US Gasoline Market
Gulf Coast crack, seasonal stocks, domestic demand and refinery output.
Inspect gasolineREFINED PRODUCTS / WORLD
How much new refining capacity starts up each year to 2030, when it actually produces, what closes, and how much extra diesel that adds, in Conservative, Base and High capacity scenarios. Test it against IEA, OPEC or your own demand growth.
BASE CASE · NET NEW CAPACITY 2026–30
+3.64 mb/dAbsorbs diesel demand growth of about 0.8% a year · ledger reviewed 2026-10-11See the verdict ↓2026–2030 · ledger reviewed 2026-10-11 · live from the ledger database
A shortage here is extra diesel the existing fleet must find by running harder or shifting yield. 1 mb/d is about 3.1 points of global utilization. 2026–2027 include the temporary war-related loss of runs.
Selected: Base Case capacity, OPEC WOO 2026 path. Broadly unchanged versus 2025. Refineries would need to run at 85.5% in 2030 versus 83.8% in 2025 (Conservative capacity 87.5%, High 84.7%).
| Scenario | Gross additions | Closures | Net new capacity | Extra diesel by 2030 | Demand growth absorbed | 2030 required utilization | Verdict |
|---|---|---|---|---|---|---|---|
| Conservative Only projects already operating, commissioning or under construction, with start dates later than the operator targets; minimal creep, no unitemized brownfield, more closures. | +1.92 | -0.75 | +1.17 | +0.60 | 0.38%/yr | 87.5% | Tightening versus 2025 |
| Base Case StorageCurve best estimate of what will actually be built and running by 2030: construction projects at realistic dates, likely planned projects (Dangote train 2, Cauvery Basin, Qilu, Iraq) with slippage, capacity creep and a moderate unitemized brownfield allowance. | +4.09 | -0.45 | +3.64 | +1.13 | 0.81%/yr | 85.5% | Broadly unchanged versus 2025 |
| High Operator schedules largely hold, more planned projects arrive before 2030, a larger unitemized brownfield pipeline and fewer closures; roughly the level of OPEC and BloombergNEF. | +5.07 | -0.35 | +4.72 | +1.45 | 1.00%/yr | 84.7% | Broadly unchanged versus 2025 |
This is a capacity forecast with explicit assumptions, not a price forecast. The Base Case is StorageCurve’s best estimate of what will actually be built and running, including likely planned projects such as Dangote’s second train. Even so, the answer depends more on diesel demand growth than on which capacity scenario is chosen. 2026–2027 also carry a temporary war-related loss of refinery runs, which is shown separately.
Crude distillation nameplate in the year each unit starts, by region, minus closures, for the selected capacity scenario. Dashed lines: published outlooks for comparison (IEA net schedule; OPEC assessed projects less closures, excluding creep). Upgrades that add diesel without distillation capacity do not appear here.
Bars: annual-average diesel added or lost compared with 2025, after commissioning ramps. Line: the selected demand growth on the common base. Where the line sits above the bars, existing refineries must run harder than in 2025.
Runs required at the 2025 average diesel yield, divided by operable capacity. Band: Conservative to High capacity. Dotted: the 2025 level and an editorial practical ceiling four points above it. 2026–2027 include war-related unavailable capacity.
Solid: OPEC observations 2021–2025. Dashed: year-end nameplate in the three capacity scenarios and the runs required to meet the selected diesel demand. The distance between capacity and runs is spare capacity.
The default Base Case remains in the tables below when charts are unavailable.
| Year | Capacity added | Capacity closed | Year-end capacity | Diesel from new capacity | Diesel from upgrades | Diesel lost to closures | Diesel lost to disruption | Biodiesel / renewable diesel | Supply change | Demand change | Supply minus demand | Required utilization % | Spare capacity |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2026 | +0.754 | 0.145 | 104.3 | +0.101 | +0.014 | -0.025 | -1.236 | +0.091 | -1.055 | +0.336 | -1.391 | 87.7 | 12.3 |
| 2027 | +0.828 | 0.075 | 105.0 | +0.312 | +0.021 | -0.045 | -0.644 | +0.133 | -0.223 | +0.676 | -0.899 | 86.3 | 14.1 |
| 2028 | +0.364 | 0.075 | 105.3 | +0.483 | +0.021 | -0.061 | -0.216 | +0.185 | +0.412 | +1.020 | -0.608 | 85.8 | 14.9 |
| 2029 | +1.523 | 0.075 | 106.8 | +0.623 | +0.040 | -0.078 | -0.072 | +0.211 | +0.724 | +1.369 | -0.645 | 85.8 | 15.0 |
| 2030 | +0.620 | 0.075 | 107.3 | +0.952 | +0.047 | -0.094 | 0.000 | +0.229 | +1.134 | +1.721 | -0.587 | 85.5 | 15.5 |
Every project, upgrade and closure, with the month it is assumed to start producing in each capacity scenario. Base Case dates are StorageCurve’s best estimates: later than the operator target where a project has slipped before or lacks financing, but never zero just because funding is not yet confirmed. Projects judged more likely than not to start before 2031 are in the Base Case; less likely ones appear only in High. Diesel uses the operator’s figure where published, otherwise the refinery-type yield.
