Dear Dr. Zimmerman, Thanks for your suggestion. I've tried different solvers including PDIPM, SCIPM, and TRALM, etc. All of these solvers still cannot converge while solving the IEEE 123 nodes distribution system for an OPF solution. So I'm wondering if it is diverging because the 123 nodes is a distribution system with a lot end branches not in a loop. Also I notice the lines in the 123 nodes have very similar impedence (which is typical in a distribution system), will this cause singular or bad scaled matrix while solving for OPF?
Thanks for your time! *Fanjun Meng * On Thu, Jun 30, 2011 at 8:07 AM, Ray Zimmerman <[email protected]> wrote: > Besides the good suggestions you got from others, you may also want to try > a different solver. I assume you are using the default MIPS solver. > Sometimes it does encounter numerical trouble where other algorithms do not. > I haven't had the opportunity to dig into things to understand under what > circumstances this happens. > > -- > Ray Zimmerman > Senior Research Associate > 211 Warren Hall, Cornell University, Ithaca, NY 14853 > phone: (607) 255-9645 > > > > On Jun 29, 2011, at 8:51 AM, Fanjun Meng wrote: > > The voltage is given 4.16kV. I convert the impedances to p.u. by myself > with 1MVA base. Actually I have checked this model in 'runpf()' and the > result is almost the same as shown in IEEE 123 node profile. So I guess the > problem might be the generation cost setting. I'm trying to fix that > problem. > > Thank you very much! > > > *Fanjun Meng > Electrical Engineering > Missouri University of Science & Technology > [email protected] <[email protected]>* > > > > On Wed, Jun 29, 2011 at 2:39 AM, Simon Schneider <[email protected]>wrote: > >> ** >> Are the nodal voltage limits set wide enough(e.g. 0.9 an 1.1)? >> Another possible source could be a bad choice of baseMVA and base voltage. >> >> Do you convert the impedances by yourself or are they already given in >> p.u? >> >> Regards, >> Simon >> Am 6/28/2011 10:05 PM, schrieb Fanjun Meng: >> >> Thanks a lot! >> >> *Fanjun Meng* >> >> >> >> On Tue, Jun 28, 2011 at 12:36 PM, Santiago Torres < >> [email protected]> wrote: >> >>> Maybe the generation costs you are using for the opf are causing >>> problems. Try to check that!. >>> >>> Regards, >>> >>> Santiago >>> >>> 2011/6/28 Fanjun Meng <[email protected]> >>> >>>> Thank you very much for your suggestion! >>>> Actually I did check the simple power flow result with constraints, the >>>> voltages, line flows, generations are all satisfied. That's why I'm >>>> confused >>>> with this problem. Is it possible that the line impedences and system >>>> topology may cause matrix sigularity during OPF calculation iteration and >>>> fail to converge? >>>> >>>> Thanks a lot! >>>> >>>> *Fanjun Meng >>>> * >>>> >>>> >>>> >>>> >>>> On Tue, Jun 28, 2011 at 10:13 AM, Santiago Torres < >>>> [email protected]> wrote: >>>> >>>>> The simple power flow has not constraints, so maybe with the settings >>>>> you are using the power flow converges and has solution, however when >>>>> you use the opf, some constraints can cause it does not converge. Check >>>>> if >>>>> the IEEE 123 node reported solution when using the simple power flow >>>>> is complying the constraints (ie voltages, power generation, power flows). >>>>> >>>>> I hope this can help. >>>>> >>>>> Santiago Torres >>>>> >>>>> 2011/6/28 Fanjun Meng <[email protected]> >>>>> >>>>>> Hello, >>>>>> >>>>>> I've been working on IEEE 123 nodes system and created a model for >>>>>> MatPower power flow. This model is basically a radial distribution system >>>>>> with a single generation at substation and works fine when I can use >>>>>> runpf(), and it returns similar result as shown IEEE 123 node system >>>>>> report. >>>>>> However, when I use the same model for runopf(), it shows >>>>>> >>>>>> >>>>>> Numerically Failed >>>>>> Did not converge in 4 iterations. >>>>>> Did not converge (0.10 seconds) >>>>>> >>>>>> The total generation is very different from total demand, and bus >>>>>> voltages are at strange values (e.g. 1.836p.u. at reference bus). Also, >>>>>> generation cost Lambda are extremely high (e.g. 17528752586.76 $/MWh @ >>>>>> bus >>>>>> 6). I have tried to add multiple DGs at different locations and change >>>>>> the >>>>>> system to a ring type system, but it didn't work out. Does anyone have >>>>>> any >>>>>> suggestions or experiences to this problem? Your help is highly >>>>>> appreciated! >>>>>> >>>>>> Have a good day!* >>>>>> >>>>>> Fanjun Meng >>>>>> * >>>>>> >>>>>> >>>>> >>>>> >>>>> -- >>>>> Dr.-Ing. Santiago Torres >>>>> IEEE Senior Member >>>>> >>>>> Post-Doctoral Fellow >>>>> School of Electrical and Computer Engineering >>>>> University of Campinas, Campinas, SP, Brazil >>>>> >>>>> http://www.dsee.fee.unicamp.br/ >>>>> >>>>> Albert Einstein, 400 >>>>> 13083-852, Campinas, SP, Brazil >>>>> >>>> >>>> >>> >>> >>> -- >>> Dr.-Ing. Santiago Torres >>> IEEE Senior Member >>> >>> Post-Doctoral Fellow >>> School of Electrical and Computer Engineering >>> University of Campinas, Campinas, SP, Brazil >>> >>> http://www.dsee.fee.unicamp.br/ >>> >>> Albert Einstein, 400 >>> 13083-852, Campinas, SP, Brazil >>> >> >> >> > >
