Awesome
GridCal
GridCal is a top tier power systems planning and simulation software. As such it has all the static analysis studies that you can think of, plus linear and non-linear optimization functions. Some of these functions are well known, while others you may have never heard of as they are a product of cutting-edge research.
GridCal started in 2015 with a clear objective: create a solid programming library and a user-friendly interface. This straightforward approach sparked many innovations — some driven by the necessity for commercial use, and others fueled by curiosity and research.
Whether you're a pro needing free tools, a researcher wanting a real-world tested platform, a teacher sharing commercial-grade software insights, or a student diving into practical algorithms, GridCal's got your back. It's a high quality product made for all of us now and for the future generations.
Installation
GridCal is a software made in the Python programming language. Therefore, it needs a Python interpreter installed in your operative system.
The GridCal project is divided in three packages:
- GridCalEngine: A package with the database and calculations logic.
- GridCalServer: A package that serves an API-rest to use GridCalEngine remotelly.
- GridCal: A package that contains the Graphical User Interface (GUI) and operates with
GridCalEngineandGridCalServerseamlessly.
To install everything, you only need to install the GridCal package and the others will beinstalled as dependencies.
<img height="200" src="doc/rst_source/figures/Packages.png"/>Standalone setup
If you don't know what is this Python thing, we offer a windows installation:
This will install GridCal as a normal windows program and you need not to worry about any of the previous instructions. Still, if you need some guidance, the following video might be of assistance: Setup tutorial (video).
Package installation
We recommend to install the latest version of Python and then, install GridCal with the following terminal command:
pip install GridCal
You may need to use pip3 if you are under Linux or MacOS, both of which
come with Python pre-installed already.
Install into an environment
python3 -m venv gc5venv
source gc5venv/bin/activate
pip install GridCal
gridcal
Run the graphical user interface
Once you install GridCal in your local Python distribution, you can run the graphical user interface with the following terminal command:
gridcal
If this doesn't work, try:
python -c "from GridCal.ExecuteGridCal import runGridCal; runGridCal()"
You may save this command in a shortcut for easy future access.
Install only the engine
Some of you may only need GridCal as a library for some other purpose like batch calculations, AI training or simple scripting. Whatever it may be, you can get the GridCal engine with the following terminal command:
pip install GridCalEngine
This will install the GridCalEngine package that is a dependency of GridCal.
Again, you may need to use pip3 if you are under Linux or MacOS.
Features
GridCal is packed with feautures:
- Large collection of devices to model electricity grids
- AC/DC multi-grid power flow
- AC/DC multi-grid linear optimal power flow
- AC linear analysis (PTDF & LODF)
- AC linear net transfer capacity calculation
- AC+HVDC optimal net transfer capacity calculation
- AC/DC Stochastic power flow
- AC Short circuit
- AC Continuation power flow
- Contingency analysis (Power flow and LODF variants)
- Sigma analysis (one-shot stability analysis)
- Investments analysis
- Bus-branch schematic
- Substation-line map diagram
- Time series and snapshot for most simulations
- Overhead tower designer
- Inputs analysis
- Model bug report and repair
- Import many formats (PSSe .raw/rawx, epc, dgs, matpower, pypsa, json, cim, cgmes)
- Export in many formats (gridcal .xlsx/.gridcal/.json, cgmes, psse .raw/.rawx)
All of these are industry tested algoriths, some of which surpass most comemercially available software. The aim is to be a drop-in replacement for the expensive and less usable commercial software, so that you can work, research and learn with it.
Resources
In an effort to ease the simulation and construction of grids, We have included extra materials to work with. These are included in the standalone setups.
- Load profiles for your projects.
- Grids from IEEE and other open projects.
- Equipment catalogue (Wires, Cables and Transformers) ready to use in GridCal.
Tutorials and examples
-
GridCal PlayGround repository with some notebooks and examples.
-
The tests may serve as a valuable source of examples.
Matpower grids
Matpower's excellent formulations and consistency has allowed this and other projects to develop, relying on its sound math. That is why GridCal reads Matpower cases out of the box, without you having to do anything special. And of course, GridCal solves all Matpower 8 provided grids, solving the continental USA case in about 1 second:
Find the results: All matpower grids.xlsx, and the benchmark code matpower_grids.py. _Results simulated with AMD 9750x and 64 GB of RAM under Ubuntu 24.04. All solved using Newton-Raphson, and only using the provided solution that comes with the files when the flat start fails.
Cool right?
API
Since day one, GridCal was meant to be used as a library as much as it was meant to be used from the user interface. Following, we include some usage examples, but feel free to check the documentation out where you will find a complete description of the theory, the models and the objects.
Understanding the program structure
All simulations in GridCal are handled by the simulation drivers. The structure is as follows:
<img height="250" src="doc/rst_source/figures/DataModelSimulation.png"/>Any driver is fed with the data model (MultiCircuit object), the respective driver options, and often another
object relative to specific inputs for that driver. The driver is run, storing the driver results object.
Although this may seem overly complicated, it has proven to be maintainable and very convenient.
