Mitchell, See below. Shri
From: <bounce-123560997-83373...@list.cornell.edu> on behalf of "Dennis, Mitchell T." <mitchelldennis...@uky.edu> Reply-To: MATPOWER discussion forum <matpowe...@list.cornell.edu> Date: Friday, April 26, 2019 at 8:38 PM To: "MATPOWER-L@cornell.edu" <MATPOWER-L@cornell.edu> Subject: Removing a Bus and Evaluating the New Network Dear All, I have a question regarding convergence. I am attempting see the effect of removing a node on a power network using MATPOWER. To do this, I first evaluate the IEEE case with 300 nodes, and then I set the branches of one of the nodes to zero. All isolated nodes (buses) should have bus type = 4. You can find which nodes are isolated and the islands formed using the function find_islands<http://www.pserc.cornell.edu/matpower/docs/ref/matpower5.0/find_islands.html>. Doing this typically causes the network to no longer converge using the power flow equations, however in some cases it will still converge. It’s highly unlikely that the power flow converges if there are isolated buses in the network. You might want to run the find_islands function to see if the network is connected or split. In an attempt to make the diverging networks converge again, I systemically reduce the load by setting the bus Pd and Qd values 5% lower and then run the convergence test again. Is there a better way to do this? Am I actually decreasing the load? You can use MATPOWER’s continuation power flow (see the chapter in the manual) instead. And will decreasing the load make the equations more likely to converge? Also, how can I tell what is making the equations not converge? Is divergence only caused by a load imbalance where the load is too high? See this FAQ<http://www.pserc.cornell.edu/matpower/#pfconvergence> on reasons a power flow may diverge and suggestions to try to make it converge. Thanks in advance Best Regards, Mitchell T Dennis