Ray's suggestion is an excellent idea. I'd like to add the following. Suppose you start with "caseA", which solves ok. Now you make only one change to obtain "caseB": you switch one particular PQ bus to PV type, by carefully specifying its Vg as the voltage value you obtained in the solution of "caseA". You preserve the same value of P (careful with signs here), and the value of Q is immaterial now. When you solve now caseB, you should obtain exactly the same solution, within numeric tolerances.
So, if you used a value of P for this new "generator" equal to -P of the former load, AND you kept Vg equal to the bus V in the original solution, you should obtain the same solution. Of course, enforcing MVAR limits (on this new generator) may alter things completely. Note that the reverse process is not true, in general: you may start from a solved case, then change a generator from PV type to PQ type (this time using the value of Q obtained from the solution when the bus was PV type) and you may end up with an infeasible loadflow. This is not entirely evident. -- Jose L. Marin Grupo AIA -- Jose L. Marin Grupo AIA On Thu, Apr 28, 2016 at 4:04 PM, Ray Zimmerman <r...@cornell.edu> wrote: > It sounds like the voltage at that bus may be very sensitive to the > reactive power injection. One thing you might try to get some idea of this > is to change that bus back to PQ with the reactive at the lower limit, then > try running a few cases with slightly perturbed values of the reactive > power at that generator and see how the voltage at the bus changes. > > Ray > > > On Apr 28, 2016, at 8:40 AM, Chris Prokop <christophprok...@gmail.com> > wrote: > > Hi, > > > I'm using Matpower (v5.0b1, but the same holds for v5.1) for a 220 kV/110 > kV/20 kV-grid quite a while. The grid has 1 reference bus (220 kV), 2 > PV-buses (220 kV) and >100 PQ-buses (110 kV & 20 kV). So far calculating > the grid using runpf with Newton has never been a problem. > > Now I've tried to change a 20 kV PQ-bus to a PV-bus with the Q-limits +0.7 > for Q_max and -0.45 for Q_min (considering mpopt = > mpoption('pf.enforce_q_lims', 1)). If I set Vg of the generator to 1.05 I > get the error: > "All 4 remaining gens exceed their Q limits : INFEASIBLE PROBLEM" > whereas when using 1.00 as Vg there is no problem (then the generator is > at its lower Q-limit, hence converted to PQ). If the problem is infeasible > the result of pfsoln in runpf are Q-values of all 4 PV generators (Slack+3 > PV) that are out of their limits. Why is there such a big difference > between the case V=1.05 and V=1.00? > > As a info: the Slack has a Q-Limit of +-10000, both of the already > existing PV buses +720/-290, according to case_info the total generation is > -300 MW+j10 Mvar, the total load 300 MW-j50 Mvar, but I've tried several > scenarios which are no problem when using the PQ instead of the PV-bus (or > the Vg=1). > > > Does anybody experience a similar problem/has an idea how to fix it? > > Nice regards, > Chris > > >