Snapshot vs. time series
GridCal has dual structure to handle legacy cases (snapshot), as well as cases with many variations (time series)
-
A snapshot is the grid for a particular moment in time. This includes the infrastructure plus the variable values of that infraestructure such as the load, the generation, the rating, etc.
-
The time series record the variations of the magnitudes that can vary. These are aplied along with the infrastructure definition.
In GridCal, the inputs do not get modified by the simulation results. This very important concept, helps maintaining the independence of the inputs and outputs, allowing the replicability of the results. This key feature is not true for other open-source of comercial programs.
A snapshot or any point of the time series, may be compiled to a NumericalCircuit. This object holds the
numerical arrays and matrices of a time step, ready for the numerical methods.
For those simulations that require many time steps, a collection of NumericalCircuit is compiled and used.
It may seem that this extra step is redundant. However the compilation step is composed by mere copy operations, which are fast. This steps benefits greatly the efficiency of the numerical calculations since the arrays are aligned in memory. The GridCal data model is object-oriented, while the numerical circuit is array-oriented (despite beign packed into objects)
Loading a grid
import GridCalEngine as gce
# load a grid (.gridcal, .m (Matpower), .raw (PSS/e) .rawx (PSS/e), .epc (PowerWorld), .dgs (PowerFactory)
my_grid = gce.open_file("my_file.gridcal")
In the case of CIM/CGMES, you may need to pass a list of files or a single zip file:
import GridCalEngine as gce
# load a grid from many xml files
my_grid = gce.open_file(["grid_EQ.xml", "grid_TP.xml", "grid_SV.xml", ])
# or from a single zip assumed to contain CGMES files
my_grid = gce.open_file("my_cgmes_set_of_files.zip")
# or load a grid from a combination of xml and zip files assumed to be CGMES
my_grid = gce.open_file(["grid_EQ.xml", "grid_TP.xml", "grid_SV.xml", "boundary.zip"])
If you need to explore the CGMEs assets before conversion, you'll need to dive deeper in the API:
import GridCalEngine as gce
fname = "tests/data/grids/CGMES_2_4_15/IEEE 118 Bus v2.zip"
logger = gce.Logger()
data_parser = gce.CgmesDataParser()
data_parser.load_files(files=[fname])
cgmes_circuit = gce.CgmesCircuit(cgmes_version=data_parser.cgmes_version,
cgmes_map_areas_like_raw=False, logger=logger)
cgmes_circuit.parse_files(data_parser=data_parser)
# print all the ac line segment names
for ac_line_segment in cgmes_circuit.cgmes_assets.ACLineSegment_list:
print(ac_line_segment.name)
# print the logs
logger.print()
GridCal supports many file formats:
- CIM 16 (.zip and .xml)
- CGMES 2.4.15 and 3.0 (.zip and .xml)
- PSS/e raw and rawx versions 29 to 35, including USA market exchange RAW-30 specifics.
- Matpower .m files directly.
- DigSilent .DGS (not fully compatible)
- PowerWorld .EPC (not fully compatible, supports substation coordinates)
Simmilarly to CGMES you may be able to use the conversion objects to explore the original formats.
Save a grid
import GridCalEngine as gce
# load a grid
my_grid = gce.open_file("my_file.gridcal")
# save
gce.save_file(my_grid, "my_file_2.gridcal")
In the case of saving a model in CGMES mode, we need to specify some extra parameters.
To simplify we can use the API function save_cgmes_file:
import GridCalEngine as gce
# load a grid
my_grid = gce.open_file("my_file.gridcal")
# run power flow (this is optional and it is used to generate the SV profile)
pf_results = gce.power_flow(my_grid)
# save the grid in CGMES mode
gce.save_cgmes_file(grid=my_grid,
filename="My_cgmes_model.zip",
cgmes_boundary_set_path="path_to_the_boundary_set.zip",
cgmes_version=CGMESVersions.v2_4_15,
pf_results=pf_results)
Creating a Grid using the API objects
We are going to create a very simple 5-node grid from the excellent book Power System Load Flow Analysis by Lynn Powell.
