After you can run SMIB stability and CCT work (Lab 5), the next plant fidelity step is a network simulator. This course uses ANDES Cui et al., 2021ANDES developers, 2026 as that reference: load a case, run power flow and time-domain simulation, then apply the same SciML evaluation ideas (scenarios, operational labels, CCT brackets, manifests) you already used on SMIB.
ANDES is open-source (GPL v3). We summarize what you need here and point to the official docs for full tutorials and derivations — we do not republish the ANDES manual in this book.
Install (short)¶
Use the course ANDES environment from the repository root:
conda env create -f envs/course_andes.yml
conda activate sciml_ps_andesCourse pin: andes==2.0.0 in envs/course_andes.yml. After create/activate, run andes prep once if prompted. Do not mix ANDES 1.x and 2.x in the same lab narrative.
Optional notebook (Lab 5 extension): lab05_network_cct_andes.ipynb.
Details: ANDES installation.
Concepts you will touch (summaries)¶
| Idea in this course | Meaning for SciML labs | Details |
|---|---|---|
| Power flow (PFlow) | Find a steady operating point before dynamics | Power flow |
| Time-domain (TDS) | Simulate faults and clearing; get trajectories for labels / CCT | Time-domain simulation |
| Case file | Documented network + machines + events | Data and formats |
| Plots / channels | Extract angles, speeds, etc. for your metrics | Plotting |
Minimal workflow in words: load case → PFlow.run() → configure fault/clearing → TDS.run() → read trajectories → apply your stability rule and CCT search. Full API options stay in the tutorials above and the reference.
Preferred class network case: Kundur¶
| Item | Choice |
|---|---|
| Case | andes.get_case("kundur/kundur_full.xlsx") |
| Why | Two-area multimachine plant; matches course stage C after SMIB |
| Official walkthrough | Critical Clearing Time Mapping |
That gallery sweeps fault clearing times across buses, classifies stable / unstable / non-convergent outcomes, and maps CCT. Use it to learn the ANDES CCT workflow. Your course extension (when assigned) still reports SciML metrics: signed CCT error, false-safe rate, and comparison to a learned surrogate — same spirit as Lab 5 on SMIB.
Self-study path (ANDES), then return here¶
Then return to Lab 5 / this network path. Tutorials 07+ (eigenvalue, sweeps, contingency, …), the Modeling guide, and Verification are further practice, not assessed core.
Other gallery examples (optional)¶
Full index: ANDES Gallery.
| Gallery | How it connects to this book |
|---|---|
| PINN surrogate (SMIB) | Further practice after Labs 1–2 — does not replace them |
| Low-inertia frequency (IEEE 14) | Project / Advanced (frequency / RoCoF), not the CCT spine |
| Forced oscillation (WECC 179) | Large-system elective / project |
| RL oscillation damping (SMIB) | Outside the core three pillars; curiosity only |
Honesty¶
Network simulation labels and CCT values are operational under a stated criterion and model set. They are not automatic certificates of real-world security. Always record ANDES version, case name, fault/clearing definitions, and the stability rule in your manifest.
Citation¶
When you use ANDES in a report, cite Cui et al. (2021). Tool links: documentation, GitHub. License: GPL v3.
- Cui, H., Li, F., & Tomsovic, K. (2021). Hybrid Symbolic-Numeric Framework for Power System Modeling and Analysis. IEEE Transactions on Power Systems, 36(2), 1373–1384. 10.1109/TPWRS.2020.3017019
- ANDES developers. (n.d.). ANDES documentation. Retrieved October 14, 2026, from https://docs.andes.app/en/stable/