Run Demo Projects
Gridalyn projects are reproducible workspaces. Each project owns a
project.yaml, a workflow.yaml, input references, generated outputs,
manifests, reports, and validation checks. The same commands work for compact
demos and larger studies.
The public documentation should teach the project pattern, not anchor the platform to one unpublished study. Use the small demos first, then move to the larger flexibility workflow only when you want an end-to-end operations example.
Recommended Order
| Project | Why run it first |
|---|---|
projects/minimal_grid_project |
Fastest smoke test for project contracts, reports, and figures. |
projects/ieee_33_bus_demo |
Familiar benchmark feeder with planning-style outputs. |
projects/synthetic_geojson_feeder |
Shows the GeoJSON-to-network generation path. |
projects/prosumer_battery_market |
Demonstrates prosumer assets, forecasts, and market clearing. |
projects/der_voltage_optimization |
Demonstrates CVXPY plus pandapower verification. |
projects/rl_voltage_control_lightsim |
Demonstrates a Gridalyn voltage-control environment backed by LightSim2Grid. |
projects/dr_agent_interaction |
Runs a day of OpenADR 3.1.0 demand-response events through the dr_program protocol over a lossy channel, in seconds. |
projects/ev_hosting_flex |
EV hosting capacity and flexibility on a Québec all-electric feeder. |
projects/admm_thermal_consensus |
Distributed ADMM coordination of cold-climate electric heating, on a 500 kVA LV feeder. |
Common Commands
Validate a project contract. --check-artifacts also requires the declared
reports and figures to exist; those are git-ignored, so drop the flag on a
fresh checkout and add it back after the run below, or the command exits
non-zero on the missing artifacts:
Inspect its planned stages:
Run it (per-stage progress streams to the terminal):
While iterating on one stage, run only that stage and its dependencies:
Check outputs and required artifacts:
Run the publication-style verification ladder:
Verify all governed demos:
verify-all covers every project in the table above and only passes a project
whose outputs already exist, so it exits non-zero until each has been run. The
six compact demos take about 80 s in total; a fully green verify-all also
needs the two long-running studies (ev_hosting_flex,
admm_thermal_consensus).
Check pinned result metrics against the project's regression baseline (every
demo ships one under baselines/results_baseline.json):
If a command fails with an import error, uv run gridalyn doctor shows which
optional capabilities are installed; domain CLIs also print the exact
pip install "gridalyn[<extra>]" command they need.
Project Outputs
Most projects follow this output layout:
projects/<name>/outputs/data/
projects/<name>/outputs/json/
projects/<name>/outputs/figures/
projects/<name>/outputs/reports/
projects/<name>/outputs/manifests/
projects/<name>/outputs/operations/
Not every project uses every folder. Small demos may only write reports and one
figure; operations demos usually write outputs/operations/ as well.
Larger Research Workflow
The EV Hosting Flexibility project is a comprehensive research arc. It starts
from project-declared inputs, builds a synthetic topology cache, generates
stochastic building and EV profiles, computes dynamic thermal limits, clears
locational flexibility contracts, validates selected actions with pandapower,
writes figures and materializes operation artifacts, and closes with
build_study_reports, which assembles the study-level report from every
stage's report and the run manifest.
Run it only when you need the full stack:
uv run gridalyn project run projects/ev_hosting_flex
uv run gridalyn project verify projects/ev_hosting_flex
Its generated artifacts live under:
It is the repo's flagship study, and it is long-running: a full source
regeneration takes roughly six hours across 23 stages (operator-verified,
receipt-pinned), while warm runs against an existing cache take minutes. See
verification.md for the staged protocol.
Synthetic Network Inputs
Projects that build synthetic networks can start from building-footprint GeoJSON. The default tutorial path is:
To prepare footprints from OpenStreetMap or Microsoft Building Footprints, see Synthetic Networks From GeoJSON.