I am not sure there is a clear obvious way to tell which constraints pose the most problem for the optimization. Perhaps doing some sensitivity analysis might help, but with different operating conditions you will get different sensitivities. There are a few papers published on convergence behavior of interior point algorithms, I haven't read them yet, that may be of help.
It is an interesting question. I hope some of the folks working on OPF and related problems can shed some light. Shri Sent from my iPad On Sep 9, 2015, at 8:25 AM, Arun Shrestha <yours.a...@gmail.com<mailto:yours.a...@gmail.com>> wrote: Hello MATPOWER Community, I am trying to understand how various constraint types can affect OPF convergence. Are some constraints easy to satisfy than others? Is there a way to figure out the effect of constraints type (in terms of total number of loadflow iteration or convergence time) on OPF convergence? For Example: Let's assume a small power system with four generating stations. The solution of an unconstrained OPF is shown below: Gen Active Power: P1gen_0, P2gen_0, P3gen_0, P4gen_0 Gen Reactive Power: Q1gen_0, Q2gen_0, Q3gen_0, Q4gen_0 Gen voltage Magnitude: V1gen_0, V2gen_0, V3gen_0, V4gen_0 Gen Voltage Angle: TH1gen_0, TH2gen_0, TH3gen_0, TH4gen_0 Next I add a user defined constraints as shown below, one at a time. Constraint #1: P1gen + P2gen <= P_const OR Constraint #2: Q1gen + Q2gen <= Q_const OR Constraint #3: V1gen + V2gen <= V_const OR Constraint #4: TH1gen + TH2gen <= TH_const I would like to know which constraint (out of 4) will result in the fastest OPF convergence. Since runopf function uses Newton Method (by default), I think P and Q constraints will converge faster than V and TH constraints (based on how power flow is solved). Is there any mathematical derivation which shows the effect of the constraints on OPF convergence? Or the OPF convergence solely depends upon the power system model/cost function used? Any help on this topic is highly appreciated. Thank you, Arun
