Please see my comments below …
> On Dec 7, 2016, at 6:14 AM, Robin Weckx <[email protected]> > wrote: > > Dear Ray > > Thank you for your help! > I was able to modify the code to work for my kundur file but I still have > some questions to make sure i fully understand how the program works. > are my comments in the code correct: > define_constants; > mpopt = mpoption('opf.ac.solver', 'MIPS', 'verbose', 2);%choose which solver > to use+which info > to print > mpc = loadcase('case30');%load file > nb = size(mpc.bus, 1);%calculate number of busses in system > ng = size(mpc.gen, 1); > mpc.gencost = ones(ng, 1) * [2 0 0 2 1 0];%modify the gencost matrix in the > case30file > r1 = runopf(mpc, mpopt);%runopf(wil give same result as pf because all cost > are equal) The result is not the same as the power flow, but rather is a loss minimization OPF. That is, generators will be redispatched and generator voltages will be optimized to minimize losses. Since loads are fixed and all generator costs are equal, the objective is to minimize the total generation, which is equivalent to minimizing costs. > mpc.gencost(:, 5) = 0;%modify gen cost so all cost of generation is 0 > mpc.N = [sparse(nb, nb) speye(nb, nb) sparse(nb, 2*ng)]; > Because there are no new elements in the matrix x I would suspect N to be an > eye matrix. I don't really understand what this formulation of N does and how > the 3 matrix defined relate to each other.(does this isolate the Voltage from > the x vector?) Yes, N is used to select the voltage magnitudes from the x vector. > mpc.fparm = ones(nb, 1) * [2 1 0 1];%define objective for each bus > mpc.Cw = 1000 * ones(nb, 1);% we multipy by 1000 to make the objective strong? Yes, it is essentially to scale the cost function to a level that the nodal prices are visible in the printed output. This is an arbitrary scaling and is not strictly necessary. > r2 = runopf(mpc, mpopt);%runopf > > the cost funciton is (V-1)^2 so > > The formula in matpower for the cost objective is: > fu(x; z) =1/2 w^THw + C^Tw > by default H is zero so the first part is gone. > C=Cw > Fparm= > d=2 power > r=1 shift > k=0 no deadzone > m=1 scale factor > > w=(V-1)^2 because of fparm. Correct. > this is how we get the costfunction > > I hope you find the time to look at my questions. > Kind regards > Robin > > > Van: [email protected] > <mailto:[email protected]> > <[email protected] > <mailto:[email protected]>> namens Ray Zimmerman > <[email protected] <mailto:[email protected]>> > Verzonden: maandag 28 november 2016 17:19 > Aan: MATPOWER discussion forum > Onderwerp: Re: change objective OPF > > Here is an example of how to do it for case30. The first run is a > loss-minimization solution, the second minimizes the voltage deviations from > 1 p.u. using the objective you specified. > > define_constants; > mpopt = mpoption('opf.ac.solver', 'MIPS', 'verbose', 2); > mpc = loadcase('case30'); > nb = size(mpc.bus, 1); > ng = size(mpc.gen, 1); > mpc.gencost = ones(ng, 1) * [2 0 0 2 1 0]; > r1 = runopf(mpc, mpopt); > > mpc.gencost(:, 5) = 0; > mpc.N = [sparse(nb, nb) speye(nb, nb) sparse(nb, 2*ng)]; > mpc.fparm = ones(nb, 1) * [2 1 0 1]; > mpc.Cw = 1000 * ones(nb, 1); > r2 = runopf(mpc, mpopt); > > norm(r1.bus(:, VM) - 1) > norm(r2.bus(:, VM) - 1) > > Notice the reduction in voltage deviations from the first run. > > Ray > > > > >> On Nov 24, 2016, at 10:46 AM, Robin Weckx <[email protected] >> <mailto:[email protected]>> wrote: >> >> Dear, >> >> I have implemented the two area kundur topology in the matpower software. I >> can successfully run a PF and OPF. My next goal is the change the objective >> of the OPF so that the voltages on the busses are as close to unity as >> possible independently from the cost of the generators. For this i modified >> my gencost matrix so that all the generators have the same cost. the next >> step is to add my objective while the constrains stay the same. Because i >> don't need to add any values because Vm is already in the x matrix this >> process should be quite straightforward but i keep failing. >> My objective function is fmin=(Vm-1)^2 for all the voltages of the busses. >> How do i implement this? >> i looked at the "add_costs" function but don't really know where i should >> use this function. >> i tried the direct implementation method of cheaper 7. should i just but the >> values of the H,N,Cw,... in de Kundur file that I use to perform >> runopf('Kundur')? >> or could i just modify the gencost matrix to make a "buscost" matrix where >> all the polynomial cost functions are as defined above? >> >> Could you please give me some more directions on how to solve this problem? >> >> thanks in advance >> kind regards >> Weckx Robin > >