| Project | Region | Status | Crude capacity kb/d | Conservative | Base Case | High | Ramp | Diesel at full rate | 2030 diesel, selected scenario kb/d | Source |
|---|---|---|---|---|---|---|---|---|---|---|
| Balikpapan RDMP Indonesia | Other Asia | Operational | +100 | Jan 2026 | Jan 2026 | Jan 2026 | 6 mo | 36% · Fuels-oriented / hydrocracker | +31 | 2026-01-12 Operator target Jan 2026 Inaugurated 12 January 2026; capacity 260 to 360 kb/d. Six-month ramp is a StorageCurve assumption. |
| Visakh residue upgrading facility India | India | Operational | Upgrade only | Jan 2026 | Jan 2026 | Jan 2026 | 9 mo | 25 kb/d design (operator or plan) | +21 | 2026-01-05 Operator target Jan 2026 3.55 MMTPA residue hydrocracker, ~93% bottoms conversion; no distillation capacity added. 25 kb/d diesel is a StorageCurve estimate (≈63 kb/d residue feed × ~40% diesel), not an operator figure. |
| Dangote debottlenecking 650→700 Nigeria | Africa | Operational | +50 | Feb 2026 | Feb 2026 | Feb 2026 | 4 mo | 32% · Standard fuels refinery | +14 | 2025-10-27 Operator target Jan 2026 Operator said debottlenecking to 700 kb/d would be complete by early 2026; its website now states 700 kb/d. The 650 kb/d base is already in 2025 capacity. |
| Barmer (HRRL) greenfield India | India | Commissioning | +180 | Jul 2026 | Jul 2026 | Jul 2026 | 12 mo | 22% · Refinery integrated with large petrochemical units | +34 | Undated source Operator target Jul 2026 9 MMTPA; operator page reports dedication on 4 July 2026. Integrated with ~2.4 MMTPA petrochemicals, so a petrochemical-integrated diesel yield is assumed. 12-month ramp is an assumption. |
| Huajin Aramco (Panjin) China | China | Commissioning | +300 | Jan 2027 | Nov 2026 | Oct 2026 | 9 mo | 22% · Refinery integrated with large petrochemical units | +56 | 2026-09-14 Operator target Sep 2026 All 32 main units mechanically complete; in final commissioning in September 2026. 1.65 Mt/y ethylene and 2 Mt/y paraxylene, so low diesel yield. Base start two months after the H2-2026 target. |
| Replan / Revap expansions Brazil | Latin America | Under construction | +44 | Jun 2027 | Dec 2026 | Oct 2026 | 6 mo | 16 kb/d design (operator or plan) | +14 | Undated source Operator target Oct 2026 Petrobras expects 44 kb/d of combined expansion in 2026. 16 kb/d diesel is a StorageCurve allocation inside Petrobras’ plan-level +134 kb/d additional diesel (with RNEST 88 and other projects 30), so the plan total is never exceeded. |
| Gujarat (Koyali) expansion 13.7→18 MMTPA India | India | Under construction | +86 | Jun 2027 | Jan 2027 | Nov 2026 | 6 mo | 32% · Standard fuels refinery | +23 | 2026-08-03 Operator target Nov 2026 89.2% complete at 30 June 2026; IOC targets November 2026. Base adds two months. 1 MMTPA ≈ 20 kb/d. |
| Barauni expansion 6→9 MMTPA India | India | Under construction | +60 | Jun 2027 | Feb 2027 | Dec 2026 | 6 mo | 32% · Standard fuels refinery | +16 | 2026-08-03 Operator target Dec 2026 91.6% complete at 30 June 2026. Recorded as the verified net +3 MMTPA; no unverified old-unit closure is modelled separately. |
| Panipat expansion 15→25 MMTPA India | India | Under construction | +200 | Sep 2027 | Mar 2027 | Dec 2026 | 9 mo | 32% · Standard fuels refinery | +54 | 2026-08-03 Operator target Dec 2026 94% complete at 30 June 2026; target December 2026 after slipping from September 2024. Base adds three months for that delivery record. |
| Numaligarh expansion 3→9 MMTPA India | India | Under construction | +120 | Jan 2028 | Jun 2027 | Jan 2027 | 12 mo | 36% · Fuels-oriented / hydrocracker | +37 | Undated source Operator target Dec 2026 Operator envisaged phased commissioning from December 2025 through FY2026–27; no confirmed CDU start found. Base assumes mid-2027 with a 12-month ramp including the new crude pipeline. |
