[PUBLISHER
HEADER HERE]
[PUBLISHER HEAD-
ER HERE]
ICEM 2012
Slower is faster: the influence of departure time distribution
on the overall evacuation performance
Gregor Lämmel
a*
, Hubert Klüpfel
b
a
TU Berlin, Transport Systems Planning and Transport Telematics, Salzufer 17-19, 10587 Berlin, Germany
b
TraffGo HT GmbH, Bismarckstraße 142a, 47057 Duisburg, Germany
Abstract
In this paper we present results for evacuation simulations with different departure time distributions. One scenario is
immediate departure, i.e. all evacuees get informed at the same time and start evacuation immediately after having
received the evacuation order. The second scenario is based on a log-normal departure time distribution. The compar-
ison of the results for both scenarios shows that a later departure (log-normal departure time distribution) leads to a
better overall evacuation performance, i.e. a decrease in total evacuation time. The reason is the lower demand on the
road network for a longer departure time span. This leads to less congestion since the demand is below the capacity
and the density below the critical density and certain roads are in the laminar or capacity regime. For the immediate
departure, the traffic demand is higher and part of the road network will be in the congested regime where the flow
breaks down and the overall time is higher despite the fact that the departure time (i.e. the time between the alarm and
the departure) is shorter. An application of this result might either be to meter the inflow on main evacuation routes
(similar to ramp metering for freeway traffic during rush hours) or to implement a phased evacuation strategy. We do
not go into detail here, since the focus of the paper is on the simulation of the phenomenon described (“slower is
faster”) with MATSim (a queuing model). In order to reproduce the effect, the queuing algorithm had to be adapted
to reproduce the flow density relation (fundamental diagram) for the congested regime more realistically than is the
case in the standard queuing model. The results presented are part of an ongoing research project (Gis based risk
assessment, information, and planning system for regional evacuation – GRIPS). The different strategies and factors
that might influence the departure time distribution will be investigated further within this research project and the
first results presented here will be scrutinized in more detail in the future.
© 2011 Published by [] Ltd. Selection and/or peer-review under responsibility of [name organizer]
Keywords: Evacuation Simulation; Multi-Agent Simulation; MATSim, Departure Time Distribution
*
Corresponding author. Tel.: +49-30-314-21376; fax: +49-30-314-26269, E-mail address: laemmel@vsp.tu-berlin.de.