Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
Multiscale Model for PedestrianDynamics
–Review and Implementation
Niklaus Leuenberger – ETH Zurich – MIT Exchange Student
Numerical Fluid Mechanics 2.29 – Spring 2019
05/15/19 Niklaus Leuenberger
Pedestrian Crossing Intersection
Photo by Ryoji Iwata on Unsplash
https://unsplash.com/photos/IBaVuZsJJTo?utm_source=unsplash&utm_medium=referral&utm_content=creditCopyTexthttps://unsplash.com/search/photos/traffic?utm_source=unsplash&utm_medium=referral&utm_content=creditCopyText
References
• Cristiani, Emiliano, Benedetto Piccoli, and Andrea Tosin. Multiscale modeling of pedestrian dynamics. Vol. 12. Springer, 2014.
05/15/19 Niklaus Leuenberger
Connection to Fluid Dynamics
05/15/19 Niklaus Leuenberger
Continuum Fluid Mechanics Rarefied Gas Dynamics Molecular Dynamics
Macroscopic Modeling Multiscale Model Microscopic Model
Governing Equations
05/15/19 Niklaus Leuenberger
Measure at time t.
Number of Pedstrians in E
at time t.
Discretization
05/15/19 Niklaus Leuenberger
Pedestrians, track the…
• position
• velocity
Lagrangian representation
Cells, has…
• velocity of a cell
• density in a cell
Eulerian representation
Discretization
05/15/19 Niklaus Leuenberger
Pedestrians, track the…
• position
• velocity
Lagrangian representation
Cells, has…
• velocity of a cell
• density in a cell
Eulerian representation
mic
rosco
pic
macro
sco
pic
Multiscale Level – Governing Equations
05/15/19 Niklaus Leuenberger
Multiscale level
Measure
Equation
microscopic macroscopic
Governing Equations
05/15/19 Niklaus Leuenberger
microscopic
macroscopic
Desired Velocity
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Multiscale Level – Governing Equations
05/15/19 Niklaus Leuenberger
Multiscale level
Measure
Equation
microscopic macroscopic
Update Position and Density
Microscopicwe update the position of each pedestrian
Macroscopic
we update the density value at each cell center
05/15/19 Niklaus Leuenberger
Result
05/15/19 Niklaus Leuenberger
Future Work
• Fix the bugs in the code.
• Run test cases to test the model qualitatively and quantitatively.
• Efficient implementation of the algorithm
• Extend the code to multiple groups of pedestrians• Crossing Flows
• Flow of a small group of pedestrians through a large crowd
05/15/19 Niklaus Leuenberger
Backup Slides
05/15/19 Niklaus Leuenberger
Algorithm
05/15/19 Niklaus Leuenberger
%% PSEUDOCODE
initialize position of pedestrians
define desired velocityfield
initialize density
% timeloop
for t=1:Nt
% Calculate the velocity couplings
v_micro_for_micro
v_micro_for_macro
v_macro_for_macro
v_macro_for_micro
% Aggregate the velocities
v_micro = theta* v_micro_for_micro + (1-theta)*lambda*v_macro_for_micro+v_desired
v_macro = theta* v_micro_for_macro + (1-theta)*lambda*v_macro_for_macro+v_desired
% Update Positions
X = X + dt*v_micro
% Update Density
update rho
end
Macroscopic Level
05/15/19 Niklaus Leuenberger
Macroscopic level
Density
Measure
Equation
Governing Equations
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Governing Equations
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger
Update Density – 1D Model
05/15/19 Niklaus Leuenberger