A Python implementation of a conic programming formulation to the power flow problem on tree networks, as described in the paper

R. A. Jabr, "Radial distribution load flow using conic programming,"
IEEE Transactions on Power Systems, vol. 21, no. 3, pp. 1458-1459, 2006.

It relies on Gurobi, a commercial (but free academic license) optimization solver, which unfortunately requires python 2.

The work I'm doing on the power grid attack problem involves testing the power flow on grids that are far from their normal operating conditions. The usual Newton's method approach can't be trusted (as in MATPOWER) for all cases; I want a certificate of infeasibility, which this method provides.

Once you have gurobi installed, try running

python jabr.py

which should produce

  1   0.853 -27.815
  2   0.985  -5.829
  3   0.873 -20.100
  4   0.853 -27.815
  5   1.000   0.000

Try it on some other examples in the cases directory:

answer = solve('cases/case15_v2.m')
for bus in sorted(answer.keys()):
    v, t = answer[bus]
    print('%3d %7.3f %7.3f' % (bus, v, 180/pi*t))

  1   1.000   0.000
  2   0.971   0.032
  3   0.957   0.049
  4   0.951   0.056
  5   0.950   0.068
  6   0.958   0.189
  7   0.956   0.216
  8   0.957   0.205
  9   0.968   0.072
 10   0.967   0.085
 11   0.950   0.131
 12   0.946   0.182
 13   0.945   0.198
 14   0.949   0.085
 15   0.948   0.087