Yes, it is possible, as long as your costs can be put in the form of (6.27)-(6-31) in the User’s Manual <http://www.pserc.cornell.edu/matpower/docs/MATPOWER-manual-5.1.pdf>. And you can simply set the appropriate parameters in the gencost matrix to zero, rather than forming user-defined cost terms to cancel out their contribution to the objective function.
It looks like your costs are simple linear functions of up and down generation shift variables. The standard optimization variables contain generation, but not separate up and down shifts in generation from a reference. You can define those by introducing a set of user-defined linear constraints, in the form of (6.25) … Pg – Pup + Pdn = Pref … where Pref is a constant reference dispatch, Pg is part of the existing optimization variable x, and Pup and Pdn are new vectors of variables (blocks in z) defined by additional columns in the A matrix. You also need to set z_min to zero to enforce ... Pup >= 0 Pdn >= 0 Then, I believe, your costs are purely linear functions of these new user-defined optimization variables, so you can use the simplest form of the user-defined cost, where parameters d_i and m_i are all 1, k_i and r_hat are all zero (so w = r), the H matrix is zero and the C vector is all ones. In that case, (6.27) and following simply reduces to … f_u(x, z) = C' * N * [x; z] All you need to do then is create N in two blocks, stacked vertically, one for your f1 and one for your f2. -- Ray Zimmerman Senior Research Associate B30 Warren Hall, Cornell University, Ithaca, NY 14853 USA phone: (607) 255-9645 > On Apr 16, 2015, at 1:04 PM, Electric <[email protected]> wrote: > > Dear Prof Ray > > I studied extended OPF function to adopt my problem to be run with MATPOWER > package as you recommended already, so I could avoid implementing standard > OPF constraints like line loadings and voltage range. As I am running a > multi-objective congestion management problem (equivalently, maybe called > multi-objective OPF), is it possible to include other objective functions > separately in the OPF cost function? > for example : > > f1= I have my own cost function, so I can minus the existed cost function to > cancel it and add my own cost functions that are still directly depended on > the optimization vector. f1=sum(X[i].Bup[i]+Y[i].Bdown[i]), where X and Y are > up and down generation shift of unit i and Bup[i] and Bdown[i] are their > corresponding bid. > > f2= security function. > f2 is also directly depended on the optimization vector. f2=sum(X[i].S[i]), > where x[i] is up generation shift of unit i and S[i] is its corresponding > sensitivity to the stability margin. > > Your help is much appreciated. > > > On Fri, Apr 10, 2015 at 9:11 PM, Electric <[email protected] > <mailto:[email protected]>> wrote: > Dear Prof Ray and Shiri > Thank you so much for your help. I am going to study opf_consfcn() > <http://www.pserc.cornell.edu//matpower/docs/ref/matpower5.1/opf_consfcn.html> > and find out how to implement the constraints I want. > Kind regards. > > On Fri, Apr 10, 2015 at 5:43 PM, Ray Zimmerman <[email protected] > <mailto:[email protected]>> wrote: > Specifically, you might look at how the power flow and line limit constraints > are implemented in opf_consfcn() > <http://www.pserc.cornell.edu//matpower/docs/ref/matpower5.1/opf_consfcn.html>. > However, for a problem that includes the standard OPF constraints I still > think it would be simpler and cleaner to set up your problem to use the > extended OPF formulation in MATPOWER, unless of course you have some > requirement the precludes that option. > > Ray > > >> On Apr 9, 2015, at 10:40 PM, Abhyankar, Shrirang G. <[email protected] >> <mailto:[email protected]>> wrote: >> >> http://www.pserc.cornell.edu//matpower/docs/ref/ >> <http://www.pserc.cornell.edu//matpower/docs/ref/> >> >> On Apr 9, 2015, at 4:00 PM, "Electric" <[email protected] >> <mailto:[email protected]>> wrote: >> >>> Dear Shri >>> Thanks for your answer. But I don't want to use runopf(). I want to write >>> these constraint manually with Ybus and equations I've written. Is that >>> possible? How should I start? >>> >>> Thank you. >>> >>> On Fri, Apr 10, 2015 at 1:24 AM, Abhyankar, Shrirang G. >>> <[email protected] <mailto:[email protected]>> wrote: >>> I don't know what your algorithm does but I think you can do the same, >>> determine optimal generation and load, using MATPOWER's optimal power flow >>> (runopf). runopf supports all the constraints that you've listed. >>> >>> >>> Shri >>> >>> >>> >>> From: [email protected] >>> <mailto:[email protected]> >>> [[email protected] >>> <mailto:[email protected]>] on behalf of Electric >>> [[email protected] <mailto:[email protected]>] >>> Sent: Thursday, April 09, 2015 3:18 PM >>> To: MATPOWER discussion forum >>> Subject: How to verify the feasibility of your solution without running >>> Power flow? >>> >>> I am using an algorithm that changes the amount of demand and >>> generation at each bus (Pd,Pg) to mitigate over load of lines and enhance >>> the security. The output of my algorithm (Pd,Pg) must be checked to insure >>> that they are feasible. So, it is obvious that by assigning this new >>> calculated values instead of old Pg and Pd (generation and demand ) at each >>> bus and running a simple power flow (pf), I can find out if the calculated >>> results are feasible. >>> However, this is an optimization problem. Therefore, feasibility of >>> the suggested values of Pd and Pg by fmincon must be verified at each >>> iteration(1000 times). Basically, it can be accomplished by running a power >>> flow in constraint function of fmincon and checking lines loading and >>> voltage range. >>> But, doing so is very time consuming and takes more than an hours to >>> calculate the feasible Pg and Pd. So I was wonder if there is another >>> faster way to check the feasibility of the calculated Pd,Pg without running >>> power flow. >>> I think, if the solution (calculated Pd,Pg) satisfies these constraints,the >>> answer is then feasible and there is no need to run power flow. >>> But I have some problems in considering the constraints 5,8,9,10. >>> >>> 1) Pgmin(j) <Pg(j) <Pgmax(j) % j indicates buses with generator >>> 2) Qgmin(j) <Qg(j) <Qgmax(j) % j indicates buses with generator >>> 3) Pdmin(k) <Pd(k) <Pdmax(k) % k indicates buses with demand >>> 4) Qd(k)=Pd(k)*tan(phi(k)) % ph(k) is load P.F at bus K >>> Nodal active balance: >>> 5) Pg(n) -Pd(n)=|V(i)| sum( |Y(h,n)| .V(h)*Cos(del(n)-del(h)-teta(n,h))) >>> where Pg(n) is nodal generation which is calculated as: >>> 6) Pg(n)=sum (Pg(j)) at bus n >>> 7) Pd(n)=sum (Pd(k)) at bus n >>> teta(n,h) is the angle of Y(n,h), del(n) and del(h) are bust angle of n and >>> h. >>> Nodal reactive balance: >>> 8) Qg(n) -Qd(n)=|V(i)| sum( |Y(h,n)| .V(h)*Sin(del(n)-del(h)-teta(n,h))) >>> >>> 9) Vmin(n)<V(n)<Vmax(n) % voltage range at all buses. >>> 10) S(i,j)<Smax(i,j). transmitted apparent power from i,j must be less >>> than capacity of line. >>> >>> Thanks in advance >>> >>> >>> > > >
