Refinery spreads: test actual yields behind a crack proxy
A crack spread is a benchmark product-value calculation, not a refinery’s reported profit. Actual crude quality, product yields, location, operating costs and sale timing can produce a substantially different result. Use the benchmark as a reference, then rebuild the economics of the refinery being researched.
Start with the physical conversion
The EIA describes crude-oil refining through separation, conversion and treatment into petroleum products. Refineries differ in their equipment and in the crude they can process. A simple benchmark does not establish that a plant produces its assumed proportions or sells products in the benchmark’s delivery market.
Keep all prices on compatible units. A product price per gallon cannot be subtracted directly from crude per barrel; a standard US petroleum barrel contains 42 gallons. Also match product specification and delivery terms. A quoted wholesale product value is not a retail pump price after distribution and taxes.
A hypothetical benchmark and actual yield
Assume crude costs 80 per barrel, gasoline 100 and distillate 110, all at compatible locations and dates. A simplified 3:2:1 benchmark buys three barrels of crude and values two gasoline barrels plus one distillate barrel. Gross spread per crude barrel is (2 × 100 + 110 − 3 × 80) / 3, approximately 23.33.
Now use a hypothetical refinery processing 100 barrels of the same-cost feed. Its simplified saleable yields are 48 gasoline barrels, 32 distillate and 15 other-product barrels valued at 60 each. The remaining five feed-equivalent barrels represent assumed internal use and loss in this teaching model.
| Hypothetical product | Saleable barrels | Price | Value |
|---|---|---|---|
| Gasoline | 48 | 100 | 4,800 |
| Distillate | 32 | 110 | 3,520 |
| Other products | 15 | 60 | 900 |
| Total | 95 | Mixed | 9,220 |
Against feed cost of 8,000, gross product value less feed is 1,220, or 12.20 per feed barrel. At assumed conversion cost of 5 per feed barrel, simplified contribution becomes 7.20. The benchmark’s 23.33 is therefore not the refinery’s contribution.
Respect the model’s engineering boundary
Real refinery volume yields need not behave like the teaching allocation because products have different densities and processing can create volumetric gains. Use actual disclosed yield conventions, feedstocks and units. Do not force real output barrels into a mass-balance identity based on this deliberately simplified example.
A discounted heavier crude can require different processing, hydrogen, energy or maintenance. A lower crude purchase price is not a free margin improvement. Inspect the plant’s ability to process the grade and whether the benchmark discount includes transport to its gate.
Add timing and utilization
A refinery purchases crude before products are sold. Inventory accounting and price changes can separate reported results from a same-day spread. Record the relevant feed-cost basis, sales period and disclosed inventory effects rather than attributing the whole gap to operating performance.
Review when headline crack spreads strengthen but realized refining margins weaken, when utilization drops, or when planned maintenance changes the product slate. Lower throughput can raise fixed cost per barrel, while the outage of one conversion unit can shift output toward lower-value products even if the crude distillation unit remains running.
The final worksheet should show feed grade, throughput, saleable product yields, location differentials, conversion costs, downtime and timing differences. It is a business-research model for an operator, not a petroleum trading strategy or a forecast of oil prices.
For diversified operators, separate refining from marketing and logistics before comparing a reported segment result with the model. A terminal fee, pipeline charge or retail fuel margin may offset a weaker refinery result without improving conversion economics. Conversely, transfer-pricing arrangements can shift the location of reported contribution within the group. Keep the physical plant calculation independent, then reconcile the disclosed segment boundary. A group improvement does not establish that the refinery captured a stronger product spread.
FAQ
Is a 3:2:1 spread an audited profit measure?
No. It is a defined benchmark calculation. Its yields and cost coverage differ from an actual refinery’s operations.
Can higher utilization worsen unit economics?
It can if the additional feed or product mix is unfavorable or if operating constraints increase costs. Test incremental economics rather than assuming every extra barrel earns the benchmark spread.