susantoj / pypower-dynamics Goto Github PK

In this branch I have been experimenting with a more flexible structure for case data:

https://github.com/rwl/PYPOWER-Dynamics/tree/pypf

I didn't want to have to update all of the OPF functions in PYPOWER so I moved the necessary PF functions into a new pypf package. There is still a dependency on PYPOWER for the remaining functions.

The new case data classes should make it straightforward to add columns for new attributes. By way of demonstration, I committed the 3 phase fault analysis function by @susantoj and added faultMVA, Rgen and Xdpp attributes.

I took the liberty of doing away with the BUS_I column. The former GEN_BUS, F_BUS and T_BUS columns are now assumed to be the zero-based bus index. This removes the need for ext2int and int2ext conversions, but may have some as yet unforeseen downsides. A convert_ppc function allows PYPOWER case data dicts to be reused and performs the necessary conversions.

Hi I have been trying to use pypower dynamics in python3.

So far I have found the following error:
IndexError: only integers, slices (:), ellipsis (...), numpy.newaxis (None) and integer or boolean arrays are valid indices

This happens becase the indices are floats instead of integers. The affected function (there might be more) is

def mod_Ybus(Ybus, elements, bus, gen, baseMVA):
    # Add equivalent generator and grid admittances to Ybus matrix
    for element in elements.values():
        Ye = 0
        
        # 4th/6th order machines and converters
        if element.__module__ in ['pydyn.sym_order4', 'pydyn.sym_order6a', 'pydyn.sym_order6b', 'pydyn.vsc_average']:
            i = gen[element.gen_no,0]
            Ye = element.Yg
        
        # External grid
        if element.__module__ == 'pydyn.ext_grid':
            i = gen[element.gen_no,0]
            Ye = 1 / (1j * element.params['Xdp'])
        
        if Ye != 0:
            Ybus[i,i] = Ybus[i,i] + Ye

    # Add equivalent load admittance to Ybus matrix    
    Pl, Ql = bus[:, PD], bus[:, QD]
    for i in range(len(Pl)):
        S_load = (Pl[i] - 1j * Ql[i]) / baseMVA
        y_load = S_load / (bus[i, VM] ** 2)
        Ybus[i,i] = Ybus[i,i] + y_load
    
    return Ybus

Recorder object only works for signals and states (not network variables)

Generator current injections do not appear to be correct when generator parameters exhibit saliency.

There is a ) bracket missing from the equation for Vfd0 of the Sauer-Pai model at the bottom of the Machine Models page of the wiki:

https://github.com/susantoj/PYPOWER-Dynamics/wiki/3.4.-Machine-Models#6th-order-synchronous-machine-sauer-pai

Currently when the program fails, it just crashes and doesn't give any meaningful guidance on what was done incorrectly.

Currently, interfacing of controllers only works for signals in machine models or other controllers. Augment to allow interfacing with network variables (e.g. voltages, currents, etc)

Faults are modelled as high shunt admittances / conductances and use the shunt columns (Gs, Bs) in the PYPOWER case file. This is okay if there are no actual shunts on the bus, but where a shunt exists, then the fault and fault clearing events overwrite the existing shunt values.

Need to fix this.

An __init__.py file is required in the pydyn directory to make Python treat it as a package.

https://docs.python.org/3/tutorial/modules.html#packages

The variable yi is not defined here:

The fix looks straightforward. I'll send a PR soon.

Add saturation to rotor field current.

Add tests for core functions

In order to get controllers to work with Python 2.7 I had to wrap the replace operation with a unicode call in the neg_token function. I suspect this is a Python 2 only issue and has no effect when using Python 3.

            if unicode(x.replace('.','0')).isnumeric() == True:

As of version 5.1, MATPOWER is available under the terms of a 3-clause BSD license and no longer uses the GPLv3. PYPOWER has been updated and released under the same BSD license.

rwl/PYPOWER@d176356

PYPOWER-Dynamics might also be updated to use the BSD license. This would bring it in line with the latest MATPOWER version and with its derived projects.

The switch would allow those unable to work with the GPL license to use PYPOWER-Dynamics and not turn to the older and less capable PyDyn project:

https://github.com/rwl/pydyn

With the exception of PSAT, I am not currently aware of any open source library capable of Small Signal Stability Analysis (SSSA).

Section 7 of "Power System Modelling and Scripting" by Federico Milano includes some useful information and snippets of code.

The eigs function from SciPy provides a convenient interface to ARPACK that would do the hard work.

Does anyone know how to find the eigenvalues for a PYPOWER-Dynamics model?