import GridCalEngine as gce
# declare a circuit object
grid = gce.MultiCircuit()
# Add the buses and the generators and loads attached
bus1 = gce.Bus('Bus 1', Vnom=20)
# bus1.is_slack = True # we may mark the bus a slack
grid.add_bus(bus1)
# add a generator to the bus 1
gen1 = gce.Generator('Slack Generator', vset=1.0)
grid.add_generator(bus1, gen1)
# add bus 2 with a load attached
bus2 = gce.Bus('Bus 2', Vnom=20)
grid.add_bus(bus2)
grid.add_load(bus2, gce.Load('load 2', P=40, Q=20))
# add bus 3 with a load attached
bus3 = gce.Bus('Bus 3', Vnom=20)
grid.add_bus(bus3)
grid.add_load(bus3, gce.Load('load 3', P=25, Q=15))
# add bus 4 with a load attached
bus4 = gce.Bus('Bus 4', Vnom=20)
grid.add_bus(bus4)
grid.add_load(bus4, gce.Load('load 4', P=40, Q=20))
# add bus 5 with a load attached
bus5 = gce.Bus('Bus 5', Vnom=20)
grid.add_bus(bus5)
grid.add_load(bus5, gce.Load('load 5', P=50, Q=20))
# add Lines connecting the buses
grid.add_line(gce.Line(bus1, bus2, name='line 1-2', r=0.05, x=0.11, b=0.02))
grid.add_line(gce.Line(bus1, bus3, name='line 1-3', r=0.05, x=0.11, b=0.02))
grid.add_line(gce.Line(bus1, bus5, name='line 1-5', r=0.03, x=0.08, b=0.02))
grid.add_line(gce.Line(bus2, bus3, name='line 2-3', r=0.04, x=0.09, b=0.02))
grid.add_line(gce.Line(bus2, bus5, name='line 2-5', r=0.04, x=0.09, b=0.02))
grid.add_line(gce.Line(bus3, bus4, name='line 3-4', r=0.06, x=0.13, b=0.03))
grid.add_line(gce.Line(bus4, bus5, name='line 4-5', r=0.04, x=0.09, b=0.02))
Power Flow
Using the simplified API:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE39_1W.gridcal')
main_circuit = gce.open_file(fname)
results = gce.power_flow(main_circuit)
print(main_circuit.name)
print('Converged:', results.converged, 'error:', results.error)
print(results.get_bus_df())
print(results.get_branch_df())
Using the more complex library objects:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE14_from_raw.gridcal')
main_circuit = gce.open_file(fname)
options = gce.PowerFlowOptions(gce.SolverType.NR, verbose=False)
power_flow = gce.PowerFlowDriver(main_circuit, options)
power_flow.run()
print(main_circuit.name)
print('Converged:', power_flow.results.converged, 'error:', power_flow.results.error)
print(power_flow.results.get_bus_df())
print(power_flow.results.get_branch_df())
Output:
IEEE14_from_raw
Converged: True error: 5.98e-08
Bus resuts:
Vm Va P Q
BUS 1 1.06 0.00 232.39 -16.55
BUS 2 1.04 -4.98 18.30 30.86
BUS 3 1.01 -12.73 -94.20 6.08
BUS 4 1.02 -10.31 -47.80 3.90
BUS 5 1.02 -8.77 -7.60 -1.60
BUS 6 1.07 -14.22 -11.20 5.23
BUS 7 1.06 -13.36 0.00 0.00
BUS 8 1.09 -13.36 0.00 17.62
BUS 9 1.06 -14.94 -29.50 -16.60
BUS 10 1.05 -15.10 -9.00 -5.80
BUS 11 1.06 -14.79 -3.50 -1.80
BUS 12 1.06 -15.08 -6.10 -1.60
BUS 13 1.05 -15.16 -13.50 -5.80
BUS 14 1.04 -16.03 -14.90 -5.00
Branch results:
Pf Qf Pt Qt loading
1_2_1 156.88 -20.40 -152.59 27.68 -2,040,429,074,673.33
1_5_1 75.51 3.85 -72.75 2.23 385,498,944,321.99
2_3_1 73.24 3.56 -70.91 1.60 356,020,306,394.25
2_4_1 56.13 -1.55 -54.45 3.02 -155,035,233,483.95
2_5_1 41.52 1.17 -40.61 -2.10 117,099,586,051.68
3_4_1 -23.29 4.47 23.66 -4.84 447,311,351,720.93
4_5_1 -61.16 15.82 61.67 -14.20 1,582,364,180,487.11
6_11_1 7.35 3.56 -7.30 -3.44 356,047,085,671.01
6_12_1 7.79 2.50 -7.71 -2.35 250,341,387,213.42
6_13_1 17.75 7.22 -17.54 -6.80 721,657,405,311.13
7_8_1 -0.00 -17.16 0.00 17.62 -1,716,296,745,837.05
7_9_1 28.07 5.78 -28.07 -4.98 577,869,015,291.12
9_10_1 5.23 4.22 -5.21 -4.18 421,913,877,670.92
9_14_1 9.43 3.61 -9.31 -3.36 361,000,694,981.35
10_11_1 -3.79 -1.62 3.80 1.64 -161,506,127,162.22
12_13_1 1.61 0.75 -1.61 -0.75 75,395,885,855.71
13_14_1 5.64 1.75 -5.59 -1.64 174,717,248,747.17
4_7_1 28.07 -9.68 -28.07 11.38 -968,106,634,094.39
4_9_1 16.08 -0.43 -16.08 1.73 -42,761,145,748.20
5_6_1 44.09 12.47 -44.09 -8.05 1,247,068,151,943.25
Inputs analysis
GridCal can perform a summary of the inputs with the InputsAnalysisDriver:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE 118 Bus - ntc_areas.gridcal')
main_circuit = gce.open_file(fname)
drv = gce.InputsAnalysisDriver(grid=main_circuit)
mdl = drv.results.mdl(gce.ResultTypes.AreaAnalysis)
