Invited Paper - Simulation and Modeling Methodologies, Technologies and Application - page 3

charging takes considerable time and currently range is more limited than with con-
ventional vehicles. It results in range anxiety and hampers customer acceptance of the
product.
To alleviate range anxiety, new vehicle electronics features are needed to help ve-
hicle operators make driving choices that avoid discharged battery situations, extend
vehicle range, and combine charging with other good uses of time. Development of
these features requires practical meta-models that can accurately predict energy con-
sumption on the public roads.
Building meta-models from field-test vehicle data requires statistical regression of
public-road vehicle data (PRVD) over very large geographic areas. At present; there
are not enough production test vehicles available to collect a sufficient amount of
data, noise factors are not well controlled, and data collection is too time consuming
to support product launch. As a result modeling and simulation are essential tools in
analysis of BEV performance.
In this work we propose implementation of traffic simulation combined with pro-
pulsion modeling for determining electric vehicle energy consumption. We use traffic
micro-simulation to create surrogate PRVD data that has many of the properties of
actual PRVD data, specifically capturing the stochastic nature of vehicles moving
through roads with traffic. The surrogate data is analyzed using propulsion simulation
to estimate the amount of energy the vehicles will consume in a specific driving ma-
neuver to derive statistical information.

2 Simulating Energy Consumption

The simulation approach used here begins with geographical information about roads
and basic parameters about the vehicle of interest. The road data is coded into a mi-
croscopic traffic simulation program in which model vehicles are input into the simu-
lation and surrogate drive data is produced. This is input into a propulsion modeling
system that outputs the energy consumption of individual vehicles in the model. This
data is collected and analyzed through statistical regression. The resulting energy
consumption data is used to calibrate an energy consumption meta-model that can be
used to estimate the energy consumption of a vehicle using surrogate data from the
traffic simulation.
Scenarios of different road networks under different external conditions are
simulated. For example, a road network could consist of a highway interchange or a
stretch of a specific kind of road. External conditions would be such things as
topography, weather, and traffic load. The traffic simulation contains BEV vehicle
and other vehicles that create the traffic conditions for the sample vehicles.
Regression analysis is used to create an energy consumption meta-model that predicts
average energy consumption as well as the stochastic distribution. By comparing
scenarios it is possible to determine main effects to build a meta-model of energy
consumption from geographical data and data available in traveller information
systems.
Advantages of the method include developing energy consumption models during
the design process before physical testing is possible. These models can be used to
improve the design and support the deployment of BEV vehicles into the consumer