• Electric

Overview#

Since version 0.24.0 SUMO includes a model for electric vehicles. It was implemented by Tamás Kurczveil and Pablo Álvarez López from the TU-Braunschweig. The core of the model is implemented in the vehicle device device.battery. Additional features are a charging station (which can be placed on any lane in the network) and a new output option --battery-output <FILE>.

You can find a test case for these implementations at [1]

Defining Electric Vehicles#

Different aspects of electric vehicles are modeled separately. This page puts the focus on modeling the battery and how it is charged and discharged. The actual energy consumption values themselves are part of the emission modelling because SUMO can use different models for that.

To track the charging status of a vehicle, it must be equipped with a battery device. This may be done using the option --device.battery.explicit <vehID1,vehID2,...> or simply setting --device.battery.probability 1 to equip all vehicles. Alternatively, the device may be specified using Generic vehicle parameters.

Additional properties of the vehicle and its electrical components must then be defined via parameters of the vehicle or its type. Some property can only be defined for the vehicle type.

These values have the following meanings (the defaults are from the Kia below):

An example of a vehicle with electric attribute (those are the values for a city bus from the original publication):

<routes>
    <vType id="ElectricBus" accel="1.0" decel="1.0" length="12" maxSpeed="100.0" sigma="0.0" minGap="2.5" mass="10000" color="1,1,1">
        <param key="has.battery.device" value="true"/>
        <param key="device.battery.capacity" value="2000"/>
        <param key="maximumPower" value="1000"/>
        <param key="frontSurfaceArea" value="5"/>
        <param key="airDragCoefficient" value="0.6"/>
        <param key="rotatingMass" value="100"/>
        <param key="radialDragCoefficient" value="0.5"/>
        <param key="rollDragCoefficient" value="0.01"/>
        <param key="constantPowerIntake" value="100"/>
        <param key="propulsionEfficiency" value="0.9"/>
        <param key="recuperationEfficiency" value="0.9"/>
        <param key="stoppingThreshold" value="0.1"/>
        <param key="device.battery.maximumChargeRate" value="150000"/>
    </vType>
</routes>

If a vehicle has a battery device (and is not tracking fuel) and no explicit emissionClass is defined, it will be assigned the emission class Energy/unknown. It will not use the default emission class derived from the vehicle class then. This is for backward compatibility and will issue a warning because in general it is preferable to set the emission class explicitly. Most of the parameters above do actually apply to this emission class and not to the battery device itself.

The initial energy content of the battery (by default 0.5*device.battery.capacity) can be set in the vehicle definitions

<routes>
    <vehicle id="0" type="type1" depart="0" color="1,0,0">
        <param key="device.battery.chargeLevel" value="500"/>
    </vehicle>
</routes>

The charging rate of the battery at a charging station is limited to model the effects of battery management controllers (e.g. charge a nearly full battery less than an nearly empty one). There are two ways to define the maximum charge rate: For a constant rate set the attribute device.battery.maximumChargeRate. If instead a maximum charge rate depending on the state of charge is wanted, it can be defined through data points between which the rate will get interpolated. The states of charge have to be given in device.battery.chargeLevelTable and the corresponding charge rates in device.battery.chargeCurveTable. If defined, the maximum charge curve takes precedence over the constant maximum charge rate device.battery.maximumChargeRate. An example definition where the charge rate decreases above 50% state of charge looks like the following:

<routes>
    <vehicle id="0" type="type1" depart="0" color="1,0,0">
        <param key="device.battery.chargeLevelTable" value="0 0.5 1"/>
        <param key="device.battery.chargeCurveTable" value="45000 45000 20000"/>
    </vehicle>
</routes>

Vehicle behavior#

By default, vehicle behavior will not be affected by battery level. Car will keep driving even when their battery capacity is at 0. To avoid this, either TraCI must be used to change speed or route based on the current battery level or the stationfinder device can be configured to monitor the battery capacity.

Charging Stations#

A charging station is a surface defined on a lane in which the vehicles equipped with a battery are charged. The basic structure and parameters of bus stops were used for the implementation of charging stations.

Charging stations are defined in additional using the following format:

<additional>
    <chargingStation chargeDelay="2" chargeInTransit="0" power="200000" efficiency="0.95" endPos="25" id="cS_2to19_0a" lane="2to19_0" startPos="10"/>
</additional>

And are represented in the simulation as shown:

Representation of chargingStation in GUI

Color of chargingStation during charge

Stopping at a Charging Station#

A stop at a charging station may occur due to traffic conditions, stopping at a defined location or stopping at an explicit chargingStation as defined below:

<routes>
    <vehicle id="v0" route="route0" depart="0">
        <stop chargingStation="myChargingStationID" until="50"/>
    </vehicle>
</routes>

Charging Station output#

There are two variants of the charging station output. One lists values for all time steps whereas the other aggregates the data into charging events (thus related to the whole time a vehicle was connected to the charging station).

