VISSIM 5.30-05 User Manual

Simulated Travel Time and General Cost
To model a growing experience of travel times the times not only from the
immediately preceding iteration should be considered but from all preceding
iterations.
► However, if you want the older measurements to have less influence, the
appropriate mathematical method for that is exponential smoothing. The
user-defined smoothing factor indicates the relative weight of the respective
recent iteration.
► Alternatively, MSA (Method of Successive Averages) can be applied
which assigns as much weight to any recent iteration as to the current
one by forming the arithmetic mean from all iterations. This way, the
influence of any further iteration is reduced. As long as no *.BEW file is
available, the parameter value is set automatically in VISSIM (not userdefined),
if this option is active. In case of an existing *.BEW file the user
has to enter the number of iterations calculated for this file. Here, also a
value below the actual number of current iterations can be entered to
increase the influence of the next iterations. This is recommended, if the
path evaluation states that the measured travel times deviate heavily
from expected travel times, cf. section 11.21.
The appropriate method can be selected and the parameters can be set via
TRAFFIC - DYNAMIC ASSIGNMENT via EXTENDED in the Dynamic Assignment
window, if option Store costs is active.
If option Store costs is active, the expected travel times for the next iteration
are stored in the VISSIM cost file (*.BEW) after an iteration of the simulation
in order to provide a base for the route choice in the next iteration.
Exponential Smoothing
If we get a new set of measured values, we compute the smoothed travel
time as the weighted sum of
► the old smoothed value (from previous iterations) and
► the newly measured value (from current iteration).
That smoothed value represents the travel time that we expect in the next
iteration.
Formally:
T
n,
κ
( 1 α ) T
n 1,
κ
α TO
n, κ
i
= − ⋅
−
i
+ ⋅
i
where
K = index of the evaluation interval within the simulation period
n = index of the assignment iteration
i = index of the edge
User Manual © PTV AG 2011 621