df = mdl.to_df()
print(df)
The results per area:
P Pgen Pload Pbatt Pstagen Pmin Pmax Q Qmin Qmax
IEEE118-3 -57.0 906.0 963.0 0.0 0.0 -150000.0 150000.0 -345.0 -2595.0 3071.0
IEEE118-2 -117.0 1369.0 1486.0 0.0 0.0 -140000.0 140000.0 -477.0 -1431.0 2196.0
IEEE118-1 174.0 1967.0 1793.0 0.0 0.0 -250000.0 250000.0 -616.0 -3319.0 6510.0
Linear analysis
We can run an PTDF equivalent of the power flow with the linear analysys drivers:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE 5 Bus.xlsx')
main_circuit = gce.open_file(fname)
# snapshot
results = gce.linear_power_flow(grid=main_circuit)
print("Bus results:\n", results.get_bus_df())
print("Branch results:\n", results.get_branch_df())
print("PTDF:\n", results.mdl(gce.ResultTypes.PTDF).to_df())
print("LODF:\n", results.mdl(gce.ResultTypes.LODF).to_df())
Simulating with a more detailed control of the objects:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE 5 Bus.xlsx')
main_circuit = gce.open_file(fname)
options_ = gce.LinearAnalysisOptions(distribute_slack=False, correct_values=True)
# snapshot
sn_driver = gce.LinearAnalysisDriver(grid=main_circuit, options=options_)
sn_driver.run()
print("Bus results:\n", sn_driver.results.get_bus_df())
print("Branch results:\n", sn_driver.results.get_branch_df())
print("PTDF:\n", sn_driver.results.mdl(gce.ResultTypes.PTDF).to_df())
print("LODF:\n", sn_driver.results.mdl(gce.ResultTypes.LODF).to_df())
Output:
Bus results:
Vm Va P Q
Bus 0 1.0 0.0 2.1000 0.0
Bus 1 1.0 0.0 -3.0000 0.0
Bus 2 1.0 0.0 0.2349 0.0
Bus 3 1.0 0.0 -0.9999 0.0
Bus 4 1.0 0.0 4.6651 0.0
Branch results:
Pf loading
Branch 0-1 2.497192 0.624298
Branch 0-3 1.867892 0.832394
Branch 0-4 -2.265084 -0.828791
Branch 1-2 -0.502808 -0.391900
Branch 2-3 -0.267908 -0.774300
Branch 3-4 -2.400016 -1.000006
PTDF:
Bus 0 Bus 1 Bus 2 Bus 3 Bus 4
Branch 0-1 0.193917 -0.475895 -0.348989 0.0 0.159538
Branch 0-3 0.437588 0.258343 0.189451 0.0 0.360010
Branch 0-4 0.368495 0.217552 0.159538 0.0 -0.519548
Branch 1-2 0.193917 0.524105 -0.348989 0.0 0.159538
Branch 2-3 0.193917 0.524105 0.651011 0.0 0.159538
Branch 3-4 -0.368495 -0.217552 -0.159538 0.0 -0.480452
LODF:
Branch 0-1 Branch 0-3 Branch 0-4 Branch 1-2 Branch 2-3 Branch 3-4
Branch 0-1 -1.000000 0.344795 0.307071 -1.000000 -1.000000 -0.307071
Branch 0-3 0.542857 -1.000000 0.692929 0.542857 0.542857 -0.692929
Branch 0-4 0.457143 0.655205 -1.000000 0.457143 0.457143 1.000000
Branch 1-2 -1.000000 0.344795 0.307071 -1.000000 -1.000000 -0.307071
Branch 2-3 -1.000000 0.344795 0.307071 -1.000000 -1.000000 -0.307071
Branch 3-4 -0.457143 -0.655205 1.000000 -0.457143 -0.457143 -1.000000
Now let's make a comparison between the linear flows and the non-linear flows from Newton-Raphson:
import os
from matplotlib import pyplot as plt
import GridCalEngine as gce
plt.style.use('fivethirtyeight')
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE39_1W.gridcal')
main_circuit = gce.open_file(fname)
ptdf_driver = gce.LinearAnalysisTimeSeriesDriver(grid=main_circuit)
ptdf_driver.run()
pf_options_ = gce.PowerFlowOptions(solver_type=gce.SolverType.NR)
ts_driver = gce.PowerFlowTimeSeriesDriver(grid=main_circuit, options=pf_options_)
ts_driver.run()
fig = plt.figure(figsize=(30, 6))
ax1 = fig.add_subplot(131)
ax1.set_title('Newton-Raphson based flow')
ax1.plot(ts_driver.results.Sf.real)
ax1.set_ylabel('MW')
ax1.set_xlabel('Time')
ax2 = fig.add_subplot(132)
ax2.set_title('PTDF based flow')
ax2.plot(ptdf_driver.results.Sf.real)
ax2.set_ylabel('MW')
ax2.set_xlabel('Time')
ax3 = fig.add_subplot(133)
ax3.set_title('Difference')
diff = ts_driver.results.Sf.real - ptdf_driver.results.Sf.real
ax3.plot(diff)
ax3.set_ylabel('MW')
ax3.set_xlabel('Time')
fig.set_tight_layout(tight=True)
plt.show()
Linear optimization
import os
import numpy as np
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE39_1W.gridcal')
main_circuit = gce.open_file(fname)
# declare the snapshot opf
opf_options = gce.OptimalPowerFlowOptions(mip_solver=gce.MIPSolvers.HIGHS)
opf_driver = gce.OptimalPowerFlowDriver(grid=main_circuit, options=opf_options)
print('Solving...')
