United States refinery closures capacity demand and fuel exports

4 minutes ago

13 min read

Research dated September 21, 2026. Annual refinery counts and capacities refer to January 1 of the named year. Completed-year operating data and 2026 forecasts are identified separately.

America has far fewer petroleum refineries than it had in the early 1980s, but the surviving fleet has preserved most of the country's crude-processing capacity. The change is substantial: the United States went from 301 operable refineries in January 1982 to 130 in January 2026. That is a net decline of 171 sites, or approximately 57%. It does not mean that exactly 171 refineries permanently closed during those years. Openings, reactivations, conversions, and changes in the surveyed population also affect the total. EIA operable-refinery history.

Comparing 254 refineries in 1982 with 130 in 2026 combines two different categories. The first number counts operating refineries; the second includes both operating and idle facilities. An accurate account must establish what is being counted before explaining why the count changed.

The evidence also corrects another familiar assertion: the United States has built refineries since the 1970s. EIA identifies a Galveston facility that started operating in 2022 as the newest refinery. Garyville, Louisiana, which began operating in 1977, occupies a narrower category: the newest refinery with significant downstream processing capacity. Both statements can be true.

Counting facilities accurately

EIA distinguishes operating capacity from idle capacity, and combines the two as operable capacity. An idle facility has not necessarily permanently exited the industry. Calendar-day capacity measures the amount that can be processed under usual conditions, accounting for interruptions and operational constraints. Stream-day capacity assumes optimal operation without an allowance for downtime. These measures should not be interchanged when comparing plants or ranking their size. EIA capacity definitions.

On a consistent operating-only basis, the comparison is 254 plants in January 1982 against 128 in January 2026, a net decline of 126. The corresponding idle counts were forty-seven and two. Calling all 130 facilities “active” obscures that distinction. EIA operating-refinery history.

The January date matters just as much. EIA's June 2026 analysis says Valero's Benicia refinery was included in the January inventory but subsequently ceased refining; its capacity was removed from monthly estimates beginning in March. The annual count therefore cannot serve as a September 2026 operating census. A current count requires a fresh reconciliation of subsequent changes, rather than casually subtracting one plant from an annual total. EIA's 2026 refinery update.

How many actually shut down

For a documented plant-level answer, EIA's Table 13 provides a permanent-shutdown register covering January 1990 through January 2026. Counting its rows produces ninety entries in the five domestic petroleum districts, plus six entries in Puerto Rico and the U.S. Virgin Islands. Three domestic entries are explicitly marked as never having operated. Removing those leaves eighty-seven listed domestic entries, including five renewable-diesel conversions. This is a defined-period register, not an all-time closure census. EIA Table 13.

The accompanying workbook preserves every entry, the source's classification notes, crude distillation capacity, downstream capacity, and separate last-operation and shutdown fields. Blank dates remain blank. A refinery that stopped processing crude in one year may have been formally classified as permanently shut in another. A facility that became a terminal or renewable-fuel plant may still employ people and handle energy products even though it no longer belongs to the conventional petroleum-refining fleet.

This creates three useful answers to three different questions: 171 fewer operable sites since 1982; ninety domestic entries in the 1990–2026 shutdown register; and 87 register entries after excluding facilities that never operated. None should be substituted for another without its definition and dates.

Fewer plants roughly comparable capacity

The long-term capacity comparison is much less dramatic than the plant-count decline. Operable atmospheric crude distillation capacity was 17,889,734 barrels per calendar day in January 1982 and 18,160,493 in January 2026. The increase was 270,759 b/cd, or 1.51%. Dividing each capacity by its matching operable count gives approximately 59,434 b/cd per site in 1982 and 139,696 in 2026. Average site capacity grew by about 135%. These are calculations from EIA's capacity and count series. EIA capacity history.

