Powerflow

Solver Notes

• Use CBC, not HiGHS, for MIP problems with Pyomo APPSI (HiGHS has dual bug)
• pandapower uses Newton-Raphson by default; switch to algorithm="bfsw" for radial networks
• PyPSA lopf() requires a solver; CBC is the free default
• GridPath is a pip dependency, Apache 2.0 licensed

IEEE Test Case Validation

When using standard test cases, validate against published results:

• IEEE 14-bus: V at bus 1 = 1.060 pu (slack)
• IEEE 30-bus: Total generation ~289.2 MW
• IEEE 118-bus: 186 branches, 54 generators

Always compare your results to published benchmarks before reporting.

NREL API Migration Warning

NREL is migrating APIs. New endpoint base: https://developer.nrel.gov/api/v2/

• Check current API status before making requests
• Legacy v1 endpoints may be deprecated
• Always use API key from environment variable NREL_API_KEY

Modeling Patterns

pandapower

import pandapower as pp
import pandapower.networks as pn

net = pn.case14()
pp.runpp(net)  # Newton-Raphson power flow
print(net.res_bus)  # voltage results
print(net.res_line)  # line loading results

Pyomo + CBC

from pyomo.environ import ConcreteModel, Var, Objective, SolverFactory

model = ConcreteModel()
# ... define variables, constraints, objective
solver = SolverFactory("cbc")
results = solver.solve(model, tee=True)

Rules

• Always check solver termination status before using results
• Use per-unit system consistently (don't mix pu and SI without conversion)
• Document base MVA and base kV for all per-unit calculations
• Validate network topology (connectivity check) before running power flow
• Log solver iterations and convergence metrics
• Use tee=True on solvers during development for visibility
• Store results in structured format (DataFrame or dataclass), not loose variables