| Cabinda refinery phase 2 Angola | Africa | Under construction | +30 | Jun 2028 | Sep 2027 | Mar 2027 | 6 mo | 32% · Standard fuels refinery | +8 | 2026-02-09 Phase 1 (30 kb/d) finished September 2025 and is in the base; phase 2 adds 30 kb/d with no published date. |
| Digboi expansion India | India | Under construction | +7 | Mar 2028 | Oct 2027 | Jul 2027 | 3 mo | 32% · Standard fuels refinery | +2 | Undated source Operator target Jul 2027 +0.35 MMTPA, expected July 2027 (secondary compilation of PPAC data). |
| Thai Oil Clean Fuel Project 275→400 Thailand | Other Asia | Under construction | +125 | Jun 2028 | Oct 2027 | Jul 2027 | 12 mo | 36% · Fuels-oriented / hydrocracker | +38 | 2026-04-22 Operator target Jul 2027 Commercial operation Q3 2028; operator hiring notice places CDU/HVU in 2027 and remaining units through 2028. Modelled as a CDU start in late 2027 with a 12-month ramp to Q3–Q4 2028. |
| Bina expansion 7.8→11 MMTPA India | India | Under construction | +64 | Jun 2029 | Nov 2028 | May 2028 | 9 mo | 22% · Refinery integrated with large petrochemical units | +12 | Undated source Operator target May 2028 Expected May 2028 per secondary compilation; BPCL reported 9.6% progress in March 2025, so Base adds six months. Large ethylene cracker attached. |
| Petrobras other diesel-capacity projects (incl. Boaventura integration) Brazil | Latin America | Under construction | Upgrade only | Dec 2029 | Dec 2028 | Jun 2028 | 9 mo | 30 kb/d design (operator or plan) | +26 | Undated source Operator target Jun 2028 Residual of Petrobras’ +134 kb/d additional diesel in its 2026–30 plan after RNEST (88) and Replan/Revap (16). Quality conversions S-500→S-10 (173 kb/d) are excluded because they add no volume. |
| Mongol refinery (Sainshand) Mongolia | Other Asia | Under construction | +30 | Dec 2029 | Dec 2028 | Jun 2028 | 9 mo | 36% · Fuels-oriented / hydrocracker | +9 | 2025-10-14 Operator target Jun 2028 India-funded 1.5 MMTPA (~30 kb/d) greenfield; civil work almost complete in October 2025; operations expected by 2028. Diesel-oriented for the domestic market. |
| Qilu upgrade China | China | Planned | +70 | after 2030 | Dec 2028 | Jul 2028 | 6 mo | 32% · Standard fuels refinery | +19 | 2025-06-17 Operator target Jul 2028 IEA 2028 entry. Sinopec projects usually deliver close to schedule, so Base keeps a late-2028 start; Conservative excludes it because construction status was not verified. |
| Missan refinery Iraq | Middle East | Planned | +70 | after 2030 | Jan 2029 | Jul 2027 | 9 mo | 32% · Standard fuels refinery | +19 | 2025-06-17 Operator target Jul 2026 IEA 2026 entry (70 kb/d). Iraqi refinery projects typically run years late (Karbala took about a decade); Base allows 2.5 years of slippage. |
| Nagapattinam (CPCL) 9 MMTPA India | India | Planned | +180 | after 2030 | Mar 2029 | Dec 2027 | 12 mo | 32% · Standard fuels refinery | +49 | Undated source Operator target Dec 2027 Cauvery Basin replacement refinery, 9 MMTPA. S&P (January 2025) expected end-2027 after a two-year approval delay; CPCL was still seeking approvals. Base assumes March 2029; High holds end-2027. |
| Hassi Messaoud refinery Algeria | Africa | Under construction | +100 | Dec 2030 | Sep 2029 | Dec 2028 | 9 mo | 32% · Standard fuels refinery | +25 | 2025-06-17 Operator target Dec 2028 IEA scheduled 2028. Construction progress was not refreshed in this review, so Base adds nine months. |
| Dung Quat expansion 148→171 Vietnam | Other Asia | Under construction | +23 | Mar 2030 | Sep 2029 | Dec 2028 | 6 mo | 32% · Standard fuels refinery | +6 | 2026-08-26 Operator target Dec 2028 Execution kicked off 21 August 2026 with a 37-month execution period; operator still cites 2028. Base follows the 37-month schedule. |
| Sinopec Tahe expansion 5→8.5 Mt/y China | China | Under construction | +70 | Jun 2030 | Sep 2029 | Mar 2029 | 9 mo | 22% · Refinery integrated with large petrochemical units | +12 | 2025-09-22 Operator target Jun 2029 Construction began September 2025; +3.5 Mt/y crude with chemicals focus; completion 2029. |