opf_driver.run()
print("Status:", opf_driver.results.converged)
print('Angles\n', np.angle(opf_driver.results.voltage))
print('Branch loading\n', opf_driver.results.loading)
print('Gen power\n', opf_driver.results.generator_power)
print('Nodal prices \n', opf_driver.results.bus_shadow_prices)
# declare the time series opf
opf_ts_driver = gce.OptimalPowerFlowTimeSeriesDriver(grid=main_circuit)
print('Solving...')
opf_ts_driver.run()
print("Status:", opf_ts_driver.results.converged)
print('Angles\n', np.angle(opf_ts_driver.results.voltage))
print('Branch loading\n', opf_ts_driver.results.loading)
print('Gen power\n', opf_ts_driver.results.generator_power)
print('Nodal prices \n', opf_ts_driver.results.bus_shadow_prices)
Run a linear optimization and verify with power flow
Often ties, you want to dispatch the generation using a linear optimization, to then verify the results using the power exact power flow. With GridCal, to do so is as easy as passing the results of the OPF into the PowerFlowDriver:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE39_1W.gridcal')
main_circuit = gce.open_file(fname)
# declare the snapshot opf
opf_driver = gce.OptimalPowerFlowDriver(grid=main_circuit)
opf_driver.run()
# create the power flow driver, with the OPF results
pf_options = gce.PowerFlowOptions(solver_type=gce.SolverType.NR)
pf_driver = gce.PowerFlowDriver(grid=main_circuit,
options=pf_options,
opf_results=opf_driver.results)
pf_driver.run()
# Print results
print('Converged:', pf_driver.results.converged, '\nError:', pf_driver.results.error)
print(pf_driver.results.get_bus_df())
print(pf_driver.results.get_branch_df())
Output:
OPF results:
Va P Shadow price
Bus 1 0.00 0.0 0.0
Bus 2 -2.22 0.0 0.0
Bus 3 -1.98 0.0 0.0
Bus 4 -2.12 0.0 0.0
Bus 5 -2.21 0.0 0.0
Pf Pt Tap angle Loading
Branch 1 -31.46 31.46 0.0 -44.94
Branch 1 -1.84 1.84 0.0 -10.20
Branch 1 -1.84 1.84 0.0 -9.18
Branch 1 0.14 -0.14 0.0 1.37
Branch 1 -48.30 48.30 0.0 -53.67
Branch 1 -35.24 35.24 0.0 -58.73
Branch 1 -4.62 4.62 0.0 -23.11
Power flow results:
Converged: True
Error: 3.13e-11
Vm Va P Q
Bus 1 1.00 0.00 1.17e+02 12.90
Bus 2 0.97 -2.09 -4.00e+01 -20.00
Bus 3 0.98 -1.96 -2.50e+01 -15.00
Bus 4 1.00 -2.61 2.12e-09 32.83
Bus 5 0.98 -2.22 -5.00e+01 -20.00
Pf Qf Pt Qt Loading
Branch 1 -31.37 -2.77 31.88 1.93 -44.81
Branch 2 -1.61 13.59 1.74 -16.24 -8.92
Branch 3 -1.44 -20.83 1.61 19.24 -7.21
Branch 4 0.46 5.59 -0.44 -7.46 4.62
Branch 5 -49.02 -4.76 49.77 4.80 -54.47
Branch 6 -34.95 -6.66 35.61 6.16 -58.25
Branch 7 -4.60 -5.88 4.62 4.01 -23.02
Hydro linear OPF
The following example loads and runs the linear optimization for a system that integrates fluid elements into a regular electrical grid.
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'hydro_simple.gridcal')
grid = gce.open_file(fname)
# Run the simulation
opf_driver = gce.OptimalPowerFlowTimeSeriesDriver(grid=grid)
print('Solving...')
