Skip to content

Running GSEE

This section covers running the PV model directly:

  • PV models: the available panel technologies, the inverter model, and their parameters (shared by all entry points).
  • Multi-site simulations: running many sites or whole grids at once on the vectorized core.
  • Legacy options: flags that reproduce the behaviour of older GSEE versions.

For running on coarse-resolution climate data, see the climate data interface.

Single-site simulations

The classic entry function is gsee.pv.run_model(), which simulates one PV system from hourly or finer time series data:

import gsee.pv

result = gsee.pv.run_model(
    data,
    coords=(22.78, 5.51),  # lat and lon
    tilt=30,               # degrees
    azim=180,              # degrees, 180 = towards equator
    tracking=0,            # 0 = fixed, 1 = 1-axis, 2 = 2-axis
    capacity=1000,         # W
)

data must be a pandas DataFrame indexed with a timezone-aware UTC DatetimeIndex at a uniform resolution, with columns:

  • global_horizontal: mean global horizontal irradiance per timestep, in W/m2.
  • diffuse_fraction: fraction of the irradiance that is diffuse (if you only have global horizontal irradiance, estimate it with the BRL model).
  • temperature (optional): ambient air temperature in degrees Celsius. 20 °C is assumed if absent.

Any uniform resolution, hourly or finer, is supported. Sub-hourly data (e.g. 30- or 15-minute) must supply the diffuse_fraction column directly, since the BRL model used to estimate it works on hourly data only.

The result is a pandas Series of mean AC power output per timestep, in W (power is in W, energy in Wh throughout GSEE). Energy per timestep is power times the timestep length in hours, so power in W and energy in Wh are numerically identical for hourly data, but not for finer resolutions. With include_raw_data=True, a DataFrame is returned that also contains the in-plane direct and diffuse irradiance, module temperature and relative panel efficiency.

Model configuration beyond the required parameters, including panel technology, inverter sizing, and system losses, is described on the PV models page.

Tip

When calling run_model repeatedly over the same time index and site (e.g. to compare configurations), compute the solar angles once with gsee.api.sun_angles_frame() and pass them via angles=. See speeding up single-site runs for details.