Hi Malcolm, This should not be happening. Can you confirm that you get the same behavior with MATPOWER 5? There was a bug in the MIPS solver prior to version 5 where a near-singular matrix could produce an extremely large Newton step, resulting in incorrectly satisfying the relative feasibility criterion for successful termination. So it was technically possible to get false positive success values. I think this was quite rare, but someone else did supply me with such a case.
Just to clarify, MATPOWER does not include a mechanism to automatically handle shortfalls. The technique mentioned in Section F.1 of the User’s Manual <http://www.pserc.cornell.edu/matpower/manual.pdf> is simply a description of how we manually set up our problem to deal with the shortfall issue in our market tests. So, when simply using off2case() and the DC OPF, you should not be seeing successful convergence for cases with insufficient generation to meet the fixed demand. If you continue to see successful convergence with MATPOWER 5, I suggest you look carefully at the solution to see if you can locate a specific infeasibility. If you are still convinced that there is a problem with the solution, please send me the case off-list so I can reproduce the problem. Ray > On Dec 21, 2014, at 7:56 AM, Malcolm Barnacle <[email protected]> > wrote: > > Hey all, > > I'm using Matpower 4.0 and a smart market function to carry out a cost based > analysis of network constraints, given various bids and offer prices. > > Specifically, i'm using the off2case function and then running a DCOPF using > MIPS on a heavily constrained network throughout the year. > > My question relates to how these two functions (off2case and the DCOPF) > appear to handle supply shortfall. During periods of high demand, the total > demand is around 57GW and the output off all contributing generators (from > the DCOPF) is around 47GW. A small dispatchable load, which is not signifcant > enough to meet the supply shortfall, is included in the smart market at a > very high price as I want to model constrained-on and constrained-off > generation only. The DCOPF converges and an output is generated despite there > being a sizeable shortfall. Reffering to the manual, it appears that a method > exists to handle supply shortfall, which involves emergency imports; where an > import is modelled as a fixed injection together with an equally sized > dispatchable load. > > Does this method exist in these two functions (i.e. off2case and the DCOPF)? > and if so, where in the code is this performed? Also, where are these imports > placed on the network (the reference bus amongst others?) and how is the > descision made to ensure that the constraints of the DCOPF are met? > > I've looked into the code of both functions and I can't seem to find this > procedure, and yet the DCOPF converges, so this process (or an alternative) > must exist somewhere. > > Kind Regards > Malcolm > > P.S. many thanks for MATPOWER, it's a very useful package and one which I > have used alot throughout my research. > >