| RNEST Train 2 Brazil | Latin America | Under construction | +130 | Jun 2030 | Oct 2029 | Jun 2029 | 9 mo | 88 kb/d design (operator or plan) | +64 | 2026-05-08 Operator target Jun 2029 +130 kb/d crude and +88 kb/d S-10 diesel by 2029 (operator design figures). Train 1’s 2025 revamp is already in the base. |
| Dangote second train (to 1.4 mb/d) Nigeria | Africa | Planned | +750 | after 2030 | Dec 2029 | Dec 2028 | 12 mo | 32% · Standard fuels refinery | +126 | 2025-10-27 Operator target Dec 2028 Announced October 2025 for ~2028 (750 kb/d replica of the existing train); licensor signed, financing open. Base assumes first crude around end-2029, a year after the operator target, reflecting strong 2026 cash flow but a first train that took about seven years. Not in Conservative. |
| Dhi Qar refinery Iraq | Middle East | Planned | +100 | after 2030 | Jun 2030 | Jul 2028 | 12 mo | 32% · Standard fuels refinery | +5 | 2025-06-17 Operator target Jan 2028 IEA 2028 entry (100 kb/d). Base assumes mid-2030; financing delays reported for Iraq’s pipeline. |
| Gulei II (Fujian, Aramco/Sinopec) China | China | Under construction | +320 | after 2030 | Oct 2030 | Jun 2030 | 12 mo | 22% · Refinery integrated with large petrochemical units | +2 | 2024-11-18 Operator target Jun 2030 Groundbreaking November 2024; fully operational by end-2030. Crude-unit start assumed Q4 2030, so it contributes nameplate in 2030 but little 2030 output. |
| Diwaniya refinery Iraq | Middle East | Planned | +70 | after 2030 | after 2030 | Jul 2028 | 9 mo | 32% · Standard fuels refinery | — | 2025-06-17 Operator target Jan 2028 IEA 2028 entry (70 kb/d). High only. |
| Hoima (Uganda) 60 kb/d Uganda | Africa | Planned | +60 | after 2030 | after 2030 | Jun 2029 | 9 mo | 32% · Standard fuels refinery | — | 2026-02-09 Operator target Dec 2028 Construction agreement March 2025; African Energy Chamber cites 2028. Financing and crude-supply timing make a pre-2031 start less likely than not: High only. |
| Hambantota (Sinopec) Sri Lanka | Other Asia | Planned | +200 | after 2030 | after 2030 | Mar 2030 | 12 mo | 32% · Standard fuels refinery | — | 2025-10-15 $3.7bn, up to 200 kb/d; agreement January 2025 but construction not confirmed as started in October 2025. A greenfield of this size needs about four years: High only. |
| BPCL Andhra Pradesh (Ramayapatnam) India | India | Planned | +180 | after 2030 | after 2030 | Jun 2030 | 12 mo | 22% · Refinery integrated with large petrochemical units | — | Undated source 9 MMTPA greenfield with petrochemicals; environmental clearance recommended, construction not started. High only. |
| Valero Benicia United States | OECD Americas | Closed | −145 | May 2026 | May 2026 | May 2026 | — | 37 kb/d lost | -37 | 2026-07-30 Operator target May 2026 Phased idling completed by end-April 2026. Production loss modelled from May; earlier phased run cuts are ignored. |
Unitemized debottlenecking and capacity creep at existing sites, added mid-year. OPEC WOO 2026 identifies 0.4 mb/d beyond its assessed projects for 2026–2030 (Base). Brownfield expansions not itemized in the ledger, added mid-year. BloombergNEF counts 2.9 mb/d of brownfield additions for 2026–2030; the itemized ledger holds about 1.4 mb/d of brownfield and creep, so up to ~1.5 mb/d is unitemized. Base assumes about a third of that gap (0.48 mb/d), High a little over half (0.85 mb/d). Unannounced economic closures from 2027, removed mid-year (OECD and small Chinese plants). OPEC forecasts 0.3 mb/d of closures and BloombergNEF 0.2 mb/d for 2026–2030. Base: 0.3 mb/d plus Benicia; record 2026 margins make near-term closures less likely. No real refinery is named or dated. Extra runs, compared with the 2025 average, from plants that started in 2024–25 and were still ramping (Dangote, Dos Bocas, Yulong second crude unit and others). Their capacity is already in the 2025 base, so only the run-up is counted. Plant-level 2025 runs were not verified.
