Hello again, and sorry for taking so long to reply. From: [email protected] Subject: Re: adding user defined costs + changing obj function Date: Tue, 16 Oct 2012 13:10:03 -0400 To: [email protected]
On Oct 15, 2012, at 1:01 PM, Evangelos Galinas <[email protected]> wrote: <snip>If I had understood correctly from a previous example found here in the mailing list I have to create these extra fields in the mpc struct namely mpc.H H mpc.N mpc.Cw in order to formulate eq 5.27. Correct? Correct, and there is also mpc.fparm matrix as well. See the notes in section 6.1. So If I'm using linear costs mpc.Cw needs to be 0, mpc.N has to have the Voltage part (14, 41) zero and the Power part (14,14) non zero (1) only at the elements 2,2 11,11 and 12,12 in order to refer to the power of the specific generator. Am I right? The simplest is probably to use an identity matrix for N and then put the linear coefficients in the appropriate elements of Cw, which should be ordered as first 41 elements are voltage angles, followed by 14 generator real power injections. However, since it is just a cost on Pg, you don't even have to use this mechanism. You can simply modify gencost accordingly. Assuming that I use a standard cost for curtailment, let's say c=0.18 ct/kwh, at which elements of the H matrix should I have this c denoted. The same as the power part of the N? and how I can denote that these cost must come from a subtract from max P of the Pg? Pmax is a constant, so adding a function of Pmax to the objective will not change the solution. The cost of curtailment simply shows up as a negative cost on Pg. the way I have developed my scripts, the Pmax of the wind turbines is not the rated power of the machine but the max infeed for every snapshot. So indeed every hour of the year would be different. I'm calculating the curtailment cost out out of the opf and adding it to the obj. function manually but I believe the results would be different if that cost was in the matpower's objective function. Don't you think? Also, about static general penalty cost to the whole system I have to denote a cost using Cw non zero right? I'm not sure what you have in mind for "penalty costs for unfeasibility". The OPF problem does not allow feasibility violation. Which constraints do you want to replace with penalty functions? Yeah, you are right. I had in mind something like if it did not converge, then assign a penalty to the objective function. But I can do it directly in my scripts, out of the opf. Regards, Vagelis