opf_driver.run()
print('Gen power\n', opf_driver.results.generator_power)
print('Branch loading\n', opf_driver.results.loading)
print('Reservoir level\n', opf_driver.results.fluid_node_current_level)
Output:
OPF results:
time | p2x_1_gen | pump_1_gen | turbine_1_gen | slack_gen
------------------- | --------- | ---------- | ------------- | ---------
2023-01-01 00:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 01:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 02:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 03:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 04:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 05:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 06:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 07:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 08:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
2023-01-01 09:00:00 | 0.0 | -6.8237821 | 6.0 | 11.823782
time | line1 | line2 | line3 | line4
------------------- | ------ | ----- | --------- | -----
2023-01-01 00:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 01:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 02:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 03:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 04:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 05:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 06:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 07:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 08:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
2023-01-01 09:00:00 | 100.0 | 0.0 | 68.237821 | 40.0
time | f1 | f2 | f3 | f4
------------------- | ---------- | --- | --- | ----------
2023-01-01 00:00:00 | 49.998977 | 0.0 | 0.0 | 50.001022
2023-01-01 01:00:00 | 49.997954 | 0.0 | 0.0 | 50.002046
2023-01-01 02:00:00 | 49.996931 | 0.0 | 0.0 | 50.003068
2023-01-01 03:00:00 | 49.995906 | 0.0 | 0.0 | 50.004093
2023-01-01 04:00:00 | 49.994884 | 0.0 | 0.0 | 50.005116
2023-01-01 05:00:00 | 49.993860 | 0.0 | 0.0 | 50.006139
2023-01-01 06:00:00 | 49.992838 | 0.0 | 0.0 | 50.007162
2023-01-01 07:00:00 | 49.991814 | 0.0 | 0.0 | 50.008185
2023-01-01 08:00:00 | 49.990792 | 0.0 | 0.0 | 50.009208
2023-01-01 09:00:00 | 49.989768 | 0.0 | 0.0 | 50.010231
Short circuit
GridCal has unbalanced short circuit calculations. Now let's run a line-ground short circuit in the third bus of the South island of New Zealand grid example from reference book Computer Analysis of Power Systems by J. Arrillaga and C.P. Arnold
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'South Island of New Zealand.gridcal')
grid = gce.open_file(filename=fname)
# Run a Line-Ground short circuit on the bus at index 2
# Since we do not provide any power flow results, it will run one for us
results = gce.short_circuit(grid, fault_index=2, fault_type=gce.FaultType.LG)
print("Short circuit power: ", results.SCpower[fault_index])
A more elaborated way to run the simulation, controlling all the steps:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'South Island of New Zealand.gridcal')
grid = gce.open_file(filename=fname)
pf_options = gce.PowerFlowOptions()
pf = gce.PowerFlowDriver(grid, pf_options)
pf.run()
fault_index = 2
sc_options = gce.ShortCircuitOptions(bus_index=fault_index,
fault_type=gce.FaultType.LG)
sc = gce.ShortCircuitDriver(grid, options=sc_options,
pf_options=pf_options,
pf_results=pf.results)
sc.run()
print("Short circuit power: ", sc.results.SCpower[fault_index])
Output:
Short circuit power: -217.00 MW - 680.35j MVAr
Sequence voltage, currents and powers are also available.
Continuation power flow
GridCal can run continuation power flows (voltage collapse studies)
import os
from matplotlib import pyplot as plt
import GridCalEngine as gce
plt.style.use('fivethirtyeight')
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'South Island of New Zealand.gridcal')
# open the grid file
main_circuit = gce.open_file(fname)
# Run the continuation power flow with the default options
# Since we do not provide any power flow results, it will run one for us
results = gce.continuation_power_flow(grid=main_circuit,
factor=2.0,
stop_at=gce.CpfStopAt.Full)
# plot the results
fig = plt.figure(figsize=(18, 6))
ax1 = fig.add_subplot(121)
res = results.mdl(gce.ResultTypes.BusActivePower)
res.plot(ax=ax1)
ax2 = fig.add_subplot(122)
res = results.mdl(gce.ResultTypes.BusVoltage)
res.plot(ax=ax2)
plt.tight_layout()
A more elaborated way to run the simulation, controlling all the steps:
import os
from matplotlib import pyplot as plt
import GridCalEngine as gce
plt.style.use('fivethirtyeight')
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'South Island of New Zealand.gridcal')
# open the grid file
main_circuit = gce.FileOpen(fname).open()
# we need to initialize with a power flow solution
pf_options = gce.PowerFlowOptions()
power_flow = gce.PowerFlowDriver(grid=main_circuit, options=pf_options)
power_flow.run()
# declare the CPF options
vc_options = gce.ContinuationPowerFlowOptions(step=0.001,
approximation_order=gce.CpfParametrization.ArcLength,
adapt_step=True,