Full report#

Option --chargingstations-output chargingstations.xml generates a full report of energy charged by charging stations:

<output>
    <battery-output value="battery.xml"/>
    <chargingstations-output value="chargingstations.xml"/>
</output>

File chargingstations.xml has the following structure:

<chargingstations-export>
    <chargingStation id="CS1" totalEnergyCharged="71.25" chargingSteps="27">
        <vehicle id="veh0" type="ElectricVehicle1" totalEnergyChargedIntoVehicle="15.83" chargingBegin="12.00" chargingEnd="17.00">
            <step time="12.00" chargingStatus="chargingStopped" energyCharged="2.64" partialCharge="2.64" power="10000.00" efficiency="0.95" actualBatteryCapacity="12962.97" maximumBatteryCapacity="35000.00"/>
            <step time="13.00" chargingStatus="chargingStopped" energyCharged="2.64" partialCharge="5.28" power="10000.00" efficiency="0.95" actualBatteryCapacity="12965.59" maximumBatteryCapacity="35000.00"/>
            <step time="14.00" chargingStatus="chargingStopped" energyCharged="2.64" partialCharge="7.92" power="10000.00" efficiency="0.95" actualBatteryCapacity="12968.22" maximumBatteryCapacity="35000.00"/>
        </vehicle>
        <vehicle id="veh1" type="ElectricVehicle2" totalEnergyChargedIntoVehicle="5.28" chargingBegin="17.00" chargingEnd="18.00">
            <step time="17.00" chargingStatus="chargingStopped" energyCharged="2.64" partialCharge="18.47" power="10000.00" efficiency="0.95" actualBatteryCapacity="11967.35" maximumBatteryCapacity="35000.00"/>
            <step time="18.00" chargingStatus="chargingStopped" energyCharged="2.64" partialCharge="21.11" power="10000.00" efficiency="0.95" actualBatteryCapacity="12978.72" maximumBatteryCapacity="35000.00"/>
        </vehicle>
        ...
    </chargingStation>
        ...

    ...
</chargingstations-export>

For the entire ChargingStation:

For the current charging vehicle

For every charging timeStep:

Charging events#

Option --chargingstations-output chargingevents.xml --chargingstations-output.aggregated true generates a report of the charging events at any charging station in the network. The generated output file chargingevents.xml has the following structure:

<chargingstations-export>
    <chargingEvent chargingStation="CS1" vehicle="veh0" type="ElectricVehicle1" totalEnergyChargedIntoVehicle="15.83" chargingBegin="12.00" chargingEnd="17.00" actualBatteryCapacity="12968.22" maximumBatteryCapacity="35000.00" minPower="10000.00" maxPower="10000.00" minCharge="2.64" maxCharge="2.64" minEfficiency="0.95" maxEfficiency="0.95" />
    <chargingEvent chargingStation="CS1" vehicle="veh1" type="ElectricVehicle1" totalEnergyChargedIntoVehicle="5.28" chargingBegin="17.00" chargingEnd="18.00" actualBatteryCapacity="12978.72" maximumBatteryCapacity="35000.00" minPower="10000.00" maxPower="10000.00" minCharge="2.64" maxCharge="2.64" minEfficiency="0.95" maxEfficiency="0.95" />
    ...

</chargingstations-export>

If the option --chargingstations-output.aggregated.write-unfinished true is set as well, vehicles still charging at the end of the simulation are included as well (but do not contain the chargingEnd attribute). Attributes with the same name can be looked up in the table above. The remaining attributes are:

battery-output#

There are three output parameters to be set in the SUMO configuration to use the battery device:

<configuration>
    <input>
        <net-file value="netFile.xml"/>
        <route-files value="routeFile.xml"/>
        <additional-files value="additionalFile.xml"/>
    </input>
    <output>
        <battery-output value="Battery.out.xml"/>
        <battery-output.precision value="4"/>
        <device.battery.probability value="1"/>
        <summary-output value="summary_100.xml"/>
    </output>
</configuration>

The battery-output generates a file with this structure:

<battery-export>
    <timestep time="0.00">
        <vehicle id="vehicle01" energyConsumed="0.00" totalEnergyConsumed="0.00" totalEnergyRegenerated="0.00"
            actualBatteryCapacity="17500.00" maximumBatteryCapacity="35000.00"
            chargingStationId="NULL" energyCharged="0.00" energyChargedInTransit="0.00" energyChargedStopped="0.00"
            speed="12.92" acceleration="0.00" x="1428.27" y="25.57" lane="01to02_0"
            posOnLane="0.00" timeStopped="0"/>
        <vehicle id=..... */
    </timestep>
    <timestep time="1.00">
        <vehicle id=.....
    </timestep>
    <timestep time=...
    ...
    </timestep>
</battery-export>

Emission Output#

The Emission model-outputs of SUMO can be used together with the battery device when setting the <vType>-parameter emissionClass="Energy/unknown".

