OPF on the radial system offered by MATPOWER

[young-hwanlee]

I ran the OPF on the radial systems offered by MATPOWER, e.g. case10ba, case51ga, case51he, case69, case70da, case74ds, case85, case94pi, case118zh, case136ma, and case141. But it's not converged except case51ga, case51he, case69, and case141. I want to successfully run the OPF on the radial systems in all cases. Could you tell me how to do?

[rdzman]

The first thing I usually try when an OPF does not converge is to turn on OPF soft limits (see Section 7.6.4 in the MATPOWER User's Manual) to see if there is an infeasibility.

For example ...

mpc = loadcase('case10ba');
mpc = toggle_softlims(mpc, 'on');
r = runopf(mpc);

... results in ...

================================================================================
|     Soft Voltage Upper Bounds                                                |
================================================================================
Bus    Voltage   Limit   Overload    mu
  #    Mag(pu)   (pu)     (pu)     ($/pu)
-----  -------  -------  -------  ---------
    1   1.052    1.000    0.052  99999.999
                        --------
               Total:      0.05

Here's what I found for the cases you listed by running them with OPF soft limits:

• case10ba - infeasible to due voltage limits
• case51ga - converges
• case51he - converges
• case69 - converges
• case70da - infeasible to due voltage limits
• case74ds - converges
• case85 - infeasible to due voltage limits
• case94pi - crashes with soft limits enabled 😳 (I need to look into this one!)
• case118zh - infeasible to due voltage limits
• case136ma - infeasible to due generator limits
• case141 - converges

[rdzman]

Looks like case94pi has an error. It has 1 row in gen, but 2 in gencost. Deleting the duplicate row in gencost results in ...

• case94pi - infeasible due to voltage limits

I will fix that error in case94pi for the next release.

[young-hwanlee]

There is another error when using ext2int and toggle_softlims. Is there anyway to use both of them?

case_num = 'case10ba';

mpopt = mpoption('out.all', 0);
mpc = ext2int(loadcase(case_num));
mpc = toggle_softlims(mpc, 'on');

result = runopf(mpc, mpopt);

results in...

Unrecognized field name "softlims".

Error in toggle_softlims>userfcn_softlims_formulation (line 255)
    s = mpc.softlims.(lim);

Error in run_userfcn (line 32)
        rv = feval(userfcn.(stage)(k).fcn, rv, varargin{2:end}, args);

Error in opf_setup (line 514)
om = run_userfcn(userfcn, 'formulation', om, mpopt);

Error in opf (line 225)
om = opf_setup(mpc, mpopt);

Error in runopf (line 75)
[r, success] = opf(casedata, mpopt);

Error in test_code (line 55)
result = runopf(mpc, mpopt);

[rdzman]

The runopf() function calls ext2int() internally. Is there a reason you need to call it explicitly beforehand?

If so, you can always remove the order field from mpc before calling runopf(), but that will prevent you from being able to call int2ext() on the result to get back to the original numbering. If you need to be able to do that, you'd have to save the order field from the mpc and restore it to the result before calling int2ext().

mpc = ext2int(loadcase(case_num));
mpc = toggle_softlims(mpc, 'on');
mpc = rmfield(mpc, 'order');
result = runopf(mpc, mpopt);