| Assumption | 2026 | 2027 | 2028 | 2029 | 2030 | Total |
|---|---|---|---|---|---|---|
| Capacity creep · Conservative | 40 | 40 | 40 | 40 | 40 | 200 |
| Capacity creep · Base Case | 80 | 80 | 80 | 80 | 80 | 400 |
| Capacity creep · High | 100 | 100 | 100 | 100 | 100 | 500 |
| Unitemized brownfield · Conservative | 0 | 0 | 0 | 0 | 0 | 0 |
| Unitemized brownfield · Base Case | 0 | 120 | 120 | 120 | 120 | 480 |
| Unitemized brownfield · High | 50 | 200 | 200 | 200 | 200 | 850 |
| Unannounced closures · Conservative | 0 | 150 | 150 | 150 | 150 | 600 |
| Unannounced closures · Base Case | 0 | 75 | 75 | 75 | 75 | 300 |
| Unannounced closures · High | 0 | 50 | 50 | 50 | 50 | 200 |
| Run-up of 2024–25 start-ups (runs vs 2025, all scenarios) | 150 | 200 | 200 | 200 | 200 | — |
| Runs lost: Gulf damage and export constraints (Base disruption) | 1,900 | 500 | 0 | 0 | 0 | — |
| Runs lost: Russian drone damage (Base disruption) | 1,000 | 1,200 | 600 | 200 | 0 | — |
| Runs lost: Crude-importing Asia (feedstock cuts) (Base disruption) | 600 | 100 | 0 | 0 | 0 | — |
Annual-average refinery runs lost compared with 2025, not destroyed nameplate. 2026: Middle East from Kpler (runs 9.9 mb/d in February, 7.3 in August; full return unlikely before Q2 2027); Russia from IEA (runs 3.8 mb/d in June, ~4.0 forecast for the rest of 2026 and 2027, against roughly 5.2–5.4 in 2025); Asia is the residual that makes the 2026 total consistent with IEA’s reported −2.6 mb/d fall in world runs after new-plant ramp-ups. Every 2027+ value is a StorageCurve assumption. Recovery restores runs; it never adds installed capacity.
Net 2026–2030 capacity in each scenario next to published outlooks, followed by a bridge from the Base Case to each outlook. Only differences that can be computed from public figures are attributed; the rest is shown as a residual, because neither OPEC’s nor BloombergNEF’s project list is public.
