Example Scripts Overview

All example scripts are in the scripts/ folder and are fully self-contained. Run any script from the repository root:

conda activate potpourri_env
python scripts/<script_name>.py

Optional output (plots, tables) is written to results/.

Fundamentals

minimal_ac_power_flow.py

End-to-end AC power flow pipeline. Loads a SimBench LV network, builds a Pyomo AC model, solves with IPOPT, and compares voltage magnitudes, angles, and line losses side-by-side with pandapower Newton-Raphson.

pandapower_to_pyomo_inspection.py

Interactive model inspector. Prints how each pandapower table (bus, line, sgen, …) maps to Pyomo sets, parameters, variables, and constraints. Shows variable counts (free vs. fixed), degrees of freedom, and admittance values. No solver required.

dc_opf.py

Linearised DC power flow and DC OPF. Step 1 compares DC bus angles against pandapower's rundcpp. Step 2 minimises external grid import subject to a line-loading constraint — solved with GLPK as a linear programme in milliseconds.

Optimisation studies

acopf_loadcase_analysis.py

Two-step AC OPF on a LV rural network: 1. Reactive-power minimisation at a fixed active dispatch. 2. Voltage-deviation minimisation for the SimBench lW and hL load cases.

generator_capability_curve_demo.py

Shows how reactive capability limits shape AC-OPF dispatch on a 5-bus self-contained feeder (no SimBench needed): - PV inverter: rectangular Q bounds from a power-factor rule (cos φ ≥ pf). - Wind inverter: tighter rectangular bounds. - Battery: circular P² + Q² ≤ S² constraint via the built-in stor_inverter_cap Pyomo constraint.

Compares Scenario A (wide limits, pf ≥ 0.70) against Scenario B (grid-code, PV pf ≥ 0.90 / wind pf ≥ 0.95). Reports Q dispatch, bus voltages, ext-grid reactive import, and the battery operating point on its capability circle.

objective_tradeoff_demo.py

Compares four AC-OPF objectives (voltage deviation, reactive generation, active import, network losses) on the same network and scenario. Shows how the choice of objective changes the optimal reactive dispatch.

constraint_activation_demo.py

Systematically activates each constraint group (voltage bounds, line thermal limits, reactive power limits) and shows how each one restricts the feasible space and forces additional PV curtailment.

time_series_snapshot_opf.py

Solves eight independent AC-OPFs for representative seasonal/diurnal snapshots (winter midday, summer noon, etc.). Illustrates the snapshot-OPF paradigm and seasonal patterns in voltage and loading.

compute_feasible_operation_region.py

Traces the boundary of the (P, Q) feasible operation region at the grid connection point using angle-based sampling. Useful for studying reactive flexibility and grid hosting capacity.

Multi-period planning

multi_period_acopf.py

24-hour AC OPF (96 × 15-min steps) without storage. All time steps are optimised jointly as one large NLP. Shows hourly ext-grid dispatch and voltage band over the full day.

battery_multi_period_opf.py

Adds Battery_multi_period storage to a 24-hour AC OPF. Compares voltage deviation with and without batteries, displays per-battery SOC trajectories, and demonstrates explicit battery parameter sizing.

hosting_capacity_opf.py

Hosting capacity analysis with binary wind placement. Sweeps the minimum turbine size (SWmin) and the wind-vs-loss trade-off parameter (eps). Enforces VDE-AR-N 4105 grid-code Q constraints.

timeseries_acopf.py

Integrates the single-period AC OPF into pandapower's run_timeseries framework. Attaches SimBench 15-min load and sgen profiles via ConstControl / DFData, defines a custom run=run_acopf callback that builds and solves a fresh ACOPF model at every time step, and records results with OutputWriter. Runs one full day (96 steps) on a LV rural network and prints a summary of voltage, line loading, and dispatch.

Benchmarking

pglib_benchmark.py

Benchmarks potpourri DC and AC OPF against the PGLib-OPF reference results (PowerModels.jl + IPOPT). Loads cases via potpourri.benchmarks.load_pglib_case, applies PGLib-compatible flags (thermal_limit='mva', free_slack_vm=True, angle_limits=True), wires the polynomial generator cost from net.poly_cost, and writes results/pglib_benchmark.{csv,md} with a BASELINE.md-style table comparing objective values.

Requires the PGLib-OPF submodule:

git submodule update --init benchmarks/pglib-opf

Validation

validate_ac_model_against_pandapower.py

Full quantitative validation on six SimBench networks (HV, MV, LV). Reports MAE, RMSE, and maximum absolute error for voltage magnitude, angle, active losses, and reactive losses.

compare_solvers.py

Compares several NEOS NLP solver backends (IPOPT, KNITRO, BONMIN, COUENNE, CONOPT, SNOPT) across six SimBench networks. Records solve time and result accuracy for each combination and saves a CSV. Requires a NEOS email: export NEOS_EMAIL=your@email.com.

performance_test_solver.py

Perfplot-based scaling benchmark. Plots solve time vs. network size (number of buses) for each NEOS solver backend. Requires the optional performance-test dependency group (pip install potpourri[performance-test]) and a NEOS email.