managing passenger handling at airport terminals · faculty of transportation and traffic sciences...

Faculty of Transportation and Traffic Sciences „Friedrich List“, Chair of Air Transport Technology and Logistics

Michael Schultz and Hartmut Fricke

9th ATM Seminar, Berlin, 15.06.2011

Managing Passenger Handling at Airport Terminalsindividual-based model for stochastic passenger behavior


| |

Structure

Structure

motivation

stochastic movement modeldevelopment and adaptation of modelvalidation using the fundamental diagram (speed-density correlation)navigation inside virtual environments

airport terminalpassenger and handling processesdata collection at Dresden and Stuttgart airport

implementationdevelopment of an application environmentproof of concept

summary and outlook

Michael Schultz Managing Passenger Handling at Airport Terminals

1


| |

Structure

MotivationMotivation


modeling of individual movements to ensure– reliable optimization of traffic and handling infrastructures– sustainable configuration of guidance systems

high variety of scenarios of model application, supported by high performance of current computer architectures

demand for adequate approaches to model individual passenger behavior– delay caused by terminal processes significantly influences performance– e.g. security check (5-12%), baggage (2%), dep. gates and boarding (5-8%)

application area: airport terminals– optimization often based on staff experience– scientific models of passenger behavior barely used

2


| |

Structure

Motivation

- Dynamic Behavior

Dynamic Movement Behavior

dynamic effects caused byindividual characteristicsdecentralized decisions and control

self driven agents with capability to independent problem solving

self organizationspontaneous patternnon-standard physicsno external rules given, only localinteractions definedhigher order of system leads toefficient use of infrastructure


3


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

Self-Organization - Formation of Bi-Directional Lanes


4


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Microscopic Model


social force model– bi-directional attraction and repulsion forces, depending on distance, direction,

relative speed and group constellation– additional contact forces within highly congested areas– continuously in time and space

cellular automat– discrete, grid-based model (cell size equals to minimum space requirement)– transition probability to adjacent cells– simple exclusion - only free cells could be used by agents

convergence of approaches: discretization to speed up model implementation vs. use of conceptual analogies

reliable passenger model demands for stochastic model (fast performing) approach

5


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

Development of Stochastic Model - Cellular Automat

stochastic model defined by specific transition matrix M = p . q3 transition states for both horizontal and vertical {-1, 0, +1}

simplistic approach for p and q

no backward movements

symmetric variance


p q M

6


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

Development of Stochastic Model (2)

horizontal and vertical probability is only 1,5-dimensional (variance movement)

turning the transition matrix(re-indexing) to cope withdiagonal movements

superposition of horizontal and diagonal matrix for 2D movements


7


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

Model Adaptation

sample configuration= 15= 0,8= 0,2

investigation of the model points outexpected value of transition matrix deviate form movement vectoradditional dependence of movementangle regarding to

• variance and• speed

demand for efficient compensation: correction functions


= 13,92‘ = 15,39‘ = 0,78‘ = 0,072

8


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

Interactions Between Agents

stochastic model defines the movement of one agent

consideration of collisions– random shuffled update– crossings not allowed

additional set of parameter– movement strategy

• stochastic choice of occupied cell: wait vs. move

– amount of movements per simulation round• inverstigated band width: 1-5 steps

– movement trace• all temporally used cells are occupied during current simulation round


9


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

Validation - Fundamental Diagram (I)


10


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

Validation - Fundamental Diagram (II)

stochastic transition model fulfill standard criterion for valid movement approach


11

Transition area3-4 steps per round

3 steps4 steps


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Navigation - Continuous Movement Direction

navigation in complex environmentdemands for efficient algorithms

automatically created navigation points as a basis for a network structure


12

magnified


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Navigation - Grid-Based Movement Direction (I)

primary approach for navigationto one specific goal

flood fill, mark the target area(cell) and step wise fill all adjacent cells of the gridcorresponding distance metricconsists of artifacts

resulting metric:cells located on the axes ofsymmetry are correctly calculatedother cells possesses a higher distance regarding tothe Euclidian reference


13


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Navigation - Grid-Based Movement Direction (II)

deduction of direction fieldcharacteristics dependson the sequence of calculated cellscomplementary fieldsoccur by counterrotatingcreation

combination of complementarydirection fields leads to correctdirection field regarding to the Euclidean distance


14


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

Passengers at Airport Terminals

airport terminal as main infrastructure for handling passengersaggregation of different traffic modes (modal split, hub-and-spoke)complex spatial process arrangements and guidancehighest security requirements in traffic sector

fundamental movement decisions of passengers based on:remaining time to departureindividual experiences and expectationstravel motivationavailable information


15

arrival check-in security boarding

38 minutes 38 minutes 73 minutes

walking in public area > 5 mindirectly to security control

time to departure (min)


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

Data Collection - Video-Based Passenger Tracking (DRS)

recording and analysis of movement trajectoriessegmentation of pictureadaptation of lightinganalysis of maskingdetermination of positionidentification of passengers

people trackerprototype applicationtrajectory extractionmanual correctionexport of passenger relatedtrajectories

16


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

Data Collection - Video-Based Passenger Tracking, Results

results

women vs. men

tourist vs. business

size of the group


17

speed (m/s)

group of 3 group of 1

measurementnormal distr.

measurementnormal distr.


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

- Processes

process time (s)

Weibull ( =2 , =20, t =15s)Weibull ( =1.5, =45, t =20s)Weibull ( =2 , =80, t =25s)

Data Collection - Process Times (STR)

focused on check-in counterarrival distributionsamount of group members andpieces of baggagepersonnel (level of experience)bulky baggagehandling disturbances

functional adaptation of measurementsstochastic distributions

2 test as quality index


18


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

- Processes

Implementation

development of software-prototypegoal: proof of concept of passenger movement and handling modelvisualizationmodel- and scenario analysesderive functional/infrastructural

(VCTP)Java-

comprehensive interface definitions

virtual terminal environmentgeometry of terminal infrastructureflight plan, passenger characteristics

Implementation - Application Environment


19


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

- Processes

Implementation

Implementation - Applications

investigations of boarding proceduresoutside-in, back-to-front, block, randomanalyses of variance- and significanceaircraft used: A320, B777, and A380

validation tests at Stuttgart airportdetermination of specific process behaviorvariation of amount of check-in counterand security lanes

passenger related process evaluationrelation of waiting to remaining time usingDresden airport as referenceanalysis of check-in and security control


20

remaining time to departure (min)


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

- Processes

Implementation

Summary andOutlook

Summary and Outlook

summarydevelopment of a reliable stochastic movement modeloptimized flood fill algorithmdata collections at airport terminals provide a solid base for calibration of passenger and handling process characteristicsvalidation of the model approach against specific airport scenariosvisual computing tools platform as versatile application environment for multi-agent simulations, people tracker as validation tool

outlookconsideration of human perception (signage, information flow, guidance)navigation of passengers considering partial informationmodel of information reception and decision making (cue driven approach)group dynamic effectsevacuation simulation


21


| |

Structure

Motivation

- Dynamic Behavior

- Self-Organization

- Microscopic Model

Model

- Development

- Adaptation

- Interaction

- Validation

- Navigation

Terminal

- Passenger

- Processes

Implementation

Summary andOutlook

Don’t panic!

Thank you.

contactMichael Schultz

fon: +49 351 46339446mail: [email protected]


22

managing passenger handling at airport terminals · faculty of transportation and traffic sciences...

Documents