| Outlook | 2026 | 2027 | 2028 | 2029 | 2030 | 2026–2030 | Definition |
|---|---|---|---|---|---|---|---|
| StorageCurve Conservative | +0.23 | +0.58 | +0.17 | -0.02 | +0.21 | +1.17 | Nameplate in the start year, incl. creep, less closures |
| StorageCurve Base Case | +0.61 | +0.75 | +0.29 | +1.45 | +0.55 | +3.64 | Nameplate in the start year, incl. creep, less closures |
| StorageCurve High | +1.03 | +0.78 | +1.46 | +0.51 | +0.95 | +4.72 | Nameplate in the start year, incl. creep, less closures |
| IEA Oil 2025 2025-06-17 | +1.15 | +0.09 | +0.44 | +0.20 | +0.50 | +2.38 | Net, includes condensate splitters; pre-dates 2026 updates |
| OPEC WOO 2026 2026-06-18 | +0.90 | +0.60 | +0.60 | +1.10 | +1.40 | +4.60 | Assessed projects less closures, calculated from rounded tables; excludes 0.4 mb/d creep |
| BloombergNEF 2026-10-01 | — | — | — | — | — | +4.20 | Net CDU nameplate 2026–2030 (2.9 brownfield + 1.5 greenfield − 0.2 closures); about half already operational or under construction; no public annual split |
| Outlook | Outlook total | Base Case, comparable | Difference | Base closes more capacity | Projects Base dates after 2030 | Unattributed residual |
|---|---|---|---|---|---|---|
| BloombergNEF | +4.20 | +3.64 | +0.56 | +0.25 | +0.51 | -0.20 |
| OPEC WOO 2026 | +4.60 | +3.24 creep of 0.40 removed, as OPEC lists it separately | +1.36 | +0.15 | +0.51 | +0.70 |
Projects that the Base Case dates after 2030 but High includes: Hoima (Uganda) 60 kb/d, Diwaniya refinery, Hambantota (Sinopec), BPCL Andhra Pradesh (Ramayapatnam). Against BloombergNEF, these projects and the Base Case’s higher closure assumption explain the whole gap; the small negative residual means the Base Case carries slightly more creep and brownfield than the bridge needs. Against OPEC about 0.7 mb/d remains unattributed. OPEC’s assessed list puts 2.5 mb/d of additions into 2029–2030, which is consistent with more Middle Eastern, African and Asian brownfield projects starting on time than the Base Case assumes. IEA Oil 2025 (+2.38 mb/d) is below the Base Case because it predates the 2026 project news and nets off 1.6 mb/d of closures over 2025–2030.
IEA and OPEC disagree on the 2025 level by 1.3 mb/d (28.3 vs 29.6), which reflects definitions, coverage and estimation rather than real growth. The model therefore never compares raw levels: each agency’s own growth from its own 2025 value is applied to one common base (IEA 28.3 mb/d). OPEC’s 31.4 mb/d becomes 30.02 mb/d on the common base (+1.72 mb/d by 2030); IEA’s 28.3 mb/d means no growth.
| Source | Published | Own 2025 level | Own 2030 level | Growth 2025–2030 | 2030 on common base |
|---|---|---|---|---|---|
| IEA Oil 2025 path | 2025-06-17 | 28.3 | 28.3 | 0.0% | 28.30 mb/d |
| OPEC WOO 2026 path | 2026-06-18 | 29.6 | 31.4 | +6.1% | 30.02 mb/d |
OPEC publishes two different 2025 global throughput numbers: 86.89 mb/d in the Annual Statistical Bulletin and 83.5 mb/d in the World Oil Outlook. IEA crude runs are about 84.1 mb/d. The ASB gap is persistent (also in 2021 and 2024), so it is a definition and coverage difference, not a real 2025 change. The model therefore never reads an absolute utilization level as meaningful: it uses one internally consistent source (ASB capacity and throughput), derives the average diesel yield from that same throughput, and judges tightness only by the change from 2025. On the crude-run basis the 2030 change is almost identical.
| Baseline | Capacity | 2025 throughput | 2025 utilization | Implied average diesel yield | 2030 required utilization (Base, OPEC demand) | Change from 2025 |
|---|---|---|---|---|---|---|
| OPEC ASB 2026 (capacity and throughput) Internally consistent: both figures from the same OPEC statistical source. ASB throughput runs 2.5–3 mb/d above crude-run series in every year checked (81.82 in 2021, 85.72 in 2024), so it reflects a broader definition or coverage, not a 2025 anomaly. | 103.66 | 86.89 | 83.8% | 31.2% | 85.5% | +1.68 pts |
| IEA crude runs on OPEC capacity IEA 2025 crude runs implied by the September 2026 OMR (81.5 mb/d in 2026 after a 2.6 mb/d fall). OPEC WOO 2026 uses a similar 83.5 mb/d. Mixes two sources, so used only as a sensitivity. | 103.66 | 84.10 | 81.1% | 32.2% | 82.8% | +1.71 pts |
| Year | OPEC capacity | OPEC throughput | Utilization | OECD capacity |
|---|---|---|---|---|
| 2021 | 100.97 | — | — | 43.13 |
| 2022 | 101.37 | — | — | 42.87 |
| 2023 | 102.58 | — | — | 42.91 |
| 2024 | 103.62 | 85.72 | 82.7% | 43.07 |
| 2025 | 103.66 | 86.89 | 83.8% | 42.21 |
Capacity. Year-end capacity = OPEC end-2025 capacity + nameplate of units starting in each year − closures. Creep, unitemized brownfield and unannounced closures are added mid-year. Every 2025-or-earlier event is already in the OPEC base and is never counted again.
