Energy · Conflict · Recovery
Last updated:
Cite as: Gabriel Collins, Energy Disruption Monitor, Collins Research Portal, https://energy.collinsresearchportal.com/ (accessed [date]).
Licence: Data available under CC BY 4.0: free to reuse with credit. Download the full dataset (CSV).
Energy infrastructure affected by the Russia–Ukraine war, including occupied Ukrainian territory.
Geographic view
525 approved records · 0 with cited coordinates · 525 without a mapped point
Current filters · event records / aggregate reports
| Region | Upstream | Midstream | Downstream | Electricity | Water / desalination | Unclassified | Quantified disruption |
|---|---|---|---|---|---|---|---|
| Russia & Ukraine | 6 / 4 | 207 / 12 | 192 / 1 | 79 / 17 | 0 / 0 | 7 / 0 | 0 |
Counts describe selected historical records, not unique facilities or strikes. Aggregate reports may overlap events. Capacity, tonnage and lost output are not summed; units and scope differ. Missing magnitude stays unknown.
| Date | Facility / location | Segment / vessel | Magnitude / capacity / size | Physical damage | Evidence basis |
|---|---|---|---|---|---|
| Oct 2, 2026 | Midstream | Magnitude not established | Unknown | Single-source | |
| Oct 2, 2026 | Power | Magnitude not established | Unknown | Single-source | |
| Oct 2, 2026 | Unclassified | Magnitude not established | Unknown | Single-source | |
| Sep 28, 2026 | Downstream | Magnitude not established | Fire reported | Single-source | |
| Sep 26, 2026 | Power | Magnitude not established | Partial | Attributed report | |
| Sep 25, 2026 | Downstream | Magnitude not established | Partial | Attributed report | |
| Sep 24, 2026 | Midstream | Magnitude not established | Partial | Attributed report | |
| Sep 22, 2026 | Power | Magnitude not established | Fire reported | Attributed report | |
| Sep 20, 2026 | Downstream | Magnitude not established | Partial | Attributed report | |
| Sep 19, 2026 | Power | Magnitude not established | Unknown | Attributed report | |
| Sep 18, 2026 | Unclassified | Magnitude not established | Unknown | Single-source | |
| Sep 18, 2026 | Midstream | Magnitude not established | Unknown | Attributed report | |
| Sep 17, 2026 | Power | Magnitude not established | Unknown | Single-source | |
| Sep 17, 2026 | Downstream | Magnitude not established | Fire reported | Attributed report | |
| Sep 16, 2026 | Downstream | Magnitude not established | Unknown | Attributed report | |
| Sep 13, 2026 | Downstream | Magnitude not established | Fire reported | Attributed report | |
| Sep 13, 2026 | Downstream | Magnitude not established | Fire reported | Attributed report | |
| Sep 13, 2026 | Power | Magnitude not established | Unknown | Attributed report | |
| Sep 10, 2026 | Upstream | Magnitude not established | Unknown | Attributed report | |
| Sep 8, 2026 | Downstream | Magnitude not established | Unknown | Workbook record | |
| Sep 8, 2026 | Unclassified | Magnitude not established | Unknown | Workbook record | |
| Sep 8, 2026 | Power | Magnitude not established | Unknown | Workbook record | |
| Sep 8, 2026 | Power | Magnitude not established | Unknown | Attributed report | |
| Sep 7, 2026 | Downstream | Magnitude not established | Unknown | Workbook record | |
| Sep 7, 2026 | Downstream | Magnitude not established | Unknown | Workbook record | |
| Sep 6, 2026 | Downstream | Magnitude not established | Unknown | Workbook record | |
| Sep 4, 2026 | Midstream | Magnitude not established | Unknown | Workbook record | |
| Sep 4, 2026 | Midstream | Magnitude not established | Unknown | Workbook record | |
| Sep 3, 2026 | Unclassified | Magnitude not established | Unknown | Workbook record | |
| Sep 3, 2026 | Upstream | Magnitude not established | Fire reported | Attributed report | |
| Sep 1, 2026 | Midstream | Magnitude not established | Unknown | Workbook record | |
| Aug 31, 2026 | Power | Magnitude not established | Unknown | Attributed report | |
| Aug 30, 2026 | Downstream | Asset capacity / size 20 million tonnes/year Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 28, 2026 | Downstream | Asset capacity / size About 15 million tonnes/year Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 26, 2026 | Downstream | Asset capacity / size About 15 million tonnes/year Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 25, 2026 | Downstream | Asset capacity / size More than 5 million tonnes/year Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 25, 2026 | Downstream | Asset capacity / size 7.2 million tonnes processed in 2024 (actual throughput, not nameplate) Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 22, 2026 | Downstream | Asset capacity / size 8.8 million tonnes/year, reported capacity Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 21, 2026 | Downstream | Asset capacity / size 12.6 million tonnes processed in 2024 (actual throughput, not nameplate) Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 20, 2026 | Midstream | Magnitude not established | Unknown | Workbook record | |
