Aluminium smelters: phase potline restarts into output
An aluminium restart announcement can describe several different milestones: funding approval, equipment preparation, energised pots, stable production or a fully restored annual run rate. Only some of these support actual output in the reporting period. Research should convert the disclosed ramp into a time-weighted production schedule before attaching an earnings assumption.
The deliverable is a quarter-by-quarter capacity bridge with an inventory reconciliation. It separates productive availability, calendar time, ownership and shipments. It is not a technical restart procedure. The worksheet uses assumed production outcomes, without advising how a potline should be energised or operated.
Locate actual production and shipment evidence
Alcoa’s fourth-quarter and full-year 2025 results release discusses aluminium production under both the quarterly and full-year Production paragraphs. It attributes increases in part to restarts at named facilities. The full-year Shipments paragraph describes a different shipment movement and the effect of an offtake arrangement ending. These separate passages demonstrate why production and shipments need distinct evidence.
Use the release to identify the reported period and the mechanism management describes. It does not establish the fictional pot count or ramp below. Obtain facility disclosures, production tables and update dates to determine whether a milestone is planned or achieved. Where only a restart target is public, future output belongs in a scenario, rather than being entered as recorded production.
For each site, record gross nameplate capacity, the productive fraction during each period, actual reported output where available and the ownership basis. Keep acquired production, external metal and contract offtake separate from production by the company’s own active pots. They can affect sales without changing the same physical capacity denominator.
Build a time-weighted main case
Assume a fictional facility has one hundred pots. For teaching purposes, each fully productive pot contributes one thousand tonnes per full year. Total fully productive annual capacity is therefore 100,000 tonnes. Assume four equal-length quarters; a real year requires actual days or appropriately weighted reporting periods.
The average productive pot counts within the quarters are twenty, forty, seventy and one hundred. These are averages of productive availability within each quarter, not simply the counts photographed on each quarter’s final day. Under the stated constant-output assumption, quarterly output equals average productive pots × 1,000 × one quarter.
| Quarter | Average productive pots | Assumed output, tonnes |
|---|---|---|
| First | 20 | 5,000 |
| Second | 40 | 10,000 |
| Third | 70 | 17,500 |
| Fourth | 100 | 25,000 |
| Year | Time-weighted average 57.5 | 57,500 |
The annual sum is 57,500 tonnes. At year-end, one hundred productive pots imply a 100,000-tonne annual run rate under the same assumptions. The fourth quarter’s 25,000 tonnes multiplied by four also equals 100,000, but describes an annualised fourth-quarter pace, not the volume produced during the completed year.
Entering 100,000 as actual full-year restart production would overstate this example by 42,500 tonnes. Even a year-end milestone achieved exactly as planned does not retroactively create output in the earlier ramp periods. Use the schedule to identify when capacity becomes productive, rather than crediting every restarted pot with a full year’s output.
Handle ownership on a separate basis
Assume the investor owns 75% of the fictional facility. Proportionate output is 57,500 × 75% = 43,125 tonnes. Gross site output remains 57,500. If the financial statements consolidate the operation, the production reporting convention may not match a simple proportionate calculation. Label the 43,125 as an ownership-based analytical measure unless the issuer defines it as its reported production basis.
Do not multiply gross output by the ownership percentage twice when a table already reports equity tonnes. Likewise, attributable capacity is not the same as a contractual right to buy third-party metal. Compare the heading and footnote before adding quantities across facilities. A total combining gross tonnes at one site and equity tonnes at another has no consistent denominator.
Reconcile output with shipments
Suppose beginning finished-metal inventory is five thousand tonnes. Own production is 57,500, externally purchased metal is ten thousand and ending finished inventory is 7,500. Assume no other movement. The volume available for shipment is 5,000 + 57,500 + 10,000 − 7,500 = 65,000 tonnes.
This simplified shipment bridge exceeds own production because external purchases and inventory movements matter. Calling all 65,000 tonnes restart output would overstate the physical contribution of the ramp. Conversely, a facility could produce more while shipment growth remains weak if it builds inventory or loses an external supply arrangement.
Revenue needs additional evidence: shipment dates, customer acceptance, product mix and realised pricing. Restart tonnes are not automatically sold tonnes, and sold tonnes are not automatically cash collected. Keep the bridge at the physical volume level until the financial recognition and price inputs are known. Do not convert the arithmetic into an earnings forecast by assuming all new output receives one unchanged margin.
Test a different boundary: quarter-end counts
Suppose the second quarter ends with forty productive pots, but only twenty are productive on average because the additions arrive late. Under the equal-quarter teaching convention, output would be five thousand tonnes rather than the ten thousand in the main case. The final-day count is unchanged; the time-weighted contribution differs.
This is why a release announcing forty pots energised at quarter-end does not establish forty productive pots for the entire quarter. Energised status may also precede stable output. A dated count and a measured production series are different evidence objects. If the source does not disclose the ramp path, show an early-versus-late activation range rather than invent a smooth monthly schedule.
Another boundary is a downstream bottleneck. Productive smelting availability alone does not prove equal saleable shipments when casting, transport or customer qualification constrains the flow. This worksheet does not solve those operating systems; it requires disclosure of the additional constraint before treating nameplate output as delivered product.
Apply the verification sheet
| Input | Evidence | Check |
|---|---|---|
| Restart status | Dated facility announcements | Planned, energised or stably productive |
| Capacity | Issuer’s site table and definition | Gross versus equity, annual basis |
| Timing | Period averages or actual production | End-date count not used as full-period average |
| Output | Site or segment production data | Comparable product and reporting period |
| Shipments | Inventory and external-metal bridge | No purchased metal counted as own output |
| Financial use | Price and recognition records | Physical ramp not relabelled as profit |
When a site announcement and an annual report appear inconsistent, check publication dates, period coverage and whether the later document revised a target. Preserve both versions and identify the exact change. Do not average planned and realised production or overwrite a missed target so that the forecast history disappears.
If quarterly average productive counts are unavailable, use actual production where disclosed. If neither exists, state a conditional range and list the missing timing evidence. A precise annualised capacity number cannot remove uncertainty about the path taken to reach it.
Update when restart dates, stable output, ownership, external supply, inventory or downstream constraints change. The conclusion should reconcile the annual output achieved with the exit pace, identify the reported ownership basis and leave shipment or earnings effects conditional where supporting evidence is absent.
Preserve calendar, product and ownership boundaries
The equal-quarter convention is a teaching shortcut. In an issuer model, weight productive capacity by actual days or use reported production. A leap year, an irregular reporting period or a late-quarter milestone can make a simple one-quarter multiplier unsuitable. Store the period start and end beside each row so the time weight can be reproduced.
Ownership also has a timeline. If the investor increases its stake during the year, applying the year-end percentage to all prior production can misstate proportionate exposure. Separate gross site output from dated ownership assumptions before calculating an analytical equity series. A financial consolidation change needs its own reporting explanation.
Match inventory at one product stage. Liquid metal, cast products and fabricated shipments may not share an unchanged weight basis after conversion or loss. If the bridge crosses stages, obtain the relevant conversion and movement evidence. Otherwise reconcile only the comparable stage and label the rest unknown. These boundaries prevent a physical restart story from absorbing acquisition effects, downstream conversion or purchased metal that belongs in another part of the operating account.
FAQ
Is a 100,000-tonne year-end run rate the year’s production?
No. Production depends on productive capacity available throughout the year.
Can shipments exceed own production without a capacity increase?
Yes. External purchases and inventory reductions can supply additional shipments.