You have about 283 generator buses that are marked as PQ buses, rather than PV
buses. This is what is producing the problem with the AC power flow not
converging immediately when run using the solution as an input. I think this
should probably be considered a bug in MATPOWER.
The other thing is that you have 4 lines with zero reactance. That is where the
NaN's come from in the DC power flow solution. And yes, the DC power flow is
simply the solution of a linear set of equations, so no iteration is necessary.
>> find(mpc.branch(:, BR_X) == 0)
ans =
3776
3784
5471
5959
--
Ray Zimmerman
Senior Research Associate
419A Warren Hall, Cornell University, Ithaca, NY 14853
phone: (607) 255-9645
On Mar 14, 2012, at 4:19 AM, Hua Bowen wrote:
> Thank you Dr.Zimmerman.
>
> Here's a single slack bus version of the same network. Oddly enough,
> the first problem you mentioned still exist.
>
> Also I noticed that the DC power flow in MATPOWER is simply done by
> matrix left division. I wonder that DC load flow doesn't need
> iteration and why the result is NaN?
>
>
> On Tue, Mar 13, 2012 at 10:01 PM, Ray Zimmerman <[email protected]> wrote:
>> Not sure what's going on here, but I wanted to pass on what I've noticed
>> initially. Something is strange. Normally the following should produce
>> something that converges immediately without iterating at all the second
>> time, but in this case it diverges ...
>>
>> opt = mpoption('OUT_BUS', 0, 'OUT_BRANCH', 0, 'VERBOSE', 2);
>> mpc = loadcase('grid');
>> r0 = runpf(mpc, opt);
>> r = runpf(r0, opt);
>>
>> I also noticed that there are 3 reference buses even though it is a fully
>> connected network (which I assume is an error). When I change two of the
>> reference buses to PV buses, the AC PF no longer converges.
>>
>> Hopefully, this may trigger some ideas for ways to track this down. But, the
>> fact that the 2nd run above diverges makes me think I should double-check
>> how the code handles cases with multiple reference buses. I probably need
>> better error checking somewhere.
>>
>> --
>> Ray Zimmerman
>> Senior Research Associate
>> 419A Warren Hall, Cornell University, Ithaca, NY 14853
>> phone: (607) 255-9645
>>
>>
>>
>>
>> On Mar 13, 2012, at 4:44 AM, Hua Bowen wrote:
>>
>> Dear all,
>>
>> I have a case with more than 5,000 nodes. I wonder why the DC power
>> flow results in meaningless solution (the angle of all nodes turns out
>> to be NaN) while AC power flow has a normal result. Some branches have
>> a reactance as small as 1e-6 p.u. and zero resistance. I thought that
>> make the B matrix ill-conditioned. But even after I changed all those
>> small reactances to 1e-2, the result didn't change. I couldn't figure
>> it out.
>>
>> The case file is included in the attachment. Thank you for your help.
>>
>>
> <grid.m>