step_min=0.00001,
step_max=0.2,
error_tol=1e-3,
tol=1e-6,
max_it=20,
stop_at=gce.CpfStopAt.Full,
verbose=False)
# We compose the target direction
factor = 2.0
base_power = power_flow.results.Sbus / main_circuit.Sbase
vc_inputs = gce.ContinuationPowerFlowInput(Sbase=base_power,
Vbase=power_flow.results.voltage,
Starget=base_power * factor)
# declare the CPF driver and run
vc = gce.ContinuationPowerFlowDriver(grid=main_circuit,
options=vc_options,
inputs=vc_inputs,
pf_options=pf_options)
vc.run()
# plot the results
fig = plt.figure(figsize=(18, 6))
ax1 = fig.add_subplot(121)
res = vc.results.mdl(gce.ResultTypes.BusActivePower)
res.plot(ax=ax1)
ax2 = fig.add_subplot(122)
res = vc.results.mdl(gce.ResultTypes.BusVoltage)
res.plot(ax=ax2)
plt.tight_layout()
Contingency analysis
GriCal has contingency simulations, and it features a quite flexible way of defining contingencies. Firs you define a contingency group, and then define individual events that are assigned to that contingency group. The simulation then tries all the contingency groups and apply the events registered in each group:
import os
import GridCalEngine as gce
folder = os.path.join('..', 'Grids_and_profiles', 'grids')
fname = os.path.join(folder, 'IEEE 5 Bus.xlsx')
main_circuit = gce.open_file(fname)
branches = main_circuit.get_branches()
# manually generate the contingencies
for i, br in enumerate(branches):
# add a contingency group
group = gce.ContingencyGroup(name="contingency {}".format(i + 1))
main_circuit.add_contingency_group(group)
# add the branch contingency to the groups, only groups are failed at once
con = gce.Contingency(device_idtag=br.idtag, name=br.name, group=group)
main_circuit.add_contingency(con)
# add a special contingency
group = gce.ContingencyGroup(name="Special contingency")
main_circuit.add_contingency_group(group)
main_circuit.add_contingency(Contingency(device_idtag=branches[3].idtag,
name=branches[3].name, group=group))
main_circuit.add_contingency(Contingency(device_idtag=branches[5].idtag,
name=branches[5].name, group=group))
pf_options = gce.PowerFlowOptions(solver_type=SolverType.NR)
# declare the contingency options
options_ = gce.ContingencyAnalysisOptions(use_provided_flows=False,
Pf=None,
contingency_method=gce.ContingencyMethod.PowerFlow,
# if no power flow options are provided
# a linear power flow is used
pf_options=pf_options)
linear_multiple_contingencies = gce.LinearMultiContingencies(grid=main_circuit)
simulation = gce.ContingencyAnalysisDriver(grid=main_circuit,
options=options_,
linear_multiple_contingencies=linear_multiple_contingencies)
simulation.run()
# print results
df = simulation.results.mdl(ResultTypes.BranchActivePowerFrom).to_df()
print("Contingency flows:\n", df)
Output:
Contingency flows:
Branch 0-1 Branch 0-3 Branch 0-4 Branch 1-2 Branch 2-3 Branch 3-4
# contingency 1 0.000000 322.256814 -112.256814 -300.000000 -277.616985 -350.438026
# contingency 2 314.174885 0.000000 -104.174887 11.387545 34.758624 -358.359122
# contingency 3 180.382705 29.617295 0.000000 -120.547317 -97.293581 -460.040537
# contingency 4 303.046401 157.540574 -250.586975 0.000000 23.490000 -214.130663
# contingency 5 278.818887 170.710914 -239.529801 -23.378976 0.000000 -225.076976
# contingency 6 323.104522 352.002620 -465.107139 20.157096 43.521763 0.000000
# Special contingency 303.046401 372.060738 -465.107139 0.000000 23.490000 0.000000
This simulation can also be done for time series.
State estimation
Now lets program the example from the state estimation reference book State Estimation in Electric Power Systems by A. Monticelli.
import GridCalEngine as gce
m_circuit = gce.MultiCircuit()
b1 = gce.Bus('B1', is_slack=True)
b2 = gce.Bus('B2')
b3 = gce.Bus('B3')
br1 = gce.Line(b1, b2, name='Br1', r=0.01, x=0.03, rate=100.0)
br2 = gce.Line(b1, b3, name='Br2', r=0.02, x=0.05, rate=100.0)
br3 = gce.Line(b2, b3, name='Br3', r=0.03, x=0.08, rate=100.0)
# add measurements
m_circuit.add_pf_measurement(PfMeasurement(0.888, 0.008, br1))
m_circuit.add_pf_measurement(PfMeasurement(1.173, 0.008, br2))
m_circuit.add_qf_measurement(QfMeasurement(0.568, 0.008, br1))
m_circuit.add_qf_measurement(QfMeasurement(0.663, 0.008, br2))
m_circuit.add_pi_measurement(PiMeasurement(-0.501, 0.01, b2))
m_circuit.add_qi_measurement(QiMeasurement(-0.286, 0.01, b2))
m_circuit.add_vm_measurement(VmMeasurement(1.006, 0.004, b1))
m_circuit.add_vm_measurement(VmMeasurement(0.968, 0.004, b2))
m_circuit.add_bus(b1)
m_circuit.add_bus(b2)
m_circuit.add_bus(b3)
m_circuit.add_line(br1)
m_circuit.add_line(br2)
m_circuit.add_line(br3)
# Declare the simulation driver and run
se = gce.StateEstimation(circuit=m_circuit)
se.run()
print(se.results.get_bus_df())
print(se.results.get_branch_df())
Output:
Vm Va P Q
B1 0.999629 0.000000 2.064016 1.22644
B2 0.974156 -1.247547 0.000000 0.00000
B3 0.943890 -2.745717 0.000000 0.00000
Pf Qf Pt Qt loading
Br1 89.299199 55.882169 0.0 0.0 55.882169
Br2 117.102446 66.761871 0.0 0.0 66.761871
Br3 38.591163 22.775597 0.0 0.0 22.775597
Export the results
A simple function is available to export the results of a driver.
