观看国产| 亚洲精品日韩av片在线观看| 国产精品.久久久| 亚洲人成网站高清观看| 国产片特级美女逼逼视频| 搡老乐熟女国产| 又大又黄又爽视频免费| 一级毛片黄色毛片免费观看视频| 国产淫语在线视频| 在线免费观看不下载黄p国产| 免费观看性生交大片5| www.色视频.com| 成人毛片60女人毛片免费| 亚洲欧美成人精品一区二区| 在线亚洲精品国产二区图片欧美 | 最近最新中文字幕大全电影3| 黄色视频在线播放观看不卡| 天美传媒精品一区二区| 日韩制服骚丝袜av| 国产永久视频网站| 内射极品少妇av片p| 久久精品国产鲁丝片午夜精品| 国产片特级美女逼逼视频| 久久人人爽av亚洲精品天堂 | 少妇 在线观看| 亚洲欧美日韩卡通动漫| xxx大片免费视频| 国产一区二区三区av在线| 精品国产露脸久久av麻豆| 亚洲精华国产精华液的使用体验| 欧美高清性xxxxhd video| 久久久久网色| 色视频在线一区二区三区| 一本色道久久久久久精品综合| 这个男人来自地球电影免费观看 | 国产乱来视频区| 午夜老司机福利剧场| 伦理电影免费视频| 国产精品av视频在线免费观看| 中文天堂在线官网| 精品酒店卫生间| 青春草视频在线免费观看| 久热久热在线精品观看| 在现免费观看毛片| 在线观看美女被高潮喷水网站| 汤姆久久久久久久影院中文字幕| 18+在线观看网站| 王馨瑶露胸无遮挡在线观看| 伦理电影免费视频| 久久久久久伊人网av| 久久99蜜桃精品久久| 亚洲欧美精品自产自拍| 亚洲欧美日韩东京热| 日韩电影二区| 高清毛片免费看| 一级黄片播放器| 午夜福利影视在线免费观看| 少妇裸体淫交视频免费看高清| 777米奇影视久久| 18禁在线播放成人免费| 午夜精品国产一区二区电影| 我的老师免费观看完整版| 美女主播在线视频| 五月玫瑰六月丁香| 青青草视频在线视频观看| 中文字幕精品免费在线观看视频 | 国产亚洲午夜精品一区二区久久| 久久久久久久久大av| 少妇人妻一区二区三区视频| 国产精品一区二区性色av| 网址你懂的国产日韩在线| 欧美老熟妇乱子伦牲交| 免费黄频网站在线观看国产| 亚洲精品国产av蜜桃| 97精品久久久久久久久久精品| 春色校园在线视频观看| 女人十人毛片免费观看3o分钟| videossex国产| 欧美高清性xxxxhd video| 少妇的逼水好多| 少妇人妻久久综合中文| 王馨瑶露胸无遮挡在线观看| 99热全是精品| 国产高清三级在线| 欧美+日韩+精品| 边亲边吃奶的免费视频| 亚洲av日韩在线播放| 肉色欧美久久久久久久蜜桃| 免费av不卡在线播放| 女性生殖器流出的白浆| 日韩国内少妇激情av| 日韩成人伦理影院| 久热这里只有精品99| 黄色欧美视频在线观看| 丝袜喷水一区| 久久精品国产鲁丝片午夜精品| 欧美日韩亚洲高清精品| 三级国产精品片| 日韩 亚洲 欧美在线| 久久久久网色| 热99国产精品久久久久久7| 午夜福利影视在线免费观看| 好男人视频免费观看在线| av在线播放精品| 观看美女的网站| 国产精品女同一区二区软件| 久久ye,这里只有精品| 欧美少妇被猛烈插入视频| 亚洲欧洲日产国产| 高清av免费在线| 国产亚洲91精品色在线| 精品午夜福利在线看| 国产精品人妻久久久影院| 18禁裸乳无遮挡动漫免费视频| 高清日韩中文字幕在线| 日本vs欧美在线观看视频 | av在线蜜桃| 日本黄大片高清| 亚洲欧美一区二区三区黑人 | 亚洲av男天堂| 成人国产av品久久久| 亚洲精品日韩av片在线观看| 国产精品无大码| 免费观看a级毛片全部| 哪个播放器可以免费观看大片| 亚洲熟女精品中文字幕| 少妇人妻一区二区三区视频| 成人黄色视频免费在线看| 国产一区二区三区av在线| 亚洲欧美日韩无卡精品| 成人黄色视频免费在线看| 日本午夜av视频| 久久久久精品久久久久真实原创| 亚洲欧洲国产日韩| 国产伦精品一区二区三区视频9| 亚洲国产av新网站| 久久久成人免费电影| 亚洲精品乱久久久久久| 久久人人爽人人爽人人片va| 