That does not prove that refinery efficiency doubled. Average capacity measures facility scale. Energy efficiency, fuel yield, operating cost, reliability, and emissions per barrel require different evidence. A larger plant may have important operational advantages, but dividing national capacity by the number of plants does not measure those advantages directly.

Nor has capacity followed a straight upward path. January 2020 capacity reached 18,976,085 b/cd. January 2026 was 815,592 b/cd lower, a decline of about 4.30%. The country therefore retained slightly more capacity than in 1982 while operating below its pre-pandemic January peak. Both comparisons belong in the article; choosing only one would hide an important part of the trend.

The recent contraction

EIA reports 135 operable refineries in January 2020, 129 in 2021, 130 in 2022, 129 in 2023, 132 in 2024, 132 in 2025 and 130 in 2026. Those changes describe the net population. A six-site decline does not, by itself, document six permanent shutdown events. A February 2022 startup cannot explain a change already recorded in January 2022. EIA operable-refinery history.

For 2025, EIA's annual event table identifies Houston Refining and Phillips 66's Los Angeles refinery. Their listed capacities were 263,776 and 138,700 b/cd, totaling 402,476. The table reports February and October 2025 as their last-operation months. Houston's last-operation date should be distinguished from later completion of its business exit. EIA Table 11.

National capacity declined by only 263,000 b/cd between January 2025 and January 2026, from 18,423,493 to 18,160,493. Subtracting that net decline from the two plants' combined capacity implies 139,476 b/cd of offsetting net capacity changes elsewhere. This residual is an accounting calculation, not proof of one identified construction project. EIA annual capacity table.

Why plants close

The early consolidation had a particular economic setting. A 1986 Government Accountability Office report explained that federal oil-market rules had encouraged small, relatively simple refineries during the 1970s. Decontrol compounded difficulties for many refiners. GAO reported that 90 of 106 net refinery shutdowns between January 1981 and December 1985 involved plants below 30,000 barrels per day. That period and definition differ from the later EIA register, so the totals should not be added. GAO's historical refinery study.

Recent decisions also need plant-specific explanations. LyondellBasell's 2025 annual-report letter describes its refining exit as reducing earnings volatility, avoiding costly maintenance investments, and lowering its emissions footprint. Those are the company's stated reasons, rather than an independent estimate of how much each factor contributed. LyondellBasell's annual-report letter.

For Los Angeles, EIA reported that Phillips 66's chief executive characterized the closure decision as a response to an increasingly challenging expected business environment, rather than an immediate response to a particular new California policy. That distinction does not establish that regulation had no influence. It does prevent a simple chronological coincidence from being treated as proof that one law caused the closure. EIA's California analysis.

The analytical implication is that maintenance needs, expected margins, feedstock access, product demand, environmental obligations, and alternative uses for capital must be assessed together. These sources do not establish the percentage of national closures attributable to any single administration, regulation, or corporate motive. A causal investigation would need contemporaneous company filings and investment decisions for each plant.

What the last new refinery actually means

The EIA FAQ names Texas International Terminals' Galveston plant, built in 2021, and first operated in February 2022, with 45,000 b/cd of capacity. Its list also includes later-era facilities in Kern, Channelview, Corpus Christi, Houston, and Galena Park. The workbook reproduces all nineteen rows, including original and 2026 capacities and ownership as printed in the source.

Garyville illustrates expansion at an established site. The FAQ reports an initial capacity of 200,000 b/cd when it began operating in 1977 and 617,000 b/cd in January 2026. It also identifies Beaumont's 2023 upgrade and Port Arthur's 2012 upgrade, with January 2026 capacities of 612,000 and 656,400 b/cd, respectively. Those figures describe total site capacity, not the amount added by each project. EIA refinery FAQ.

A refinery's original opening date consequently says little by itself about the age or capability of its current equipment. Investment can add processing units and remove bottlenecks inside an existing boundary without creating another refinery in the national count. Conversely, a small new distillation facility cannot automatically replace the range of fuel-making capabilities at a large integrated plant.