import GridCalEngine as gce
fname = os.path.join("data", "grids", "IEEE39_1W.gridcal")
grid = gce.open_file(fname)
# create the driver
pf_driver = gce.PowerFlowTimeSeriesDriver(grid=grid,
options=gce.PowerFlowOptions(),
time_indices=grid.get_all_time_indices())
# run
pf_driver.run()
# Save the driver results in a zip file with CSV files inside
gce.export_drivers(drivers_list=[pf_driver], file_name="IEEE39_1W_results.zip")
You could save many drivers in the same zip file passing then into the list drivers_list.
Also there is a function to save from the results objects themselves:
import GridCalEngine as gce
fname = os.path.join("data", "grids", "IEEE39_1W.gridcal")
grid = gce.open_file(fname)
# run with the API shortcut functions
pf_results = gce.power_flow(grid)
pf_ts_results = gce.power_flow_ts(grid)
# Save the driver results in a zip file with CSV files inside
gce.export_results(results_list=[pf_results, pf_ts_results], file_name="IEEE39_1W_results.zip")
Client - Server operation
To use the gridcal server, you need to install the GridCalServer python package. Once this is done,
the gridcalservercommand will be available on the system.
To launch the server, simply type gridcalserver. This will launch a GridCal server on the machine,
on port 8000. This is https://localhost:8000
An example on how to send a grid from a script to the server:
import os
import asyncio
import GridCalEngine as gce
# path to your file
fname = os.path.join('..', '..', '..', 'Grids_and_profiles', 'grids', "IEEE57.gridcal")
# read gridcal file
grid_ = gce.open_file(fname)
# define instruction for the server
instruction = gce.RemoteInstruction(operation=gce.SimulationTypes.NoSim)
# generate json to send
model_data = gce.gather_model_as_jsons_for_communication(circuit=grid_, instruction=instruction)
# get the sever certificate
gce.get_certificate(base_url="https://localhost:8000",
certificate_path=gce.get_certificate_path(),
pwd="")
# send json
reply_from_server = asyncio.get_event_loop().run_until_complete(
gce.send_json_data(json_data=model_data,
endpoint_url="https://localhost:8000/upload",
certificate=gce.get_certificate_path())
)
print(reply_from_server)
Tests
GridCal uses pytest for automatic software testing.
If you make changes to GridCal that you plan to submit, first make sure that all
tests are still passing. You can do this locally with pytest.
If you have added new functionality, you should also add a new function that tests this
functionality. pytest automatically detects all functions in the src/tests folder
that start with test_ and are located in a file that also starts with test_ as
relevant test cases. Unit test (for pytest) are included in src/tests. As defined in pytest.ini, all
files matching test_*.py are executed by running pytest.
Files matching *_test.py are not executed; they were not formatted specifically for
pytest but were mostly done for manual testing and documentation purposes.
Additional tests should be developped for each new and existing feature. pytest
should be run before each commit to prevent easily detectable bugs.
Bug reporting
You have found a bug in GridCal or have a suggestion for a new functionality? Then get in touch with us by opening up an issue on the issue board to discuss possible new developments with the community and the maintainers.
Contributing
All contributions to the GridCal repository are made through pull requests to the
devel branch. You can either submit a pull request from the develop branch of your
fork or create a special feature branch that you keep the changes on. A feature branch
is the way to go if you have multiple issues that you are working on in parallel and
want to submit with seperate pull requests. If you only have small, one-time changes
to submit, you can also use the devel branch to submit your pull request.
However, it is best to discuss your contribution before the pull request is ready to be officially submitted. We only accept high quality contributions that align with the project design. Those are heavily reviewed, and you may expect joint work with us if your proposal is deemed good enough.
An easier alternative to contribute is to use the GridCal objects and functions to produce your contribution in a script-like fashion. Again, if that meets the functional and quality standards that we impose, we'll take care of the integration.
All contributions must come with testing.
Contact
- Join the Discord GridCal channel for a friendly chat, or quick question.
- Submit questions or comments to our form.
- Submit bugs or requests in the Issues section.
- Simply email santiago@gridcal.org
License
GridCal is licensed under the Mozilla Public License 2.0 (MPLv2)
In practical terms this means that:
- You can use GridCal for commercial work.
- You can sell commercial services based on GridCal.
- If you distrubute GridCal, you must distribute GridCal's source code as well. That is always achieved in practice with python code.
- GridCal license does not propagate, even if you use GridCal or pieces of it in your code. However, you must retain the individual files licensing.
Nonetheless, read the license carefully.
Disclaimer
All trademarks mentioned in the documentation or the source code belong to their respective owners.