| Aug 19, 2026 | Downstream | Magnitude not established | Unknown | Workbook record | |
| Aug 19, 2026 | Downstream | Asset capacity / size 17 million tonnes processed in 2024 (throughput, not nameplate) Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 14, 2026 | Midstream | Asset capacity / size 9 million tonnes/year (three 3-million-tonne units) Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 13, 2026 | Downstream | Asset capacity / size Up to 74 million barrels/year, Ukrainian military figure Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 11, 2026 | Downstream | Magnitude not established | Unknown | Workbook record | |
| Aug 10, 2026 | Midstream | Asset capacity / size About 6 million tonnes/year LPG production (industry sources) Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 10, 2026 | Downstream | Asset capacity / size 17 million tonnes processed in 2024 (throughput, not nameplate) Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 8, 2026 | Downstream | Asset capacity / size 8.5 million tonnes/year Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 8, 2026 | Downstream | Asset capacity / size Over 6 million tonnes/year Capacity / size context; not measured loss. | Unknown | Workbook record | |
| Aug 6, 2026 | Downstream | Asset capacity / size 15 million tonnes/year Capacity / size context; not measured loss. | Unknown | Workbook record |
Compare reported recovery milestones and forecasts. Unknown repair duration stays unknown. Partial restart, full restoration and reconstruction are separate outcomes; elapsed-day values are not engineering predictions.
| Facility | Recovery basis | Reported outlook / milestone | Assessment date |
|---|---|---|---|
| Unknown | No sourced repair or restart estimate established. | Oct 12, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Oct 2, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Oct 2, 2026 | |
| Unknown | No sourced repair or restart estimate was available. | Sep 28, 2026 | |
| Unknown | The Kommersant headline reports restoration, but the original body, timing and coverage were unavailable. | Sep 26, 2026 | |
| Unknown | No sourced repair duration or restart date was provided. | Sep 25, 2026 | |
| Unknown | No sourced repair or restart estimate was provided. | Sep 24, 2026 | |
| Observed restart | The cable line was restored and unit 3 was restarted and synchronized on September 26. | Sep 26, 2026 | |
| Reported estimate | Industry sources expected repairs and restoration to take several weeks. This is not an operator-confirmed timetable. | Sep 21, 2026 | |
| Unknown | Restoration work was ongoing without a sourced completion time. | Sep 19, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 19, 2026 | |
| Unknown | No repair or full-restoration timetable established by the retrieved report. | Sep 18, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 19, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 17, 2026 | |
| Reported estimate | Industry sources estimated at least one month for repairs. Rosneft did not comment; this is not an operator-confirmed timetable. | Sep 16, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 13, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 13, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 13, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 10, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | No repair or full-restoration timetable established by the retrieved report. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 3, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | No sourced repair or restart estimate established. | Sep 1, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: Reuters sources could not estimate restart timing; Sept. 4 Ukrainian General Staff still reported shutdown. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No verified repair or restart date found. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: Reuters Aug. 24: preliminary repair estimate of 1–2 weeks; completion not verified. | Sep 8, 2026 | |
| Unknown | No separately verified repair or restart assessment added during import. Any timing statements remain in the original workbook fields below. | Sep 8, 2026 | |
| Unknown | No separately verified repair or restart assessment added during import. Any timing statements remain in the original workbook fields below. | Sep 8, 2026 | |
| Unknown | No separately verified repair or restart assessment added during import. Any timing statements remain in the original workbook fields below. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: No firm repair timetable | Sep 8, 2026 | |
