pandapower

Overview

pandapower is an open-source Python library for automated analysis and optimization of power systems. It stores network data as pandas DataFrames, provides Newton-Raphson and other power flow solvers (including C++ backends via lightsim2grid and PowerGridModel), and supports advanced studies including OPF, short circuit (IEC 60909), three-phase unbalanced flow, and state estimation.

Version: v3.4.0 Language: Python License: BSD 3-Clause Authors: University of Kassel (e2n) and Fraunhofer IEE

Quick Start

import pandapower as pp

# Create network
net = pp.create_empty_network(f_hz=50.)

# Add buses
b_hv = pp.create_bus(net, vn_kv=110., name="HV Bus")
b_mv = pp.create_bus(net, vn_kv=20., name="MV Bus")

# Add external grid (slack/reference)
pp.create_ext_grid(net, bus=b_hv, vm_pu=1.02)

# Add transformer (uses built-in standard type library)
pp.create_transformer(net, hv_bus=b_hv, lv_bus=b_mv, std_type="25 MVA 110/20 kV")

# Add load
pp.create_load(net, bus=b_mv, p_mw=10.0, q_mvar=2.0)

# Run AC power flow
pp.runpp(net)

# Inspect results (stored in net.res_* DataFrames)
print(net.res_bus[["vm_pu", "va_degree"]])
print(net.res_trafo[["loading_percent"]])
print(f"Converged: {net.converged}")

Core Concepts

• Network as DataFrames: All elements stored as pandas DataFrames (net.bus, net.line, net.load, etc.); results in net.res_* tables after power flow.
• Consumer sign convention: Positive p_mw means consumption for loads; positive p_mw means generation for generators/sgens.
• Standard types: Built-in library of line and transformer types; custom types supported via create_std_type().
• Per-unit system: Voltages in per unit (p.u.) with sn_mva as base; power in MW/Mvar; impedances in ohm/km.
• In-place results: runpp() and other solvers write results to net.res_* tables; check net.converged after each run.
• Modular subpackages: pandapower.topology, pandapower.plotting, pandapower.shortcircuit, pandapower.estimation, pandapower.timeseries, pandapower.control are separate namespaces.

API Reference

Common Workflows

• Build network from scratch: See api-network.md, then workflows.md
• AC power flow: pp.runpp(net) — See api-powerflow.md
• Optimal power flow: add cost functions + limits, then pp.runopp(net) — See workflows.md
• Short circuit: pp.shortcircuit.calc_sc(net, fault="3ph") — See api-powerflow.md
• Plot network: pp.plotting.simple_plot(net) — See api-plotting.md
• Use benchmark network: pp.networks.case14(), pp.networks.mv_oberrhein() — See api-network.md
• Time series simulation: See api-toolbox.md and workflows.md
• N-1 contingency analysis: See workflows.md

Key Considerations

• Convergence: Always check net.converged after running power flow. Non-convergence often indicates voltage angle issues — try init="dc" or algorithm="iwamoto_nr".
• Geodata format: Networks created before v2.7 use legacy bus_geodata/line_geodata DataFrames; run pp.plotting.geo.convert_geodata_to_geojson(net) to upgrade.
• OPF requires cost functions: runopp() will fail without at least one cost function (create_poly_cost or create_pwl_cost); all controlled elements need min_p_mw/max_p_mw limits.
• Solver backends: Install lightsim2grid or power-grid-model for 10-100x speedup on large networks; pandapower uses them automatically when available (pip install pandapower[pgm]).
• Three-phase flow: runpp_3ph() is imported from pandapower.pf.runpp_3ph, not the top-level namespace.
• Short circuit: calc_sc() lives in pandapower.shortcircuit; single-phase faults require transformer zero-sequence parameters (vk0_percent, vkr0_percent).