Thank you for your help. Now I am aware that commenting line 201 was wrong. Your suggestions and the link will help a great deal.
Best regards, Mirko Todorovski From: [email protected] [mailto:[email protected]] On Behalf Of Shri Sent: Thursday, May 02, 2013 3:40 PM To: MATPOWER discussion forum Cc: [email protected] Subject: Re: Flat start On May 1, 2013, at 5:21 PM, Mirko Todorovski wrote: Dear Shri, Thank you for your answer, but I think that adding mpc.gen(:,VG) = 1.0; changes the generator set voltages so different case is obtained, which happens to be solvable by Newton’s method. I presumed that you wanted to run these cases with the set point for generator voltages = 1.0 since this is what your code in (1) also did. I would like to keep generator set voltages as they are in case3012wp and solve it with a flat start. I have noticed that this is a problem only in case3012wp and case2737sop. In all other MATPOWER cases the procedure (2) works fine (with line 199 commented and lines 200 and 201 uncommented in runpf). This is a classical case of the Newton's method not converging because the initial guess is far away from the solution. You could try i) Increase the number of Newton iterations. ii) Reduce the loading/generation till you get a converged solution and then run a continuation power flow from the reduced loading/generation to the loading/generation given in the case file. iii) There are techniques of altering the Newton step length as described here A COMPARISON OF THE OPTIMAL MULTIPLIER IN POLAR AND ... <http://www.google.com/url?sa=t&rct=j&q=tate%20+%20power%20flow%20+%20thesis&source=web&cd=9&ved=0CF8QFjAI&url=http://www.ele.toronto.edu/~zeb/publications/thesis.pdf&ei=fGuCUdysGsugyQGqgoHgBw&usg=AFQjCNGzsINKq_fk_SrVg8lVzqoCP8f3ng&bvm=bv.45921128,d.aWc> that could help. You can do (i) by using mpoption. For (ii) and (iii), you'll have to write your own additional code. Shri Best regards, Mirko Todorovski From: [email protected] [mailto:[email protected]]On Behalf Of Shri Sent: Wednesday, May 01, 2013 11:19 PM To: MATPOWER discussion forum Cc: [email protected] Subject: Re: Flat start On May 1, 2013, at 3:31 PM, Mirko Todorovski wrote: Dear all, I would like to test Newton’s method performance under flat start conditions using test systems from MATPOWER. The experiments were done using case3012wp in the following way: 1. I removed the comment on line 199 in runpf and put comments on lines 200 and 201, therefore the initial state is calculated as V0 = ones(size(bus, 1), 1); i.e. the flat start is applied. When I execute the command runpf('case3012wp',mpoption,'r1.txt') for minimum and maximum voltage magnitude and angle I get the following: Minimum Maximum ------------------------- -------------------------------- Voltage Magnitude 0.880 p.u. @ bus 2445 1.044 p.u. @ bus 926 Voltage Angle -53.54 deg @ bus 2733 1.38 deg @ bus 310 which is a correct solution. 2. I put back the comment on line 199 and remove comments on lines 200 and 201 in runpf and execute the following commands: define_constants; mpc = loadcase('case3012wp'); mpc.bus(:,VM) = 1; mpc.bus(:,VA) = 0; runpf(mpc,mpoption,'r2.txt'); I got the message that Newton’s power flow did not converge. Note that the initial guess for generator bus voltages is calculated based on the voltage set points from mpc.gen ( line 201): V0(gbus) = gen(on, VG) ./ abs(V0(gbus)).* V0(gbus); ^^^^^^^^^ Thus for the code you have in (2), V = 1.0 for the pq buses but it is ~= 1.0 for ref and pv buses. The following code will produce the intended flat start initial guess. define_constants; mpc = loadcase('case3012wp'); mpc.bus(:,VM) = 1; mpc.bus(:,VA) = 0; mpc.gen(:,VG) = 1.0; runpf(mpc,mpoption,'r2.txt'); Shri Shouldn’t these two approaches give the same results? Best regards, Mirko Todorovski