Tracking fuel consumption for non-electrical vehicles#

By setting option --device.battery.track-fuel, equipped vehicles with a conventional drive train (emissionClass other than Energy) will monitor their fuel level based on the fuel consumption of their respective emission class. All capacity values are then interpreted as mg instead of Wh. Also, the chargingStation power is re-interpreted as mg/s when charging fuel.

TraCI#

The internal state of the battery device can be accessed directly using traci.vehicle.getParameter and traci.vehicle.setParameter. Charging stations can be inspected and updated using the respective getter and setter functions inside traci.chargingstation.

Furthermore, the function traci.vehicle.getElectricityConsumption() can be used to access the consumption of the vehicle if the emissionClass="Energy/unknown" is declared.

Query of the current state of charge#

The current state of charge of the battery can be computed as the quotient of the actual battery charge and the maximum battery capacity:

capacity = float(traci.vehicle.getParameter(vehID, "device.battery.capacity"))
currentCharge = float(traci.vehicle.getParameter(vehID, "device.battery.chargeLevel"))
stateOfCharge = currentCharge / capacity

Calculating the remaining Range#

After the vehicle has been driving for a while, the remaining range can be computed based on previous consumption and distance:

mWh = traci.vehicle.getDistance(vehID) / float(traci.vehicle.getParameter(vehID, "device.battery.totalEnergyConsumed"))
remainingRange = float(traci.vehicle.getParameter(vehID, "device.battery.chargeLevel")) * mWh

To compute the remaining range on departure, the value of mWh (meters per Watt-hour) should be calibrated from a prior simulation.

Reducing the power of a charging station#

If too many consumers connect to the eletrical grid, it may not be able to supply the nominal power of the charging station. A temporary power drop of 20% can be modeled using the following sample code:

prevPower = traci.chargingstation.getPower(chargingStationID) # remember for restoring the full power later
traci.chargingstation.setPower(chargingStationID, prevPower * 0.8)

Model Details#

All information about the implemented device (including details on the vehicle energy consumption and charging model) can be found in the following publication.

Publications#

• Kurczveil, T., López, P.A., Schnieder, E., Implementation of an Energy Model and a Charging Infrastructure in SUMO. In: Behrisch, M., Krajzewicz, D., Weber, M. (eds.) Simulation of Urban Mobility. Lecture Notes in Computer Science, vol. 8594 , pp. 33--43. Springer, Heidelberg (2014)

Example Configurations#

Kia Soul EV 2020#

The values are provided by courtesy of Jim Div based on his own calibration.

<!-- vehicle mass = 1682kg curb wt + average 2 passengers / bags -->
<vType id="soulEV65" minGap="2.50" maxSpeed="29.06" color="white" accel="1.0" decel="1.0" sigma="0.0" emissionClass="Energy/unknown" mass="1830">
    <param key="has.battery.device" value="true"/>
    <param key="airDragCoefficient" value="0.35"/>       <!-- https://www.evspecifications.com/en/model/e94fa0 -->
    <param key="constantPowerIntake" value="100"/>       <!-- observed summer levels -->
    <param key="frontSurfaceArea" value="2.6"/>          <!-- computed (ht-clearance) * width -->
    <param key="rotatingMass" value="40"/>               <!-- guesstimate, inspired by PHEMlight5 PC_BEV -->
    <param key="device.battery.capacity" value="64000"/>
    <param key="maximumPower" value="150000"/>           <!-- website as above -->
    <param key="propulsionEfficiency" value=".98"/>      <!-- guesstimate value providing closest match to observed -->
    <param key="radialDragCoefficient" value="0.1"/>     <!-- as above -->
    <param key="recuperationEfficiency" value=".96"/>    <!-- as above -->
    <param key="rollDragCoefficient" value="0.01"/>      <!-- as above -->
    <param key="stoppingThreshold" value="0.1"/>         <!-- as above -->
</vType>

Observations: - Simulation efficiencies of 6.3 - 6.7 km driven per kWh consumed agree with measured efficiencies of 6.4 - 6.8 (stddev 0.4 - 0.8) in short and medium range simulations with realistic traffic - abstract scenarios without junctions and other cars overestimate the efficiency by a large factor (~ twice as many km/kWh)