If there are islands, and you get the singular matrix warning, I suspect the PTDF returned could be incorrect. If there are no islands, then maybe it is due to some strange line parameters (zero reactance lines or something). In that case, I really don't know what to make of the PTDF.
If you are interested I have some code that I'll be including in the next release of MATPOWER that detects the islands in a network (and extracts them into separate cases). I could send it to you off-list. -- Ray Zimmerman Senior Research Associate 419A Warren Hall, Cornell University, Ithaca, NY 14853 phone: (607) 255-9645 On Mar 1, 2012, at 7:46 AM, Trivik wrote: > Even after removing links and generators from the struct I get *this* > warning. Do you think the PTDF is reliable enough once I have removed all the > malfunctioning links and nodes from the network? It is a huge network (N = > 1254, L = 1822). It is difficult to analyse by hand. > > On Wed, Feb 29, 2012 at 4:39 PM, Trivik <[email protected]> wrote: > Even after removing links and generators from the struct I get *this* > warning. Do you think the PTDF is reliable enough once I have removed all the > malfunctioning links and nodes from the network? It is a huge network (N = > 1254, L = 1822). It is difficult to analyse by hand. > > > On Wed, Feb 29, 2012 at 4:06 PM, Trivik <[email protected]> wrote: > I am doing some analysis with load redistribution. So when I remove certain > high load nodes, there are of course islands because I eventually end up > removing half the network and the average degree is about 3.2. The reason > surely could be the islanding effect. > > > On Wed, Feb 29, 2012 at 3:23 PM, Ray Zimmerman <[email protected]> wrote: > As long as you call int2ext() after changing the line status and bus type, so > that they actually get removed from the matrices it should be sufficient. I > did a little test with a network with two unconnected islands and it does > complain of a singular matrix, as I expected. Have you checked if that is > what is causing the problem for you? > > -- > Ray Zimmerman > Senior Research Associate > 419A Warren Hall, Cornell University, Ithaca, NY 14853 > phone: (607) 255-9645 > > > > > On Feb 29, 2012, at 9:01 AM, Trivik wrote: > >> Yes. I am making the line status zero and also making the removed buses as >> isolated. But apart from that is there anything else that I should change in >> the vectors of the mpc struct? I am still getting that warning with a >> smaller PTDF matrix generated this time. >> >> On Tue, Feb 28, 2012 at 5:11 PM, Ray Zimmerman <[email protected]> wrote: >> The context of the error message is important. In this case, it has nothing >> to do with the ill-conditioning of a matrix encountered by MIPS in the AC >> OPF solution, rather the ill-conditioning of a matrix encountered in forming >> the PTDF. I suppose it's possible that the same underlying network >> conditions could cause both. >> >> Are you sure that the remaining system is fully-connected? I suspect that >> the PTDF calculation may encounter singular matrices if the network is not >> connected, though I haven't tested it. >> >> I suggest converting the indexing to a consecutive internal ordering (and >> removing out-of-service branches, buses, etc) by calling ext2int() after >> setting the line statuses to zero and before calling makePTDF(). >> >> -- >> Ray Zimmerman >> Senior Research Associate >> 419A Warren Hall, Cornell University, Ithaca, NY 14853 >> phone: (607) 255-9645 >> >> >> >> >> On Feb 28, 2012, at 7:29 AM, Ravikumar V wrote: >> >>> Hi Trivik Verma, >>> >>> This is Matlab error message appears when the matrix inverse is not >>> possible because of singular matrix. This shows that the power system >>> becomes unhealthy or ill-conditioning because of opening several lines. >>> Kindly refer previous posts on this topic. >>> >>> http://www.mail-archive.com/[email protected]/msg01678.html >>> >>> >>> >>> On Tue, Feb 28, 2012 at 4:05 PM, Trivik <[email protected]> wrote: >>> Could anyone help me?? >>> >>> I am viewing a power grid using MATPOWER and for my simulations I need to >>> remove a node together with its links. By doing so, I am changing >>> mpc.branch(links,:) = 0. After removing the respective links, I am >>> calculating PTDFs again. The first time is fine, but in the event of >>> redistribution of flows, the next time I am removing about 10 percent of >>> the nodes and calculating PTDFs. I get a warning as follows, >>> >>> Warning: Matrix is close to singular or badly scaled. >>> Results may be inaccurate. RCOND = 4.060450e-019. >>> >>> Do I have to remove something else from the mpc struct as well? >>> >>> cheers, >>> Trivik >>> >>> >>> >>> ------------------------------------------------------------- >>> WITH REGARDS, >>> >>> V. Ravikumar Pandi, >>> Masdar Institute, Abu Dhabi, >>> United Arab Emirates - 54224. >>> mobile : +971 5522 78027. >>> ------------------------------------------------------------- >>> HAVE A NICE DAY >>> ------------------------------------------------------------- >>> >> >> > > > >