Scenarios. Each ledger entry carries its own start month in Conservative, Base Case and High; “after 2030” means it adds nothing in that scenario. Base Case dates are judged per project from progress, delivery record and financing, not by a blanket rule. The allowances also differ by scenario.
Timing. Each unit produces nothing before its start month, then ramps linearly to full rate over its ramp period. Annual values are the average of twelve monthly values, so a December start adds almost nothing to that year’s output even though year-end capacity jumps.
Diesel. New diesel = crude capacity × new-capacity utilization × refinery-type yield, or the operator’s design diesel increment × utilization where one is published. Closures lose capacity × pre-closure utilization × regional yield. Temporary disruption removes runs × regional yield. Biodiesel and renewable diesel growth comes from IEA Oil 2025. Every diesel term is a change against 2025.
Demand. Each agency’s own growth from its own 2025 level is applied to the IEA 2025 base of 28.3 mb/d. OPEC intermediate years are a geometric interpolation by StorageCurve.
Result. Average diesel yield = (2025 diesel demand − counted biofuels) ÷ 2025 throughput of the chosen baseline. Required runs = 2025 runs + new-unit runs − closed runs − disrupted runs + (demand change − supply change) ÷ that yield; required utilization divides by operable capacity. At least 2 points above 2025 is “tightening”, at least 4 points above is “tight”, and at least 2 points below is “surplus capacity building”. The thresholds are editorial. Break-even growth is the demand growth that holds 2030 utilization at the 2025 level.
Limits. Required runs assume diesel sets refinery runs and the average yield does not change; refiners can also shift yield towards diesel, and each percentage point across about 85 mb/d of runs is roughly 0.85 mb/d of diesel. Regional trade, product quality, hydrogen and logistics are not modelled. A global surplus does not guarantee supply in every region.
Updating. The ledger lives in database tables (events, settings and a change log). Every change is validated before it is saved, and the page republishes on the next scheduled update. The reviewed seed and export format is scripts/refining_ledger.json. Ledger last changed 2026-10-11 17:49:55 UTC.
About this series
Capacity coming on stream only matters if it produces diesel when demand needs it. The forecast follows each unit from start-up through its ramp, nets off closures and temporary losses, and then asks how hard existing refineries would have to run to meet the demand path you choose.
Inspect the connected refined-products view How the five-year curve is calculated
More in refined products
Related views within the same market category.
Gulf Coast crack, seasonal stocks, domestic demand and refinery output.
Inspect gasolineJet fuel crack versus Brent with stocks and four-week physical flows.
Inspect jet fuelWeekly stocks, seasonal ranges, days of supply and PADD comparisons.
Inspect US stocksMonthly closing stocks, seasonal comparisons and member-state data.
Inspect EU stocksIndependent output, demand, stocks and export samples with reporting dates.
Check diesel balanceDiesel crack, seasonal distillate stocks, demand, production, exports and refinery utilization, with an optional tightness summary.
Inspect diesel marketSource ledger
OPEC 2025 base, operator project updates, IEA and OPEC demand vintages. Ledger reviewed 2026-10-11.
Read sources and method ↓Questions, answered from the method
How the forecast is built, what it includes and how to read it.
In the Base Case, capacity starting up in 2026–2030 adds enough diesel to absorb about 0.8% a year of demand growth at 2025 refinery utilization (0.4–1.0% across the Conservative and High scenarios). The IEA Oil 2025 path, which is flat, would leave surplus capacity: required utilization falls to 80.3% in 2030. The OPEC WOO 2026 path, about 1.2% a year, needs 85.5%, against 83.8% in 2025.
The Base Case adds +3.64 mb/d of net crude distillation nameplate over 2026–2030 (2026 +0.61, 2027 +0.75, 2028 +0.29, 2029 +1.45, 2030 +0.55). Conservative adds +1.17 mb/d and High +4.72 mb/d. Indian expansions and Huajin dominate 2026–2027; Dangote’s second train makes 2029 the largest year in the Base Case.
The Base Case is a best estimate, not an inventory of decided projects: it includes likely planned projects such as Dangote’s second train and the Cauvery Basin refinery. It is 0.56 mb/d below BloombergNEF mainly because it closes more capacity and dates Hambantota, BPCL Andhra, Diwaniya and Hoima after 2030. Against OPEC, about 0.7 mb/d of brownfield additions remains unattributed. The High scenario (+4.72 mb/d) is roughly at the agencies’ level.