| Unknown | No separately verified repair or restart assessment added during import. Any timing statements remain in the original workbook fields below. | Sep 8, 2026 | |
| Reported estimate | As reported 2026-08-13: Repairs could take up to six months, according to the governor; forecast, not completed restoration. Follow-up to NEW August 11 strike; join to that event. Governor cited imported equipment and sanctions delaying repairs. Do not add a second strike count. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: Indefinite closure reported August 11; no firm repair timetable | Sep 8, 2026 | |
| Unknown | No separately verified repair or restart assessment added during import. Any timing statements remain in the original workbook fields below. | Sep 8, 2026 | |
| Unknown | No separately verified repair or restart assessment added during import. Any timing statements remain in the original workbook fields below. | Sep 8, 2026 | |
| Unknown | Workbook recovery note: Fire extinguished August 8; operational recovery unknown | Sep 8, 2026 | |
| Unknown | Workbook recovery note: Fire extinguished August 6; restart timing not reported | Sep 8, 2026 |
Ten dated follow-ups are preserved. The Omsk correction and Orsk forecast are also linked to their incident assessments. Market-wide observations are context, not additional strikes or additive capacity losses.
Prepared drafts for Russia & Ukraine. Review, approve & publish, or dismiss. Pending drafts remain private.
Inspect thermal detections near a reported attack, view daily imagery, and record what the evidence supports.
Method references: NASA data availability and API · Detection limitations · NASA imagery service. This comparison method is an analytical workflow, not a NASA attack-detection product.
Query regional thermal observations, or import a FIRMS CSV. The original CSV and acquisition metadata are retained. This does not automatically confirm an attack or assign damage severity.
Filed under Russia & Ukraine; check that the file covers this region.
Maximum 2 MB. Required columns: latitude, longitude, acq_date, acq_time. Sensor and confidence metadata are preserved when supplied.
Check whether observations overlap the reported place and time. Regional hotspot counts alone cannot match a facility.
Compare with pre-event heat patterns. Routine flares, industrial heat and vegetation fires can produce detections.
Check cloud and overpass coverage. No hotspot does not exclude an attack or damage without sustained fire.
Record the conclusion with its evidence and limitations in the incident assessment. Neither fire radiative power nor pixel count estimates lost capacity or repair time.
VIIRS: approximately 375 m pixels. Global near-real-time products are generally available within 3 hours of observation. Observation latency is not incident-detection latency. NASA methodology.
Shared source register for all regions, covering discovery and verification. Connection labels describe the implemented access route; collection success is shown separately in the run log. Syndication and translated editions do not create independent evidence.
Regional news discovery
Europe, Latin America, US/Canada, Africa, Asia and Middle East feeds provide a second discovery route when GDELT is unavailable. Asia and Europe feeds need geographic review; feed availability does not establish complete energy coverage.
News discovery
Finds articles across machine-translated reporting. Titles, publication metadata and original links enter the review queue. A news hit is not an incident or independent confirmation.
Local reporting
Original-language headline collection. Check dates and original claim provenance before adding an incident.
Russian state wire
Official and regional claims. State affiliation is material; duplicated wire stories are one source lineage.
Russian reporting
Useful operational and regional reporting. Public pages and paid wires have different access and reuse terms.
Russian reporting
Facility shutdown, production and repair reporting. Preserve anonymous-source attribution; do not equate published claims with inspection.
Iranian oil ministry news
Oil, gas, refining and reconstruction reporting. Operator and government perspectives; verify damage and restart statements against other evidence.
Iranian reporting
Original Persian and English headlines. Language editions can repeat the same claim and do not constitute independent corroboration.
Official national news
QatarEnergy and official disruption statements. Extract original statement dates and distinguish lost output from nominal facility capacity.
Operator
Production, storage, transmission and CHP updates. Operator-reported impacts; site identifies CC BY 4.0 for its content.