夫妻午夜视频| 美女xxoo啪啪120秒动态图| 久久人人爽av亚洲精品天堂 | 中文字幕久久专区| 久久久精品免费免费高清| 亚洲欧美成人精品一区二区| 99re6热这里在线精品视频| 精品少妇黑人巨大在线播放| 精品酒店卫生间| 欧美日韩综合久久久久久| 亚洲第一区二区三区不卡| 三级国产精品欧美在线观看| 国产爱豆传媒在线观看| 午夜福利在线在线| av免费观看日本| 亚洲综合色惰| 高清av免费在线| 我要看日韩黄色一级片| 人人妻人人添人人爽欧美一区卜 | 国产老妇伦熟女老妇高清| 99热6这里只有精品| 国产精品久久久久久精品古装| 久久久久久久国产电影| 最近最新中文字幕大全电影3| 欧美区成人在线视频| 国产亚洲欧美精品永久| 啦啦啦啦在线视频资源| 久久人妻熟女aⅴ| 国产成人精品久久久久久| 亚洲av中文av极速乱| 五月伊人婷婷丁香| 久久久成人免费电影| 国产国拍精品亚洲av在线观看| 亚洲av电影在线观看一区二区三区| 亚洲欧美日韩卡通动漫| 黑人高潮一二区| 日韩强制内射视频| 亚洲aⅴ乱码一区二区在线播放| 亚洲国产最新在线播放| 男人爽女人下面视频在线观看| 国产亚洲一区二区精品| 久久精品国产亚洲av天美| 国产日韩欧美在线精品| 熟女av电影| 国产高清不卡午夜福利| 日韩一本色道免费dvd| 国产一区二区在线观看日韩| 黑人高潮一二区| 成人漫画全彩无遮挡| 久久午夜福利片| 国产美女午夜福利| 免费黄网站久久成人精品| 日本黄大片高清| 美女cb高潮喷水在线观看| 黄色欧美视频在线观看| 国产女主播在线喷水免费视频网站| 国产亚洲91精品色在线| 亚洲欧美中文字幕日韩二区| a级一级毛片免费在线观看| 午夜福利在线在线| 国产精品精品国产色婷婷| 一区二区三区乱码不卡18| 国产乱来视频区| 亚洲av成人精品一区久久| 十八禁网站网址无遮挡 | 在线播放无遮挡| 精品国产一区二区三区久久久樱花 | 亚洲欧美成人精品一区二区| 国产熟女欧美一区二区| 黄片无遮挡物在线观看| 777米奇影视久久| 日日摸夜夜添夜夜爱| 人妻少妇偷人精品九色| 国产淫语在线视频| 精品久久久久久电影网| 五月天丁香电影| 99久久精品一区二区三区| 婷婷色麻豆天堂久久| 观看美女的网站| 欧美高清性xxxxhd video| 熟女电影av网| 亚洲欧洲日产国产| 欧美 日韩 精品 国产| 亚洲精品一二三| 久久久久久九九精品二区国产| www.色视频.com| 国产高清有码在线观看视频| 最近2019中文字幕mv第一页| 中文在线观看免费www的网站| 不卡视频在线观看欧美| 亚洲精品亚洲一区二区| 大码成人一级视频| 少妇猛男粗大的猛烈进出视频| 麻豆乱淫一区二区| 国产成人a区在线观看| 亚洲成色77777| 国产 一区精品| 国国产精品蜜臀av免费| av卡一久久| 男人狂女人下面高潮的视频| av专区在线播放| 亚洲第一区二区三区不卡| 亚洲欧美成人综合另类久久久| 成人18禁高潮啪啪吃奶动态图 | 国产成人aa在线观看| 中文在线观看免费www的网站| 日韩,欧美,国产一区二区三区| 这个男人来自地球电影免费观看 | 国产av码专区亚洲av| 少妇被粗大猛烈的视频| 丰满迷人的少妇在线观看| 人妻少妇偷人精品九色| 亚洲人成网站高清观看| 大陆偷拍与自拍| 日本wwww免费看| 久久青草综合色| 妹子高潮喷水视频| 亚洲精华国产精华液的使用体验| 日韩欧美 国产精品| 精品午夜福利在线看| 国产精品一及| 久久影院123| 亚洲精品成人av观看孕妇| 日韩一区二区视频免费看| 国产免费一级a男人的天堂| 亚洲精品乱久久久久久| 午夜老司机福利剧场| 久久精品国产亚洲av天美| 五月开心婷婷网| 哪个播放器可以免费观看大片| 免费大片18禁| 色视频www国产| 人人妻人人爽人人添夜夜欢视频 | 国产一级毛片在线| 亚洲国产精品999| 只有这里有精品99| 欧美成人a在线观看| 亚洲国产日韩一区二区| av卡一久久| 免费大片黄手机在线观看| 日本av手机在线免费观看| 内射极品少妇av片p| 午夜免费鲁丝| 蜜桃在线观看..