NORA: Network Operations and Realtime Analysis

An open model of the North Sea gas network that anyone can improve: the fields, platforms, processing plants, terminals and pipelines that carry Norwegian, British, Dutch and Danish gas to Europe's markets.

Open NORA · contribute

What it does

• Four levels of detail. Level 1 is the network overview: plan status, exit points, pressures, active alarms and units. Level 2 is one operating unit on the Map or in the Flow view. Level 3 is an asset: what flows in and out, its readings and state. Level 4 shows the recent readings behind the values, missing readings, source details, the network data row and the modelled flow.
• Map. The network at its true coordinates on a real map (MapLibre with free OpenFreeMap tiles). Click any asset for its details and a link to its data row.
• Flow. The whole Gassco network as a gas flow overview drawn with pyvis: fields on the left, terminals and markets on the right, on a dotted paper canvas, with flows written on the lines.
• Analysis. networkx computes how much gas the network can deliver to each market (max flow) and how much is lost if any single asset trips (N-1). The Load and Risk modes show both, on the map and in the flow overview.
• Decision support. When a plan is short, the overview names the likely active event, alternative exits with spare booked capacity, and the basis and uncertainty of the suggestion. Synthetic data is labelled; verify advice against the live system.
• Live replay. Time series of flows, from a synthetic week to real daily flows from Energinet, play back as if they were arriving live: lines fill and empty, dots move with the gas, alarms fire when a flow drops. Add a real dataset with one command; nora live serves it as a real Server-Sent Events feed.
• One-file sharing. network.html is a single file you can email or host anywhere; the flow overview even works offline. The whole site is static and rebuilt on every merge.
• Any screen in a control room. One page fits a video wall, an operator desk and a tablet. It picks a display profile and one UI scale from the window: large and dark on a wall, with the pointer hidden when idle. Settings, Display can change both, and so can the link: ?display=wall|desk|compact, ?scale=150, and ?rotate=60 to cycle Overview, Map and Flow on an unattended screen.

Quick start

git clone https://github.com/Rayha33/nora-network.git && cd nora-network
make setup        # finds Python 3.11+ (macOS's own python3 is too old) and creates .venv
make check        # lint, validate, Python and JS tests, build: what CI runs
make serve        # builds the site and opens http://127.0.0.1:8000

You need Python 3.11 or newer and git. Node 20+ is optional: with it, make check also runs the JavaScript tests. make help lists every command.

Or press the Codespaces badge above for a ready-made workspace in the browser.

.venv/bin/nora stats                    # what the network delivers, and its weak points
.venv/bin/nora live                     # replay a stream as a live feed
.venv/bin/nora live --now --stream synthetic-week  # a showcase: the stream re-timed to now, looping
.venv/bin/nora live --ingest            # take live readings pushed as JSON (docs/live-data.md)
.venv/bin/nora stream new my-data --from flows.csv   # add your own dataset

Add your own data

you have	do this
a correction or a source	edit data/nodes.csv or data/pipelines.csv in the browser (how)
a time series of flows	nora stream new NAME --from file.csv, check the mapping (guide)
a live API	yield frames from a generator and serve them like nora.live.replay (AGENTS.md)

Connecting Origo and GBS

Origo (hourly nominations) and Gassco Booking, GBS (booked capacity), have no public API, so NORA's connectors are built against documented mocks. Start them with python scripts/mock_origo.py and python scripts/mock_gbs.py, set NORA_ORIGO_URL/NORA_ORIGO_TOKEN and NORA_GBS_URL/NORA_GBS_TOKEN, then nora integrations fetch origo --to shipper-mock writes a stream the app replays. The mocks' JSON is NORA's own, not Origo's or Gassco's real format, and their data is labelled synthetic. With real credentials, only the transport and the format change: see docs/integrations.md.

How it fits together

data/*.csv ─► data.py (validate) ─► graph.py (networkx) ─► analysis.py (max flow, N-1)
                                                       │
data/streams/ ─► stream.py (import, synthesise) ───────┼─► render.py ─► network.html (Map: MapLibre, Flow: pyvis)
                                                       └─► site.py ─► index.html + data/*.json
                                    live.py ─► /stream (Server-Sent Events) ─► network.html?live=/stream

Every pull request is checked and built. Merging to main publishes the site at https://factorlab.app/nora-network/. Details for contributors and AI assistants are in CONTRIBUTING.md and AGENTS.md.

What diagnostics show

Level 4 currently reports the selected asset's recent readings, missing readings, dataset and asset source metadata, the source CSV row, and its flow in the network model. It does not yet trace a live SCADA interface or tag through every ingestion and control-room system; those integrations and lineage need to be supplied and verified.

Roadmap

The issues are the roadmap. Directions:

• Realtime. Connect live sources such as the ENTSOG Transparency Platform.
• Integrations. The Origo and Gassco Booking connectors run against mocks; the real systems need their API documentation and credentials, and CSV exports remain an import path (docs/integrations.md). The public Gassco feed reports country aggregates, not terminal-specific measurements.
• Data quality. Sources for every row.
• Scenarios. What-if outages and demand cases run in the browser.
• Operations. Line pack, pressure and alarms closer to a real control room.

About the data

Positions, capacities and lengths are approximate figures compiled from public sources, and many rows still need a citation. Map data is from OpenStreetMap contributors, with tiles by OpenFreeMap. NORA is for learning, exploration and discussion, and is not suitable for operational decisions.

MIT licensed. See LICENSE.