Capacity crude throughput and finished fuel are different measurements

The best completed-year comparison available in the retrieved historical series is 1982 against 2025. Refinery and blender net input of crude oil rose from 11.774 million barrels per day to 16.369 million, an increase of 4.595 million b/d, or approximately 39%. This is actual crude throughput. It should not be labeled gasoline production or total finished product output. EIA crude-input history.

Dividing annual crude input by January capacity does not establish EIA's official refinery utilization rate. EIA defines refinery utilization using gross input to atmospheric distillation units relative to operable capacity. Numerators, time periods, and capacity bases must match. EIA utilization definitions.

Refining also changes the mix and volume of products. Distillation separates crude into fractions; additional processing changes those fractions into more useful products. Processing gain can increase measured liquid volume because the resulting products have different densities. It does not create energy from nothing. EIA's explanation of refining.

How demand changed

EIA's historical product-supplied series puts total U.S. petroleum demand at 15.296 million b/d in 1982 and 20.736 million in 2025. The increase was 5.440 million b/d, or approximately 35.6%. The same series records a lower 15.231 million b/d in 1983. EIA product-supplied history.

The composition matters. Finished motor gasoline demand increased from 6.539 million b/d in 1982 to 8.935 million in 2025. Distillate demand increased from 2.671 million to 3.908 million b/d. Distillate includes more than highway diesel, so describing the entire category as trucking demand would overstate what these data establish. Gasoline history, distillate history.

For jet fuel, the retrieved kerosene-type series reports 0.804 million b/d in 1982 and 1.737 million in 2025. Comparisons with broader jet-fuel totals must account for the treatment of naphtha-type fuel. The workbook labels the kerosene-type historical series explicitly. EIA kerosene-type jet-fuel history.

Keeping the 2026 forecast separate

EIA's September 2026 outlook forecasts 16.61 million b/d of crude refinery input and 20.58 million b/d of petroleum consumption for the full year. Its product forecasts are 8.76 million b/d for gasoline, 3.83 million for distillate and 1.72 million for jet fuel. Domestic crude production is forecast at 13.83 million b/d. These are forecasts, not completed-year observations. The retrieved STEO also uses a different 2025 consumption figure from the historical series, so the workbook preserves the source vintages rather than silently forcing agreement. September 2026 STEO Table 4a.

Demand minus crude input is not an automatic import requirement or a measure of missing refinery capacity. A petroleum balance also includes natural-gas liquids, renewable fuels, processing gain, net trade, inventory movements, transfers, and statistical adjustments. Some natural-gas liquids serve petrochemical or heating markets directly. They are not all blended into gasoline. The balance must be reconciled product by product before drawing conclusions about self-sufficiency.

The supplyversusdemand comparison what the fourmillionbarrel gap measures

The contrast between earlier capacity and today's demand deserves an explicit comparison. However, the verified figures do not show that refineries produced four million barrels per day more than the country needed in 1982. They show the following, in millions of barrels per day:

Measure	1982	2025 actuals	2026 forecast / January capacity
Operable crude distillation capacity, January 1	17.890	18.423	18.160
Actual crude input to refineries and blenders; 2026 forecast	11.774	16.369	16.610
Total petroleum demand; 2026 forecast	15.296	20.736	20.580
January capacity minus full-year demand	+2.594	-2.313	-2.420
Crude input minus total petroleum demand	-3.522	-4.367	-3.970

Calculations use EIA capacity history, crude-input history, petroleum-demand history, and September 2026 forecasts. Capacity is a January snapshot; input and demand are annual daily averages. The last two rows compare different physical measures and are descriptive differences, not complete supply balances.

The roughly four-million-barrel figure is supported as the difference between crude refinery input and total petroleum demand: 4.367 million b/d in 2025 and a forecast 3.970 million b/d in 2026. It cannot be called a four-million-barrel shortage of finished fuel. Petroleum supply also includes natural-gas liquids, biofuels, and processing gain, alongside imports, exports, and inventory changes.