| 亚洲真实伦在线观看| 国产视频首页在线观看| 亚洲人与动物交配视频| 99久久精品一区二区三区| 22中文网久久字幕| 精品久久国产蜜桃| 久久精品夜色国产| 97超碰精品成人国产| 一区二区三区精品91| 午夜免费鲁丝| 午夜精品国产一区二区电影| 国产白丝娇喘喷水9色精品| 亚洲国产av新网站| 亚洲综合精品二区| 精品久久久噜噜| av网站免费在线观看视频| 精品一区在线观看国产| 久久99热6这里只有精品| 亚洲在久久综合| 婷婷色综合大香蕉| 午夜福利视频精品| videos熟女内射| 国产男女超爽视频在线观看| 亚洲精品乱久久久久久| 精品一区二区三区视频在线| 成人午夜精彩视频在线观看| 国产黄色免费在线视频| 午夜免费鲁丝| 亚洲美女搞黄在线观看| 肉色欧美久久久久久久蜜桃| 国产男女内射视频| 秋霞在线观看毛片| 亚洲最大成人中文| 日日撸夜夜添| 人妻少妇偷人精品九色| 日日摸夜夜添夜夜添av毛片| 美女主播在线视频| 国语对白做爰xxxⅹ性视频网站| 青春草视频在线免费观看| 日产精品乱码卡一卡2卡三| 色视频在线一区二区三区| 高清毛片免费看| 日本黄大片高清| 激情 狠狠 欧美| 国产爽快片一区二区三区| 精品熟女少妇av免费看| 男女边吃奶边做爰视频| 国产精品伦人一区二区| 久久国产精品大桥未久av | 亚洲精品亚洲一区二区| 久久97久久精品| 在线观看一区二区三区激情| 91午夜精品亚洲一区二区三区| 免费观看性生交大片5| 国产 一区 欧美 日韩| 亚洲av中文av极速乱| 一二三四中文在线观看免费高清| av在线老鸭窝| 国产精品偷伦视频观看了| 美女内射精品一级片tv| 国产欧美日韩精品一区二区| 哪个播放器可以免费观看大片| 18禁裸乳无遮挡免费网站照片| a级一级毛片免费在线观看| 美女主播在线视频| 国产精品嫩草影院av在线观看| 国产伦在线观看视频一区| 久久久久久久久大av| 亚洲精品成人av观看孕妇| 你懂的网址亚洲精品在线观看| 国产免费一区二区三区四区乱码| 亚洲电影在线观看av| 亚洲婷婷狠狠爱综合网| 亚洲中文av在线| 午夜免费鲁丝| 色5月婷婷丁香| 丰满乱子伦码专区| 青春草视频在线免费观看| 免费不卡的大黄色大毛片视频在线观看| 丝瓜视频免费看黄片| 99热网站在线观看| 国产69精品久久久久777片| 欧美日韩一区二区视频在线观看视频在线| 久久影院123| 国产色婷婷99| 久久精品久久久久久噜噜老黄| 久久久久精品性色| 成人免费观看视频高清| 国产高清国产精品国产三级 | 日本与韩国留学比较| 男人和女人高潮做爰伦理| 伦精品一区二区三区| 2018国产大陆天天弄谢| 老熟女久久久| 国产午夜精品久久久久久一区二区三区| 春色校园在线视频观看| 不卡视频在线观看欧美| 男女边吃奶边做爰视频| 亚洲av在线观看美女高潮| 亚洲真实伦在线观看| 老司机影院成人| 三级经典国产精品| 啦啦啦视频在线资源免费观看| 欧美区成人在线视频| 色婷婷久久久亚洲欧美| 男女边摸边吃奶| av不卡在线播放| 亚洲av日韩在线播放| 国产精品久久久久久精品电影小说 | 一区在线观看完整版| 亚洲精品中文字幕在线视频 | 十八禁网站网址无遮挡 | 青青草视频在线视频观看| 人妻 亚洲 视频| 一本色道久久久久久精品综合| 国产爽快片一区二区三区| 国产精品无大码| 亚洲精品成人av观看孕妇| 人妻夜夜爽99麻豆av| 国产免费一区二区三区四区乱码| 伊人久久精品亚洲午夜| 搡老乐熟女国产| 久久久精品免费免费高清| 自拍欧美九色日韩亚洲蝌蚪91 | 最近2019中文字幕mv第一页| 精品熟女少妇av免费看| 97热精品久久久久久| 网址你懂的国产日韩在线| 嫩草影院新地址| 国产成人一区二区在线| 亚洲精品国产色婷婷电影| 欧美97在线视频| 一本色道久久久久久精品综合| 久久精品国产亚洲av天美| 3wmmmm亚洲av在线观看| 国产无遮挡羞羞视频在线观看| 汤姆久久久久久久影院中文字幕|