There are many factors that affect the computation time required for an OPF. The size of the problem, that is the number of variables and constraints (determined by number of buses, branches, generators, dispatchable loads, piece-wise linear cost segments, etc.) is only one of the many factors. Computation time is also affected by how densely connected the network is, how many constraints are binding, how good the initial starting point is, characteristics of the hardware being used (cache sizes, number of cores, etc), cost function parameters, and on and on. So there simply is no simple computation time vs. # of buses relationship.
And no, there is no pre-compilation of any sort for the included test cases. User provided case files are handled just like the included cases. -- Ray Zimmerman Senior Research Associate 211 Warren Hall, Cornell University, Ithaca, NY 14853 phone: (607) 255-9645 On Jul 6, 2010, at 11:03 AM, Waqquas Bukhsh wrote: I am interested in knowing the time vs bus relation in the OPF formulation of MATPOWER. I guess that time should scale at least linearly (if not quadratically) with the increase in number of buses. Indeed only the number of buses is not responsible for the time; number of branches, number of generaors and loads also add to the constraints. But we can very well imagine by increasing the number of busses; branches and generators/loads also increase. Following is some stats from the given test cases and time taken by MATPOWER (mips) to converge. 30 Bus: converged in 1.68 seconds 118 Bus: converged in 2.02 seconds 2737 Bus: converged in 29 seconds 2746 Bus: converged in 31.98 seconds The time is not increasing linearly. Does MATPOWER uses some sought of pre compilation for its already given test cases? That means user given test case may take a longer time to converge. If this is so then it may explain my question. Kind regards, Waqquas