For 1982, January crude distillation capacity exceeded annual petroleum demand by about 2.594 million b/d, but actual crude throughput was below demand. Capacity was potential processing ability, not barrels actually produced. Comparing that earlier capacity with today's actual throughput would create an apparent surplus-to-deficit reversal by changing the measure halfway through.

A supported historical point remains: comparing 1982 with 2025, total petroleum demand grew by 5.440 million b/d, while actual crude refinery input grew by 4.595 million b/d. Their numerical gap widened by 0.845 million b/d. That is a reason to examine the other supply streams and trade balance, rather than assume that the entire difference represents unmet domestic needs. The claim of a four-million-barrel production surplus in 1982 would require a different, identified production series to substantiate it.

How much fuel is exported and what would fifty mean

Exports belong in the refinery story because U.S. production serves overseas buyers as well as domestic consumers. The frequently cited 30% figure needs a precise label: EIA reported that U.S. exports in 2024 equaled about 30% of domestic primary energy production, measured by energy content. Its rounded figures were thirty-one quadrillion British thermal units exported against 103 quadrillion produced. This covers the overall energy system, including crude oil, petroleum products, natural gas, and coal. It does not establish that 30% of gasoline, diesel or all refinery output was exported. EIA's explanation of the 30% figure.

For the latest completed year in the retrieved petroleum series, 2025, gross exports of crude oil and petroleum products averaged 10.785 million barrels per day. Crude oil accounted for 3.961 million, leaving approximately 6.824 million b/d of petroleum products. That broader product amount included 3.079 million b/d of hydrocarbon gas liquids; it should not all be described as transportation fuel produced by refineries. EIA's separately defined finished-petroleum-products category totaled 3.050 million b/d and includes products beyond motor fuels. EIA annual export data.

The actual exported product breakdown

The 2025 annual-average breakdown below uses EIA's finished-petroleum-products category. The five liquid-fuel categories together accounted for 2,359,000 barrels per day. Petroleum coke and the other products are listed separately to show how that fuel subtotal relates to the broader reported total.

Exported product	Barrels per day	Share of reported finished-product exports
Distillate fuel oil, including diesel and heating oil	1,250,000	41.0%
Finished motor gasoline	778,000	25.5%
Kerosene-type jet fuel	216,000	7.1%
Residual fuel oil	113,000	3.7%
Kerosene, separately reported	2,000	0.1%
Five liquid-fuel categories subtotal	2,359,000	77.3%
Petroleum coke	545,000	17.9%
Lubricants	112,000	3.7%
Asphalt and road oil	19,000	0.6%
Other oils for petrochemical feedstock	10,000	0.3%
Waxes	4,000	0.1%
Miscellaneous products	2,000	0.1%
EIA reported finished-products total	3,050,000	100.0%

The subtotal is included in the total, not additional to it. Individually rounded components sum to 3,051,000 b/d, which is 1,000 above the reported total. Petroleum coke is solid; its reported barrel-equivalent volume should not be described as gallons of liquid fuel. Percentages in this table describe the export mix, not the percentage of U.S. production exported. EIA 2025 annual export breakdown.

Within distillate exports, 1,081,000 b/d contained no more than fifteen parts per million sulfur, 97,000 b/d contained over 15 through 500 ppm, and 72,000 b/d contained over 500 ppm. These are components of the 1,250,000 b/d distillate total, rather than additional exports. The low-sulfur category represented approximately 86.5% of distillate exports. EIA distillate export categories.

The workbook also itemizes categories outside that finished products total: hydrocarbon gas liquids, including propane; biofuels, including ethanol and renewable diesel; unfinished oils; and gasoline blending components. These remain separate because their production sources and processing stages differ. Crude oil is excluded from the refined-product breakdown.

