Security Analysis (doc in progress)
The documentation of this section is in progress. It is rather incomplete for the moment, and only expose the most basic features.
If you are interested in collaborating to improve this section, let us know.
Warning
This function might give wrong result for lightsim2grid version 0.5.5 were they were a bug : when some contingencies made the grid non connex, it made all the other contingencies diverge. This bug has been fixed in version 0.6.0 and this is why we do not recommend to use this feature with lightsim2grid version < 0.6.0 !
Goal
This class aims to make faster (and easier) the computations of a security analysis (which is the results of some powerflow after the disconnection of one or more powerlines)
This function is much (much) faster than its pure grid2op counterpart. For example, on the case 118, to simulate all n-1 contingencies you can expect a ~20x speed ups compared to using the grid2op obs.simulate(…, time_step=0) while obtaining the exact same results (see section Benchmarks)
It can be used as:
import grid2op
from lightsim2grid import SecurityAnalysis
from lightsim2grid import LightSimBackend
env_name = ...
env = grid2op.make(env_name, backend=LightSimBackend())
security_analysis = SecurityAnalysis(env)
security_analysis.add_multiple_contingencies(...) # or security_analysis.add_single_contingency(...)
res_p, res_a, res_v = security_analysis.get_flows()
# in this results, then
# res_p[row_id] will be the active power flows (origin side), on all powerlines corresponding to the `row_id` contingency.
# res_a[row_id] will be the current flows, on all powerlines corresponding to step "row_id"
# res_v[row_id] will be the complex voltage, on all bus of the grid corresponding to the `row_id` contingency.
# you can retrieve which contingency is id'ed `row_id` with `security_analysis.contingency_order[row_id]`
For now this relies on grid2op, but we could imagine a version of this class that can read to / from other data sources.
Note
A more advanced usage is given in the examples\security_analysis.py file from the lightsim2grid package.
Benchmarks
Here are some benchmarks made with:
system: Linux 5.11.0-40-generic
OS: ubuntu 20.04
processor: Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
python version: 3.8.10.final.0 (64 bit)
numpy version: 1.18.5
pandas version: 1.1.4
pandapower version: 2.7.0
lightsim2grid version: 0.6.0
grid2op version: 1.6.4
Where lightsim2grid has been installed from source with all optimization enabled.
This benchmark is available by running, from the root of the lightsim2grid repository:
cd examples
python3 security_analysis.py
For this setting the outputs are:
For environment: l2rpn_neurips_2020_track2_small (177 n-1 simulated)
Total time spent in "computer" to solve everything: 18.5ms (9573 pf / s), 0.10 ms / pf)
- time to compute the coefficients to simulate line disconnection: 0.04ms
- time to pre process Ybus: 2.50ms
- time to perform powerflows: 15.84ms (11172 pf / s, 0.09 ms / pf)
In addition, it took 0.83 ms to retrieve the current from the complex voltages (in total 9160.4 pf /s, 0.11 ms / pf)
Comparison with raw grid2op timings
It took grid2op (with lightsim2grid, using obs.simulate): 0.42s to perform the same computation
This is a 21.6 speed up from SecurityAnalysis over raw grid2op (using obs.simulate and lightsim2grid)
It took grid2op (with pandapower, using obs.simulate): 6.39s to perform the same computation
This is a 330.9 speed up from SecurityAnalysis over raw grid2op (using obs.simulate and pandapower)
All results match !
In this case then, the SecurityAnalysis module is more than 22 times faster than raw grid2op ( with obs.simulate as a way to compute the outcome of a contingency)