In the t/t_get_losses.m file, the loss sensitivities are computed analytically and numerically (just to check the correctness of the analytically computed ones). The analytical computation of the sensitivities is certainly much more computationally efficient. Adding them across all lines does give you sensitivities of total losses, which I believe is what you want. The only part I’m not convinced about is subtracting epsilon from both PD and QD at the same time. This sensitivity of losses to simultaneous and equal changes PD and QD seems an odd sort of sensitivity to me.
Ray > On Jan 29, 2015, at 9:40 AM, nivedita arunachalam > <[email protected]> wrote: > > Dear Ray, > I am working on optimal placement and optimal sizing of solar > generators on standard ieee 30 bus system. I determine the size of generators > using different algorithms and for optimal placement I use two methods weak > bus placement and loss sensitivity factor method. I calculate the Loss > sensitivity factor numerically alone as on the file you sent me. But I > subtract the epsilon values to both PD and QD at the same time. And > calculated the LSF. To order the sensitivity I added all the line losses > togather and ordered them. I got to know from the codings that the matrix I > got as numLSF is from making changes to each bus and gives line losses. This > is want I understood and process. Am I right or not. Please suggest me. >