It is the share of operable capacity refineries would have to run to meet the selected diesel demand. OPEC publishes two 2025 throughput figures, 86.89 mb/d in its Statistical Bulletin and 83.5 mb/d in its World Oil Outlook; the gap is a definitional one that recurs every year. The model uses the Bulletin’s internally consistent capacity and throughput, derives the diesel yield from the same figure, and judges only the change from 2025, which is nearly identical on a crude-run basis.
No. It is a structural capacity forecast with stated assumptions. Prices, inventories, trade flows and product specifications decide whether a particular market is short, and a global surplus does not guarantee supply in every region.
Each unit produces nothing before its Base Case start month and then ramps linearly to full rates over its ramp period, typically 6–12 months. Annual figures are the average of twelve monthly values, so a unit starting in December raises year-end capacity but adds almost nothing to that year’s diesel.
Crude capacity × utilization at full ramp (85% by default) × a diesel yield for the refinery type: 36% fuels-oriented, 32% standard, 22% where large petrochemical units take much of the barrel. Where the operator publishes a diesel figure, such as RNEST Train 2’s 88 kb/d, that figure replaces the generic yield. A 300 kb/d petrochemical-integrated refinery therefore adds about 56 kb/d of diesel at full rates.
Residue upgraders and hydrocrackers can add diesel without any new distillation. They are separate ledger entries with zero capacity and a diesel increment, for example the Visakh residue upgrading facility. Quality conversions such as S-500 to S-10 diesel add no volume and are excluded.
Every ledger entry has its own start month in Conservative, Base Case and High. Conservative keeps only projects already under construction, with later dates; the Base Case adds planned projects judged more likely than not to start before 2031, at realistic dates; High lets operator schedules hold and adds the remaining planned pipeline. Creep, unitemized brownfield and closure allowances also differ by scenario.
Creep is debottlenecking at existing sites, 80 kb/d a year in the Base Case, matching OPEC’s 0.4 mb/d for 2026–2030. Unitemized brownfield covers expansions too small or too opaque to list one by one: 0.48 mb/d in the Base Case and 0.85 mb/d in High, against roughly 1.5 mb/d implied by BloombergNEF’s brownfield total. Run-ups are extra runs from 2024–25 start-ups still ramping in 2025.
Named closures with a date, such as Valero Benicia in 2026, plus an allowance for unannounced economic closures from 2027: 0.3 mb/d in the Base Case, 0.6 in Conservative and 0.2 in High. Closures before 2026, such as Grangemouth, Wesseling and Houston, are already in the 2025 base and are not deducted again.
As a separate temporary layer of lost refinery runs, not lost capacity. 2026 uses reported data: Middle East runs fell from 9.9 to 7.3 mb/d between February and August, and Russian runs were about 3.8–4.0 mb/d against roughly 5.2–5.4 in 2025. From 2027 the recovery path is an assumption you can change.
No. Restarting a damaged refinery restores runs that were lost; it never adds installed capacity. That is why the capacity chart is unaffected by the disruption selector while diesel supply and required utilization change.
Because roughly 3.5 mb/d of refinery runs were lost in the Gulf, Russia and crude-importing Asia, while the demand paths are structural trends that predate the war. In practice the gap was closed by record diesel prices, lower consumption and stock draws, which this model does not forecast.
Choose the persistent disruption case. It keeps 1.2 mb/d of Russian runs offline through 2030 and slows Gulf repairs. Required utilization elsewhere rises accordingly; it is a stress case, not a prediction of the war’s course.
The agencies already disagree on 2025 by 1.3 mb/d (28.3 versus 29.6), mostly because of definitions and estimation. The model applies each agency’s own growth to one common 2025 base of 28.3 mb/d, so OPEC’s 2030 figure becomes 30.0 mb/d on that base, an increase of 1.7 mb/d, while IEA’s path is flat.
Choose “Custom annual growth” and enter a rate between −3% and +5%. It compounds from the 2025 common base. In the Base Case, growth above about 0.8% a year pushes required utilization above its 2025 level; slower growth leaves more spare capacity.
Yes. Agency diesel demand includes biofuels, so IEA’s projected growth in biodiesel and renewable diesel, about 0.25 mb/d between 2025 and 2030, is added to supply. By default 90% counts as diesel; the rest is assumed to go to biojet and heating oil.
Projects, closures, allowances and disruption cases are stored in database tables with a change log. Each edit is validated against the model’s rules before it is saved, and the page republishes on the next scheduled update. The ledger was last reviewed on 2026-10-11.
Yes. The JSON download contains every ledger entry with its source and date, the allowances, disruption cases, demand presets and thresholds, so every number on the page can be reproduced.