Dear Ray,
thank you for your reply. I have updated shunt load with voltage coefficient. However, I am still finding some issues with final results. I have replicated my system in two independent software platforms. One is the Matpower and the second is the RTDS (real-time digital power system simulator). When I set my system loads to be 100% constant power loads, the power flow results from both simulators are nearly identical. Newton's method power flow converges in 4 iterations and the voltages differences between RTDS and Matpower is only approx. 0.1% and it is consistent at all buses. I consider it as an extremely good result and I am very happy with it. However, when I add some portion of a constant impedance loads (ratio 70/30 or 80/20), the voltage error increase to 0.6%-0.8% (and shunt power accordingly). It is 7x bigger error caused only by changing the proportions of constant impedance to constant power loads. At the each investigated case, Matpower calculated higher value of voltages. Do you have any idea what might the source of the error (bearing in mind that for fully constant power loads the error is almost non-existent)? How can I possibly eliminate the issue? BR, Jaroslaw ________________________________ From: [email protected] <[email protected]> on behalf of Ray Zimmerman <[email protected]> Sent: 06 April 2016 00:27 To: MATPOWER discussion forum Subject: Re: Matpower 5.1 and constant impedance loads Hi Jarek, I don’t think any code changes should be necessary for solving the problem you want to solve. That is, if you put the constant impedance loads in the GS and BS columns of the bus matrix, the calculations should be correct, including all of the resulting voltages and flows. The only issue is that the power delivered to these elements is not reported anywhere as “load”, so you’ll have to do that manually after the fact, but that is a trivial calculation once you have the bus voltages. For example, if the nominal (i.e. at 1 p.u. voltage) active and reactive constant impedance loads are given in the nb x 1 vectors Pdz0 and Qdz0, then you can calculate the constant impedance loads at the solution as follows ... define_constants; mpc = loadcase(...); mpc.bus(:, GS) = mpc.bus(:, GS) + Pdz0; mpc.bus(:, BS) = mpc.bus(:, BS) - Qdz0; r = runpf(mpc); Pdz = Pdz0 ./ r.bus(:, VM) .^ 2; Qdz = Qdz0 ./ r.bus(:, VM) .^ 2; The constant power loads can be found using total_load() … [Pds, Qds] = total_load(r.bus, r.gen, 'bus'); So, no need for any of your steps beyond a) under #2, except I suppose f) could be done by simply doing … r.bus(:, PD) = r.bus(:, PD) + Pdz; r.bus(:, QD) = r.bus(:, QD) + Qdz; Ray On Apr 5, 2016, at 1:43 AM, Jaroslaw Krata <[email protected]<mailto:[email protected]>> wrote: Dear Matpower authors, I wish to use the Matpower software in my PhD research. I need to calculate power flow for the distribution system that consists of many constant impedance and constant power loads. After I read in the manual that " (...) the constant impedance portions can be modeled as a shunt element (...)", I thought naively that it's gonna be easy ;-). I went through other messages from this mailing list and I understand that I need to changes the code of the Power Flow calculation. In particular, I was trying use this advise https://www.mail-archive.com/matpower-l%40cornell.edu/msg03254.html but it seems that since 2014 the code of Matpower has changed and this advice is not easy applicable to the newest version of the Matpower. Thus, I would like to ask you for help of the required code changes in the Matpower 5.1. I wish to achieve two targets: 1. correct power flow calculation assuming const P and const Z loads 2. output branch and power matrices showing updated (real) values of power from const Z loads. I started as follows: a) I assume that whole constant power loads are assigned to the bus matrix - Pd and Qd columns while all constant impedance loads are assigned to Gs and Bs columns b) I modified makeSbus() function by adding lines related to const P and Z line1: [Sbus, Sbus_gen, Sbus_constP, Sbus_constZ] = makeSbus(baseMVA, bus, gen) %adding Sbus_gen, Sbus_consP and Sbus_constZ vectors line36 to 38: Sbus = ( Cg * (gen(on, PG) + 1j * gen(on, QG)) ... %% power injected by generators - (bus(:, PD) + 1j * bus(:, QD)) ... %% plus power injected by const P loads ) / baseMVA; %% converted to p.u. replaced by: Sbus = ( Cg * (gen(on, PG) + 1j * gen(on, QG)) ... %% power injected by generators - (bus(:, PD) + 1j * bus(:, QD)) ... %% plus power injected by const P loads - (bus(:, GS) + 1j * bus(:, BS)) )/ ... %% plus power injected by const Z loads baseMVA; %% converted to p.u. Sbus_gen = (Cg * (gen(on, PG) + 1j * gen(on, QG))) / baseMVA; Sbus_constP = (bus(:, PD) + 1j * bus(:, QD)) / baseMVA; Sbus_constZ = (bus(:, GS) + 1j * bus(:, BS)) / baseMVA; c) I pass all Sbus vectors to newtonpf() function [V, converged, i] = newtonpf(Ybus, Sbus, Sbus_gen, Sbus_constP, Sbus_constZ, V0, ref, pv, pq, mpopt) d) In newtonpf(), I calculate 'mis' variable as line 54 and line 106: mis = V .* conj(Ybus * V) - (Sbus_gen - Sbus_constP - Sbus_constZ.*(Vm.^2)) (replacing the default mis = V .* conj(Ybus * V) - Sbus;) e) How should I change the dSbus_dV() function?? f) How to update bus matrix to show updated value of const Z power? g) Should I change other functions as well (e.g. makeYbus() function)? I will be grateful for your help and suggestions, BR, Jarek
