Usage¶

Background¶

The general premise of the battery_sizing_model package is a series of objects or “blocks” that are combined to create the battery simulation. The blocks that make up a simulation are the classes Battery, DegradationModel, ECM, SOCIntegrator, and OCVTable.

The blocks are instantiated with whatever attributes or parameters necessary and then are combined into the BatterySimulator class.

Running a Simulation¶

To run a simulation a power series with some constant timestep must be provided and the simulation blocks must be created with their necessary parameters. An example is shown below:

python

# import the package
import battery_sizing_model.blocks as blocks

# declare the timestep
dt = 60

# create an equivalent circuit block
ecm = blocks.ECM(
    num_rc_pairs=3,
    R0=0.00851,
    R=[1.25430794e-03,2.75299761e-03,2.17698493e-03],
    C=[1.65306594e+02,1.57444756e-01,1.42594864e+00],
    timestep=dt
    )

# create an open circuit voltage lookup table block
ocv = blocks.OCVTable()

# create a state of charge integrator block
integrator = blocks.SOCIntegrator()

# create a SOH estimation model block
deg_model = blocks.DegradationModel()

# create a Battery block
battery = blocks.Battery(
    cell_chemistry="NCA",
    pack_capacity=200e3, # Wh
    cell_charge_energy_eff=0.9,
    cell_discharge_energy_eff=0.9
    )

# add the blocks to simulation instance
sim = blocks.BatterySimulator(
    ecm=ecm,
    ocv=ocv,
    integrator=integrator,
    deg_model=deg_model,
    battery=battery,
    soc=0.5
    )

# run the simulation assuming that there is a power
# series named ``power``
sim.run(power=power, timestep=dt)

A similar example is provided on GitHub in the examples section along with example data files for what the expected power series would be. For more information on attributes that can be set in the objects see the API Documentation.

Gotchas¶

There are a few assumptions that were made during the development of this simulation model that need to be discussed.

First, the Battery class converts the power signal to power series to a cell level current in Amps using the nominal cell voltage of the pack. The significance of this is that the currents at high voltages will be higher than real and the currents at low voltages will be lower than real. It’s a reasonable assumption to make; however, it needs to be understood when analyzing results like over or under voltage alarms.

Second, the degradation model in the DegradationModel class is not strictly monotonically decreasing. If the input power series has long durations of differing mean DOD or mean SOC then the model may see a significant recovery in SOH that likely wouldn’t happen in real life. It is up to the user to recognize these situations and think critically about them. Improving this aspect of the model is something to be targeted in future work.