Grid Interconnection Queue Tracker

~8,200 projects are actively seeking grid interconnection in the US alone (1,312 GW generation + ~749 GW storage), with delays of three to seven years common across ISOs/RTOs — and the OBBBA's December 31, 2027 deadline for wind/solar projects to be placed in service to keep federal tax credits makes queue position acutely time-sensitive right now. This tracks real, project-level interconnection queue data — not fabricated aggregate estimates — across all 7 major US ISOs and UK National Grid.

Phase 1 source audit — the most fragmented source landscape in this program, verified live

Interconnection queue data is published per-ISO/RTO/TSO, not centrally — the PRD itself flagged this as the most fragmented of the five actors in this batch. Here's what direct testing found:

Source	Verdict	Basis
LBNL "Queued Up"	✅ Primary source — richer than expected	Confirmed live: not just an aggregate/summary file as the PRD anticipated, but a genuine project-level dataset — 38,203 real rows across all 7 US ISOs (CAISO, ERCOT, ISO-NE, MISO, NYISO, PJM, SPP) plus two non-ISO utility aggregates, real columns (q_id, q_status, q_date, IA_phase_clean, mw_1, etc.) mapping close to 1:1 onto this actor's schema. CC BY 4.0 licensed.
Individual US ISOs (PJM, MISO, ERCOT direct)	❌ Not built for v1	Given LBNL already provides real project-level detail for all 7 ISOs in one comprehensive file, separately scraping each ISO's own bespoke format (Excel/PDF/proprietary, confirmed each ISO publishes differently) wasn't pursued — a documented scope decision based on what Phase 1 found, not a gap. interconnection.fyi (by GridTracker) was found and confirmed real and valuable during research — a free, daily-updated aggregator tracking 43,887 requests with per-project status-change history — but its actual project list loads via a mechanism this build didn't fully reverse-engineer in the time available; flagged here as a genuine, verified v2 candidate, not silently omitted.
UK National Grid ESO	✅ Included	Confirmed live: connecteddata.nationalgrid.co.uk is a real, standard CKAN open-data platform with a documented public API (/api/3/action/package_show) returning clean JSON with all 45 real per-substation CSV download URLs — no HTML scraping needed for discovery, no login, plain HTTP throughout.
EU TSOs	❌ Scoped out, per the PRD's own suggestion	27 member states, no single aggregator — the PRD itself flagged this as a v2 expansion candidate, confirmed reasonable rather than second-guessed.

Two real, honest limitations — not smoothed over

• LBNL is an annual snapshot, not a real-time feed. The "2026 Edition" reflects data as of the end of 2025 — LBNL's own real update cadence, not a shortcut taken here. For faster-moving monitoring, interconnection.fyi's daily updates would be the better source — a genuine reason to revisit it in v2.
• UK data has no queue-entry-date or estimated-in-service-date columns at all. Confirmed live across multiple real CSV files — queueEnteredDate/estimatedInServiceDate are always null for UK records, not fabricated to look complete like the US data. UK's Status field is used for both queueStatus (normalized) and studyStage (verbatim) since the source doesn't separate those two concepts the way LBNL does.

Two real bugs worth knowing about, both caught by live testing

• UK region_summary came back empty. The status string "Accepted not yet Connected" contains the substring "connected", and a naive status-mapping check order classified it as operational (matching the generic "connected" check) before it ever reached the more specific "not yet connected" / "accepted" check — silently filtering out every active UK project. Fixed by reordering the checks; verified live afterward that real active UK projects come through correctly.
• LBNL worked locally, then failed consistently once deployed — the same "worked on my machine, blocked on Apify's cloud" pattern found elsewhere in this program (Instagram, in the Creator Rate-Card actor). Confirmed via the real run log: Crawlee's own block-detection fired on a literal HTTP 403 from Cloudflare, not a passable "Just a moment..." JS challenge. Confirmed further that Apify's default (datacenter-class) proxy pool didn't fix it either — still 403ing across multiple rotated IPs — pointing to Cloudflare blocking datacenter IP ranges broadly, not just Apify specifically. Switching to Apify's RESIDENTIAL proxy group (a genuinely different IP class) fixed it, confirmed live with a real successful run returning real ERCOT project data afterward.

Modes

Mode	Behavior
region_summary	Active projects for a region, capped at 300 (see below), optionally filtered by resourceType
project_lookup	A specific project by projectId within a region
monitor	Same as region_summary, plus statusChangedSinceLastRun — computed by comparing queueStatus + studyStage + estimatedInServiceDate against this actor's own stored history from a prior run (reuses deltaStore.ts, originally built for Actor 26's Tariff Monitor, unmodified — the same module is now used by 3 actors in this program)

Results are capped at 300 rows per run — LBNL's PJM region alone has thousands of active projects; a full unfiltered dump isn't the useful shape for either a dashboard or an n8n alert workflow. Use resourceType to narrow further.

Output schema

{
  "region": "string",
  "projectId": "string",
  "resourceType": "solar | wind | storage | other",
  "capacityMw": "number | null",
  "queueStatus": "active | withdrawn | operational | suspended | null",
  "studyStage": "string | null",
  "queueEnteredDate": "string | null",
  "estimatedInServiceDate": "string | null",
  "statusChangedSinceLastRun": "boolean | null",
  "sourceUrl": "string",
  "checkedAt": "ISO timestamp",
  "status": "success | failed",
  "errorMessage": "string | null"
}

statusChangedSinceLastRun is null outside monitor mode, and false (not true) the very first time a project is ever seen — there's nothing to compare against yet, so "changed" would be a false signal, not an honest one.

Setup note

Playwright/Chrome base image required for LBNL specifically — its webpage sits behind Cloudflare (confirmed live, a plain HTTP GET 403s) even though the actual .xlsx file, once its current URL is discovered, downloads fine over plain HTTP. UK National Grid needs no browser at all.

Pricing

Pay-per-event. apify-actor-start and apify-default-dataset-item (Project Status Returned) are built-in, auto-billed events.

Event	Charged when	Price
Actor Start (apify-actor-start)	Actor run starts (auto-billed)	$0.10
Project Status Returned (apify-default-dataset-item)	a project queue status is returned with study stage, capacity, and estimated in-service date (auto-billed)	$0.006
Status Change Detected (status-change-detected)	monitor mode: a project's queue status changed since the last run (not a withdrawal)	$0.025
Withdrawal Detected (withdrawal-detected)	monitor mode: a project is detected as withdrawn from the queue	$0.03
Market Queue Snapshot (market-queue-snapshot)	a full regional queue snapshot is returned (region_summary/monitor), once per run	$0.01
Project Not Found (project-not-found)	project_lookup mode: the queried project ID isn't found in that region's queue	$0.001

A region-level fetch failure (e.g. source unreachable) pushes a failed record but isn't charged — no configured event matches that case.

n8n integration

• Workflow A (trigger): scheduled run against a developer's or investor's tracked projects/regions, frequency matched to each source's real update cadence (LBNL: effectively annual; UK: check per your own needs).
• Workflow B (processing): alert on statusChangedSinceLastRun: true — stage advancement or withdrawal are both materially important signals for developers and investors.