Environmental Engineering Reference
In-Depth Information
and electrical systems into a single analytical framework. The ABM keeps track of
energy consumption vehicles have at each time interval as well as their location in
the network, and this valuable temporal and spatial data provided by the ABM is then
employed as a load forecast by the TCOPF program. This way the bottom-up ABM
perspective, representing vehicle owners and their travel behaviour, can be combined
with the top-down view from utilities and other stakeholders which can overview net-
work conditions and therefore optimise at system-level vehicle charging over time.
Linking the two models addresses a key issue in the smart-grid jigsaw puzzle by
beginning to study the interrelationship between transport and power sectors.
7.2.1 Agent-based model for vehicles
The ABM describes the behaviour of owners of EVs, modelled as autonomous
agents, and their activities bring the agent to various parts of the city at different
times of day. These activities range between work and leisure activities and, for exam-
ple, include going to work, meeting friends or going to shops. The model keeps track
of the position of each vehicle for each time interval analysed as well as the state of
charge (SOC) in the battery which is discharged according to the travels of the agent
following the characteristics of the vehicle.
The first input of the ABM model consists of driver profiles, defining the agent
features which populate an urban area. These profiles can include, for instance, people
with jobs, those who have to bring children to school, pensioners, etc. Each driver
has a number of characteristics, such as home location, work place (if applicable)
and number of children. A number of profiles can be created and 'cloned' following
a certain distribution to populate a large area with any number of PHEV owners
the user decides. Additionally, the characteristics of the fleet of vehicles used in the
model are given as an input, including the battery capacity and travel efficiency in
kilowatt-hour per kilometre terms.
Furthermore, the model requires a city layout as input. This data is read from
geographical Information System (GIS) shape files with the coordinates of build-
ings, roads and energy service networks. The model uses the city layout to determine
the journeys of the agents based on their activities as well as the nearest substation
for when a vehicle is plugged into the grid. This PHEV information calculated for
each time interval is key to incorporate the previously unknown temporal and spatial
elements to the optimal power flow problem.
The activities performed by the agents in the model can vary from one another
depending on their particular profiles as well as the time of day and the day of the
week being assessed. For instance, those with a job drive to their work place while
those who have kids drop them off at school at specific times. Meanwhile, other activ-
ities can be planned around these core tasks, representing constraints the agents need
to fulfil. Random distributions are employed to determine the individual departure
times as well as the type and location of the next activity.
An assumption of the model is that when an agent is not driving, the vehicle is
plugged to the network and connected to the charging infrastructure. The location of
the nearest substation is determined based on where the vehicle is parked. At each
Search WWH ::
Custom Search