(First of all, this is not a valid MATPOWER case file … the gen matrix needs to have 21 columns).
I’m pretty sure these are simply numerical issues. Your network has a huge range for branch reactances (from 1e-4 up to 7.2415), and I’m sure the zero cost generation doesn’t help. It turns out that some solvers do converge (IPOPT, MINOPF, TRALM) using the default starting point, and others (MIPS and KNITRO) converge if you use the power flow solution to initialize the OPF. Ray E.g. >> load mpc >> mpopt = mpoption('verbose', 2, 'opf.ac.solver', 'MIPS', >> 'out.suppress_detail', 1, 'opf.init_from_mpc', 1); >> r = runopf(mpc, mpopt); >> MATPOWER Version 5.0, 17-Dec-2014 -- AC Optimal Power Flow MATLAB Interior Point Solver -- MIPS, Version 1.1, 17-Dec-2014 it objective step size feascond gradcond compcond costcond ---- ------------ --------- ------------ ------------ ------------ ------------ 0 2.71 0.00320952 0.005 394.161 0 1 13.561367 0.45467 0.0131881 49.7247 129.098 0.00108484 2 2.5810166 0.2459 0.000347117 5.81263 30.6224 0.00109655 3 3.0946997 0.0073205 1.34869e-07 0.5504 3.18123 5.13551e-05 4 3.1139811 0.00033848 8.72888e-10 0.0214386 0.318187 1.92754e-06 5 3.1489261 0.00061552 2.90301e-09 0.018371 0.0318179 3.49342e-06 6 3.0091397 0.0024621 4.645e-08 0.0117079 0.00598559 1.39742e-05 7 2.710015 0.0068676 2.23731e-07 0.0199562 0.000603092 2.99035e-05 8 2.7116241 2.7338e-05 5.10117e-12 8.49417e-06 6.03182e-05 1.60873e-07 9 2.7101543 2.5902e-05 5.14787e-12 8.06346e-07 6.03182e-06 1.46939e-07 10 2.7100156 2.4442e-06 1.27526e-13 7.54021e-09 6.03182e-07 1.3865e-08 Converged! Converged in 0.06 seconds Objective Function Value = 2.71 $/hr ================================================================================ | System Summary | ================================================================================ How many? How much? P (MW) Q (MVAr) --------------------- ------------------- ------------- ----------------- Buses 192 Total Gen Capacity 25.0 -25.0 to 25.0 Generators 4 On-line Capacity 25.0 -25.0 to 25.0 Committed Gens 4 Generation (actual) 3.5 1.7 Loads 73 Load 3.4 1.6 Fixed 72 Fixed 3.3 1.6 Dispatchable 1 Dispatchable 0.1 of 0.1 -0.0 Shunts 0 Shunt (inj) -0.0 0.0 Branches 191 Losses (I^2 * Z) 0.05 0.16 Transformers 191 Branch Charging (inj) - 0.0 Inter-ties 0 Total Inter-tie Flow 0.0 0.0 Areas 1 Minimum Maximum ------------------------- -------------------------------- Voltage Magnitude 0.957 p.u. @ bus 171 1.000 p.u. @ bus 1 Voltage Angle -2.53 deg @ bus 158 0.00 deg @ bus 1 P Losses (I^2*R) - 0.01 MW @ line 1-2 Q Losses (I^2*X) - 0.09 MVAr @ line 1-2 Lambda P 0.00 $/MWh @ bus 1 0.00 $/MWh @ bus 171 Lambda Q 0.00 $/MWh @ bus 1 0.00 $/MWh @ bus 171 > On Jan 10, 2015, at 5:02 AM, Simone Cochi <simoneco...@gmail.com> wrote: > > Dear all, > I can’t get the convergence for the OPF on a radial distribution network > described by the mpc file that I attached. The power flow with this case file > converge and all the voltage busses are between without violating any > costraints. The OPF doesn’t converge with the standard voltage constraints > but if I relax them, the opf converge with voltage very high voltage levels. > If I relax all the busses voltage constraints and I set Vmin=Vmax for the bus > 1 (primary substation) the of converge and the voltage profiles are as > expected. Can someone explain this behavior? > > Simone > <mpc.mat>