Hi Jiashen,
You simply need a good LP solver. Check out the Optional Packages
<http://www.pserc.cornell.edu/matpower/#optionalpackages> to see if any will
work for you. CPLEX and Gurobi are probably the best ones and there are free
licenses for academics. The MIPS solver included in MATPOWER was designed for
non-linear problems, like the AC OPF. Since an LP is a special case, it can
often solve LPs too, but it is not very numerically robust as an LP solver,
which is what you are encountering. I’m not sure exactly what it is that makes
the difference between the cases that it solves and those it doesn’t.
Ray
> On Mar 13, 2015, at 6:38 AM, Jiashen Teh <[email protected]>
> wrote:
>
> Dear All,
>
> I attached a simple 6 bus network in the above. It is a very well known test
> network in the realm of reliability studies.
>
> I would like to know why when I increase the load and generation up to a
> certain point, the DCOPF sometimes converges , sometimes do not.
> Take for example, generation and load increase by 1.9 times do not converge.
> However, increment by 2.3 times converges. In both cases, the network is
> under perfect conditions.
>
> Why is this case?
>
>
> A simple testing algorithm:
>
> Yours sincerely,
>
> Jiashen Teh
> PhD Student
> Electrical Energy & Power Systems Group, School of Electrical & Electronic
> Engineering
> Ferranti Building (B18), The University of Manchester, M13 9PL, United Kingdom
> Tel: +44 (0) 161 306 2263; Mobile: +44 (0) 792 322 4864
>
> <SixBus.m><A simple testing algorithm.txt>
