First of all, the OPF algorithms in MATPOWER are only expected to work reliably 
on convex cost functions (which you do not have). It turns out that using 
OPF_ALG_DC = 100 (for BPMPD) does give nearly the same answer for the DC OPF.

-- 
Ray Zimmerman
Senior Research Associate
211 Warren Hall, Cornell University, Ithaca, NY 14853
phone: (607) 255-9645



On May 11, 2011, at 7:42 PM, Santiago Chamba wrote:

> Hello all,
>  
> In this moment, I am understanding the mechanism about of the minimization of 
> cost with curves cost and load benefit. For this reason, I have posed a 
> problem with generators and dispatchable loads in a two nodes system with the 
> following equations:
>  
> f1 = 0.05*p(1)^2+100*p(1;  in node 1 (generator 1); limits   0<p(1)<100
> f2 = -0.05*p(2)^2+100*p(2);  in node 1(dispatchable load 1);  limits   
> -100<p(1)<0
> f3 = -0.06*p(3)^2+120*p(3); in node 2 (generator 2); limits   0<p(1)<100
> f4 = 0.06*p(4)^2+120*p(4) ; in node 2 (dispatchable load 2); limits   
> -100<p(1)<0
>  
> With this conditions, I run DC OPF and AC OPF and found the following results 
> of generation and distpachable load:
>  
>                                                 DC OPF                AC OPF
> generator 1                         89.0265                90.8313
> dispatchable load 1        -89.0265              -0.0000
> generator 2                       92.4779                 0.0000
> dispatchable load 2        92.4779              -90.8238
>                                                 
> To compare these results, I used the Matlab’s Toolbox, specifically fmimcon, 
> to optimize these curves and the results are:
>                               
> x =
>    90.9091    0.0000   -0.0000  -90.9091
> fval =
>  -909.0909
>  
> Why are they so different the DC and AC results?.
>  
> PD: The problem is designed to optimized the curves f1 and f4, because the 
> curves f2 and f3 are the reflecting of the curves f1 and f4 (this case is 
> only to understand the mechanism). The problem is attached.
>  
> Thanks so much for the help.
>  
> Santiago Ch.
>  
>  
> <mer.m>

Reply via email to