Global Energy Monitor - Global Oil Infrastructure Tracker (GOIT)

Oil & Gas Refining & Marketing - the map of how crude and NGL actually move between producing basins and the coasts. The Global Oil Infrastructure Tracker is Global Energy Monitor's asset-level census of liquids transmission pipelines. At the June 2026 reference release it holds 1,634 crude oil pipeline projects totalling roughly 487,000 kilometres, plus 292 natural gas liquids projects adding about 11,100 more - and it carries each one as an individual pipeline phase rather than a whole system, which is the difference between a map and a model.

Every phase sits somewhere on an eight-value lifecycle scale - Proposed, Construction, Shelved, Cancelled, Operating, Idle, Mothballed, Retired - beside its length, country and region, start year, oil-versus-NGL type, and an owner chain resolved from direct operator up to ultimate parent. A second layer ties each project to a wiki page documenting background, financing, environmental impacts and the opposition campaigns that routinely decide whether pipe gets laid.

Inclusion is deliberate rather than accretive: transmission lines longer than 100 kilometres, with shorter lines admitted when newly proposed or under construction, and distribution and gathering lines kept out entirely. That makes the tracker a trunk-network census - the layer where takeaway capacity and stranded-asset questions live.

Get a sample of this dataset cut to the corridors, owners or statuses you care about.

The tracker publishes two shapes: aggregate kilometre tables and the asset-level records underneath them. Both matter, so here is the June 2026 release's verified top line followed by the record layout one row per pipeline phase:

# GOIT at the June 2026 reference release - verified aggregate figures
oil pipeline projects : 1,634       ~487,000 km of line
ngl pipeline projects :   292       ~11,100 km of line
status vocabulary     : Proposed, Construction, Shelved, Cancelled,
                        Operating, Idle, Mothballed, Retired
built-in summary cuts : kilometres by Country / Owner / Region /
                        Start Year & Type - oil and NGL kept separate

# one row per pipeline phase - record layout as delivered
project_name_and_phase : <pipeline system, phase-level identifier>
status                 : Proposed | Construction | Shelved | Cancelled
                         | Operating | Idle | Mothballed | Retired
length_km              : <traced or route-approximated kilometres>
country                : <country>      region: <regional grouping>
start_year             : <year>         type  : Oil | NGL
owner_parent_chain     : <direct owner> -> <ultimate parent> (<pct>; >5% interests reported)
route_geometry         : <line geometry - GeoJSON / GeoPackage>

Read the aggregate block as the market structure: roughly 487,000 kilometres of crude line against about 11,100 of NGL tells you immediately that this is a crude-movement story with a thin-goods sidebar. Read the record block as the analytical unit: because rows are phases, a single system can appear twice - one phase Operating, the next still Proposed - and only phase-level rows catch that. The status enum is doing quiet work too: Shelved and Cancelled are distinct values, not missing data, which is precisely what makes build-out failure analysable. Request a sample and real rows replace every placeholder above, pinned to the projects you name.

Eight fields define each pipeline phase. They divide into identification (phase name), state (status on the fixed eight-value scale), measurement (length, approximated from at least a low-resolution route where research cannot establish it), geography (country and region), timing (start year), commodity (the oil-versus-NGL enum), control (an owner chain resolved to ultimate parents with percentage interests above a 5% reporting threshold), and route geometry traced in QGIS from public-domain route maps or approximated from endpoints.

Everything else folds under additional fields on request rather than being promised blind: the wiki-page narrative attached to each project, a supplemental file of detailed ownership-chain information, and kilometre rollups derived across any combination of the summary dimensions. One negative finding is worth stating plainly - capacity and throughput are outside the published scope, so nobody should go looking for barrels-per-day columns that do not exist. Definitions below come from the published methodology and FAQ and are pinned against real rows during sample preparation, so the dictionary you buy is the dictionary you tested.

Geography - global by intent. The tracker aims to be comprehensive for transmission lines longer than 100 kilometres carrying oil or NGL wherever they are, with the deepest detail on operating and in-development projects and distribution and gathering lines excluded by design. Corridor questions resolve without stitching together national maps.

Temporal - dated editions that pin the entire inventory at a moment, with the recommended citation referencing the June 2026 release. Because the start-year attribute reaches back decades, the operating fleet itself carries history: build-out eras, regional booms and the age distribution of today's lines are all computable from the records.

Granularity - asset level, one row per pipeline phase, rolling up cleanly into the built-in kilometre summaries by country, owner, region and start-year-and-type with oil and NGL kept separate. Against Datadory's wider catalog - 1,744 datasets averaging 7.81 on the quality rubric - this record scores 8/10, lifted by global asset-level depth and route geometry, held back by nothing worse than inferred column labels.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your team already works in and set the cadence yourself - a one-time pull of the full inventory lands as cleanly as a recurring feed of the newest edition. Geometry ships in the standard geospatial formats beside the tabular records, so GIS teams and warehouse teams take the same delivery apart differently. Every delivery travels with the field dictionary and validation rows, and the schema holds steady between refreshes. Say which corridors matter when you get a sample of this dataset; the sample goes out first and the schema in the sample is the schema you ship against.

• Midstream market mapping - screen ~1,900 phases by country, owner, status and type to size who controls which corridors; part of the shelf serving market sizing.
• Takeaway and build-out analysis - Construction and Proposed phases against the operating fleet show where export and refinery-feed capacity is heading before flow statistics ever register it.
• Stranded-asset and transition-risk screening - Shelved, Cancelled, Idle and Mothballed are explicit status values, so the dormant tail is queryable rather than archaeological; see ESG emissions analysis workflows.
• Corridor and routing analysis - route geometry loads into GIS for proximity and exposure work; patterns in supply chain mapping apply directly.
• Ownership and counterparty research - direct-owner-to-parent chains convert a list of pipes into a list of who really holds them.
• Citation-grade journalism and research - per-project financing and opposition documentation gives footnotes that survive review; see citation grade research.

Ranked by relevance in Datadory's persona tagging:

1. Market Researchers & Consultants (relevance 3): infrastructure sizing and owner-share analysis for market studies - market researchers use cases.
2. Investors & Quant Researchers (relevance 2): construction-stage pipeline inventories as forward indicators for midstream and downstream capital cycles - investors quants use cases.
3. Journalists, Academics & Students (relevance 3): documented project histories behind every headline pipeline story - journalists academics use cases.
4. Data Scientists & ML Engineers (relevance 2): clean enums and dates that join against other energy tables without hand-matching - data scientists use cases.
5. Developers & Builders (relevance 2): flat tables plus standard geospatial formats that drop into GIS stacks and warehouses alike - developers builders use cases.
6. Competitive Intelligence & Product Teams (relevance 2): status-change watching across owners and countries - competitive intel product teams use cases.

Three things worth knowing upfront.

First, this is a length census, not a volume ledger. Kilometres by status, owner and geography - no throughput, no capacity. Where measured volumes exist they come from regulators instead: the Canada Energy Regulator's pipeline throughput and capacity data is the flow-side complement for Canadian trunks, and pairing the two is the standard move when a corridor question needs both the pipe and what moves through it.

Second, routes vary in resolution. Geometry ranges from straight-line approximation to high-resolution QGIS traces, so treat computed distances as bounded by the weakest route in your selection - we flag resolution when we cut samples.

Third, definitions are pinned, not assumed. Column labels come from the published methodology and summary-table dimensions, so anything ambiguous gets resolved against real rows before delivery. The head-to-head with PPAC India's ready reckoner shows where an asset census and a national monthly diverge, and the oil gas refining marketing data hub ranks the rest of the slice.