FEA Based Level 3 Assessment of Deformed Tanks with Fluid Induced Loads
DPG Berlin - SOE 18 - talk v1.2.4
1. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Simulation of dynamical information
spread during pedestrian evacuation
M. Kämpf, C. Napierala, J.W. Kantelhardt
DPG Frühjahrstagung 2012 - 29. März - Berlin
Martin-Luther Universität Halle-Wittenberg
Institut für Physik
FG für Theoretische Physik
SOE 18.1
2. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Measurement of Crowd Properties :
London - Lord Mayors Show 2011
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
3. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Introduction :
• We study crowd dynamics
and information flow during
evacuation processes.
• We are focused on the flow
of information and on
quality of information in a
system without any
centralized communication
and control infrastructure.
• We want to meassure, what
can be known inside such a
self organized system at any
time and place and how this
knowledge can be distributed
efficently.
Lord Mayor Show
London - 2011
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
4. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Extended Properties
Dynamic Lattice Structure
Social Network
Communication Network
• all elements have the
same properties
• only simple interactions
• no interaction with the
area, accept of walls
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
5. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
A Complex System ?
• Such a system consists of fairly simple components.
• We can not analyze the properties of
agent-structure-information-interactions
by decoupling these aspects.
• As it is a Complex System - we are looking
for emerging properties / phenomena which
are not obviously visible by looking
on isolated parts.
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
6. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Our approach :
A lattice gas model is combined with
an agent based simulation
• simple geometry, extendable to real
world like structures
• agents movement is limited to 6
directions in space and to one
movement per time step
• agents move according to simple rules
based on properties of the area and
their knowledge
• information flow is modeled as an
additional layer on top of the agents'
movement based on epidemic models
Density: 3%, 9%,
30% 95%
restricted
area
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
7. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Context of our Research :
• Real world data sets for comparison of results
–Evacuation exercise at University Krakow
–Note Bianca in Malta
–Lord Mayor's Show London
• Comparison with epidemic models
–SIR and SIS model
• Numerical Simulations in a discrete space
–compared with results from continous space model
• Construction of interconnected networks
–Building structure, and ad-hoc connected agents
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
8. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
S = susceptible, I = infectious, R= Recovered
• Valid for a large numbers of individuals
• Nonlinear system
• No generic analytical solution
Epidemic Models : SIR & SIS Model
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
9. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Meassurement of Infection Ratio
InfectionRatio
InfectionRate
We count the number of infected
(or: informed) agents
For an intermediate fit interval
the average slope is determined
numericaly
10. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
First results : Rate of Infection (low density)
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
11. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
First results : Rate of Infection (high density)
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
12. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Conclusion:
• Information flow has a strong dependence
on the density of agents with communi-
cation devices, but hardly depends on the
structure of the building.
• Different communication strategies
influence the information flow more
than a restricted geometry.
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
13. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Quality of Information :
View of an agent :
How does the quality of
information available
to an agent evolve in time?
General view :
What amount of truth can be
known at a certain point in
space and time?
What parameters affect this
information?
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
14. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Outlook: Interconnected Networks
From "Lattice Gas Model" to a Network
–collective properties of all places in
a stairway define the edge properties
–possible flow on a crossing point
defines node properties
Combined Networks
–lane-network and communication network are inter connected
by the agents which are walking within the lane network but
rely on the information transmitted through the
communication network.
SOE 18.1 - Simulation of dynamical information spread during pedestrian evacuation
15. Martin-Luther Universität Halle-Wittenberg · Institut für Physik · FG Theoretische Physik
Mirko Kämpf
Martin Luther Universität Halle-Wittenberg
Naturwissenschaftliche Fakultät II
Institut für Physik
Von-Seckendorff-Platz 1
06120 Halle
Germany
phone: (+49) 345 55 25447
fax: (+49) 345 55 25446
e-mail: mirko.kaempf@physik.uni-halle.de
Editor's Notes
Abstract:
The collecting and spreading of up-to-date information is crucial for optimizing the evacuation of pedestrian crowds from buildings or large gatherings. If no global communication system is operative (e.g., during emergency situations), information might be collected automatically by people’s future smart-phones (e.g., by accelerometers) and spread using local ad-hoc communication between them. To simulate and characterize the dynamics of such information spread, we combine an agent-based lattice-gas simulation of the evacuation process with a network analysis approach. The model system is a combination of two coupled networks, one representing the paths in the building and one representing the communication network between the agents’ smart-phones. We study how fast a change of the system’s state is recognized, how fast this information is spread and what quality of information is available at a certain location and at a certain time, depending on several parameters. We also compare our results with well-known epidemic infection models (SIR and SIS) and with previous results of simulations in an unrestricted geometry. Furthermore, we analyze the time evolution of the agent’s information status to find out if we can detect and avoid interference effects.
Radiorange erklaeren!
Spread of new information can be compared to the spread of an infection.
But right now we just look on the simplest model … to find out, what parameters do
Affect the infection rate.
The quality of information is reduced if it is not longer accurate or too old.
This has to be handled like a transition to the uninformed phase.