For comparison with U.S. refinery output, the major transportation-fuel ratios are:

Product, 2025 annual average	Gross exports, million b/d	Refinery and blender net output, million b/d	Exports divided by net output
Finished motor gasoline	0.778	9.462	8.2%
Distillate fuel oil	1.250	4.993	25.0%
Kerosene-type jet fuel	0.216	1.825	11.8%

These calculated ratios compare matching product categories and years. They are gross export volumes relative to net refinery-and-blender output, rather than a tracing of which domestically produced barrels were shipped abroad. Imports and inventory changes remain separate parts of each product's balance. Distillate includes diesel and other distillate uses. EIA exports, EIA refinery and blender production.

The person commissioning this research reports that, in a conversation with them and Ben, a Montana congressman said exports were being discussed rising from 30% to 50%. This is an attributed account of a private conversation. The congressman's name, the meeting date, the fuel or energy category, the denominator and the proposed timetable have not been supplied. The research has not independently identified a public proposal establishing that target. The account therefore merits follow-up, but cannot yet be presented as an announced national policy to export half of America's refined fuel.

If both percentages refer to the same measure, a rise from 30% to 50% is an increase of twenty percentage points. At unchanged production, it would mean 66.7% more export volume. That arithmetic is conditional, not a forecast. It also does not establish the effect on domestic fuel availability: production, imports, inventories, and the product mix could change. An increase by 50% would be different again: a 30% starting share would become 45% at fixed production, rather than 50%.

The question for further reporting is concrete: which products would be exported, from which facilities, under what proposal, and with what assumptions about domestic demand and replacement supply? The concern about exports competing with domestic buyers deserves analysis using those details. Combining an all-energy percentage with refinery closures would conceal the very supply pressures the investigation is trying to explain.

Why location can matter more than the national total

A closure's practical effect depends on where replacement fuel can come from. EIA estimates that the Los Angeles closure removed roughly 5% of West Coast capacity, while Houston's larger closure represented about 3% of Gulf Coast capacity. The West Coast has limited pipeline connections to the Gulf's large refining system. National capacity can therefore appear adequate while one regional market becomes more exposed to disruption. EIA's regional closure analysis.

Fuel specifications and processing capabilities add constraints. Refineries cannot instantly switch their entire output toward whichever fuel is scarce. EIA notes that even substantial differences in gasoline and diesel margins have produced only modest changes in refinery yields. The agency also identifies the West Coast's limited alternative supplies as a reason outages can produce pronounced price swings. EIA on production and petroleum-product prices.

This does not mean every closure produces an inevitable or permanent price increase. Inventories, imports, demand, transport capacity, and remaining plants' available output all influence the response. A credible price claim needs an identified market and period, actual fuel prices, and a method for separating the closure's effect from crude prices and other disruptions. The refinery count alone cannot do that work.

The evidence supports a clear account: a smaller fleet of larger facilities has retained considerable processing capability, while recent closures have reduced capacity from the 2020 peak and created particular regional concerns. The useful questions are how much usable capacity remains, which products it can make, where those products can be delivered, and how quickly the system can respond when another plant goes offline.

SOURCES

1. U.S. Energy Information Administration (EIA) — U.S. operable refinery count history

https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=8_NA_8O0_NUS_C&f=A

2. EIA — U.S. operating refinery count history

https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=8_NA_8OO_NUS_C&f=A

3. EIA — U.S. operable crude oil distillation capacity history

https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=8_NA_8D0_NUS_4&f=A

4. EIA — Refinery capacity definitions

https://www.eia.gov/dnav/pet/TblDefs/pet_pnp_cap1_tbldef2.asp

5. EIA — Refinery Capacity Report, Table 13: Permanently shut down refineries

https://www.eia.gov/petroleum/refinerycapacity/table13.pdf

6. EIA — Refinery Capacity Report, Table 11: Refinery changes