It's hard to say without more details. Some of the OPF solvers (MIPS, PDIPM) use their own starting point, but MINOPF in particular should use the solved power flow as a starting point and if you have VERBOSE turned on you should be able to see that the feasibility criterion is very small. If not, you may have missed some constraints. The basic ones are branch flow limits (MVA limits by default), generator P and Q upper and lower bounds, voltage magnitude upper and lower bounds. There may also be generator capability curves and branch angle difference limits as well as additional user-defined constraints if they are specified in the data.
-- Ray Zimmerman Senior Research Associate 419A Warren Hall, Cornell University, Ithaca, NY 14853 phone: (607) 255-9645 On Sep 14, 2011, at 11:42 AM, David Bekaert wrote: > Dear, > > I have a (rather large) case that converges when a power flow is carried out. > However, an optimal power flow does not converge. > > I checked the solution of the power flow for all limits, and gave this as a > starting point to the optimal power flow. > I was expecting that the optimal power flow would give at least a feasible > solution (as the PF-equations can be solved, and limits are respected by the > starting point). > Even if the algorithm is not allowed to iterate, and thus stays on its > starting point, it does not converge. > > I tried several solvers. PDIPM gives as error message "IPM: numerically > fails: unable to solve Ax=b" > > Can somebody hints into what direction I have to look for solving this issue? > What is most likely? Is the problem related to my case or do I miss some > OPF-constraints? > > Thanks for helping, > David > >