Energy sector synthesis
Energy Dimension reviews summarize Ukraine attacks, restoration and energy markets using official sources. Weekly synthesis, not a complete repair ledger.
Conflict event cross-check
Russia and Ukraine event coverage with automated energy tags. API requires myACLED/OAuth; publication rights must be checked. Not connected.
Nuclear monitoring
Nuclear safety and external power interruption/restoration updates. Scope is nuclear-related, not the whole energy sector.
Satellite thermal evidence
VIIRS NOAA-20/21 observations can support fire assessment. Flares, cloud, overpass timing and geolocation affect interpretation. No automatic attack confirmation.
Official social source
Grid conditions, outages and restoration. Automated social collection is not connected; AI processing requires appropriate platform and content permissions.
Official social source
Fire and response statements. Fire extinguishment does not establish that the affected facility has restarted.
Official social source
Official channel linked by QNA. Preserve the original statement and distinguish reports from independent physical verification.
Social discovery
Current API uses pay-per-use pricing. Not connected. Curated accounts and licensing must be configured before automated collection.
Commercial outages
Unit outages, de-rates and restarts across energy segments. Russia/Iran coverage and public redistribution rights require verification. Not connected.
Historical research baseline
Published analysis provides historical context. Supplied Ukraine and Gulf workbooks dated 8 September 2026 are now reconciled into the dashboard, with original row references, source URLs and review holds. Imported labels remain workbook claims until separately assessed.
Shared assessment method for all regions.
Collect original news links and multilingual headlines. Preserve source affiliation, language and publication date. Publication time is not automatically incident time.
Combine reports about the same facility and incident. Keep source lineage, event date uncertainty and repeat strikes explicit. Do not sum overlapping facility or portfolio records.
Separate physical damage, operating disruption and cause attribution. A corroborated outage does not necessarily mean a corroborated attacker.
Record repair forecasts with source and assessment date. Revise when new evidence appears. Preserve partial restart, full restoration and reconstruction as different milestones.
Severity categories are qualitative summaries of the cited evidence. Unknown is not zero damage. Estimates need explicit labels; none are generated from asset class alone.
Event records may cover multiple components of one incident; they are not unique-facility totals. Aggregate reports are labeled separately and can overlap individual events. The system does not sum nameplate capacity into “capacity destroyed,” or combine barrels, cubic meters, tonnes and megawatts.
Historical backfill reviewed September 8, 2026, with a scope starting February 24, 2022. Current facility operating condition remains unassessed unless a later, dated report explicitly establishes it.
Daily schedule not yet activated
News collection adds leads to the review queue. Editorial assessments remain separate. A source failure is shown as unavailable; it never becomes “no incidents.” The built-in collection covers RSS and GDELT metadata, not paywalled article bodies.
Dashboard address: energy.collinsresearchportal.com. The dashboard is publicly readable. Editing, pending reports and raw satellite evidence require the editor account.
Substack: link to the hosted dashboard from a post or navigation item. Substack posts do not accept arbitrary custom HTML or iframes; use a dashboard link in a post or navigation item.
NASA API use needs a free MAP_KEY stored on the server. Social automation needs permitted access: X is not connected, and Telegram links are reviewed manually. Telegram API terms.
Ukraine workbook · 480 rows · 431 added · 28 merged · 21 held
Gulf workbook · 121 rows · 113 added · 8 merged · 0 held
The inventory begins on 24 February 2022. It is a selective baseline, not an exhaustive census. Ukraine strike waves and portfolio damage reports are marked as aggregates; they may overlap individual facility reports and should not be summed into a strike count.
The supplied Collins Ukraine and Gulf workbooks were imported on 8 September 2026. All 601 Master Data rows are accounted for: 544 added records, 36 matches to existing records and 21 held for review. Their 12 original review items also remain outside event counts. New rows retain a “Workbook record” evidence label; imported source claims have not been independently reverified. Some years and regions remain less densely covered. Missing records do not imply that no event occurred.
JSON matching the example, up to 20 MB. All records are validated first, then saved in batches of 40. Completed batches remain saved if a later batch fails. Existing IDs are updated and previous versions retained. Workbook provenance is included in exports.