sogno-platform / pyvolt Goto Github PK

At the end of this quickstart example script, the calculated voltage values are compared with precalculated references.
However, doing an assertion of complex floats is not a good idea and usually fails because of numerical tolerances.
So I'd suggest to do the following:

voltages_ref = [(1-7.970485900477431e-27j), (0.9521818868802214-0.11692768153747995j),
                (0.9642955926931457-0.09862127081290231j), (0.8796973782245792-0.15318580971335868j),
                (0.8758799767843979-0.1554670528853566j), (0.8774339089327876-0.1545390713879984j),
                (0.8704521134131005-0.1574589214738466j), (0.8719578342107204-0.15661905676450852j),
                (0.8731024990161087-0.1561497673751958j), (0.8727417602545003-0.15623359013901236j),
                (0.8740410220986286-0.1565630049874985j), (0.9563016474701451-0.09917826765833906j),
                (0.9592141716734833-0.09896267637101246j), (0.8702137462858025-0.15760036065945185j),
                (0.9239489705253996-0.13105032262255972j)]
epsilon = 1e-4
assert ([val-epsilon for val in voltages] <= voltages_ref <= [val+epsilon for val in voltages]), \
    "Results do not match reference results."

Please let me know if I should push that change.

Currently, almost all quickstart and validation examples are broken, because they depend on different (old) versions of the cimpy package.
This should be fixed.

As a reference, the most recent master branch version of the cimpy package should be used:
https://github.com/cim-iec/cimpy

When a branch of type PowerTransformer is created, there is a bug in the function _get_primary_connection. This function shoud not return the PowerTransformerEnd with the higheist nominal voltage but also the PowerTransformerEnd with r>0 and x>0. Independently of the nominal voltage of the PowerTransformerEnd, the parameters x, r can be equal to zero, which results in a singular admittance matrix.

If the underlying CIM data is reloaded ( using function network.System.load_cim_data(...) ), the power flow algorithm fails because the network model is not created properly in this case.

Describe the bug
Installing the Pyvolt package in an up-to-date environment and executing one of the example scripts leads to an error.
This is caused by changes in Numpy that has removed support for numpy internal datatypes (np.int, np.complex etc.) and replaced them with Python default datatypes.

To avoid this problem in the future Versions should be pinned.
While at it the project could/should be transitioned to the currents standard of PEP621 for project metadata (see e. g. Setuptools explaination

To Reproduce
Steps to reproduce the behavior:

1. Pull repo and cd into it
2. (create venv)
3. pip install .
4. python examples/quickstart/run_nv_powerflow.py

If a node inside the network typology is an instance of BusType.PV, the powerflow algorithm either produces abstruse results or does not converge at all.

According to file network.py, a node is declared as "PV" type if at least one synchronous machine is connected to it:

I am aware of the ToDo remark, so please don't get me wrong...

So, I checked the code in nv_powerflow.py and I am wondering about the following (unfortunately, I don't know the working principle of the algorithm in detail):

• According to line 9, variable "z" takes current values.
• However, in line 48, a power value is actually assigned to "z".
• Also, why do we assign a real value to z[m + 1] ? Because usually z[m] takes the real values and z[m+1] the imaginary ones if I understand it correctly.
• That given, is the code in lines 78 and 79 correct?

And by the way, if I set a node containing a synchronous machine to type "PQ" instead of "PV", the algorithm seems to work properly.
But this is most certainly not the way to go, I guess...

Best regards,
Sebastian

Problem description
While using the algorithm of the state estimation our developement team noticed something unusual in the behaviour of the function.
Two behaviours occured with our input:

• State estimation function went into an infinite loop in this function epsilon value is never converging to the wanted result.
• When the epsilon value converges the return are some values that are not accurate and descriptive of the grid (like Nan and infinite values)

To Reproduce
Build an example like this one but instead of the Voltage magnitude and phase input provide in input Voltae magnitude(MeasType.V_mag), active and reactive power.

Expected behavior
Return a state of all the input grid with relevant values trough the Result class included in the pyvolt.result module, where all nodes and branches informations after estimation are contained.

Additional context
The issue has already been discussed with Jan Dinkelbach, member of the community, who gave to the team support and noticed that this behaviour is not normal and could be fixed.