Reconcile foundry yield before blaming the melting process

When a foundry ERP report shows that a 1,000.0 kg furnace charge produced only 680.0 kg of accepted castings (a 68.00% charge-basis net good yield), plant leadership often assumes 320.0 kg of metal was lost in the furnace or poured into scrap. In practice, foundry yield is governed by distinct denominators: charge-basis net good yield (accepted dressed casting weight divided by total furnace metallic charge, 680.0 kg ÷ 1,000.0 kg = 68.00%), mould-poured casting yield (rough or accepted casting weight divided by liquid metal actually poured into moulds, 720.0 kg ÷ 960.0 kg = 75.00% gross box yield or 680.0 kg ÷ 960.0 kg = 70.83% net poured yield, excluding the 12.0 kg pigged ladle heel), and furnace metal recovery (total metallic weight recovered across castings, runners, risers, rejected castings, and pigged heel divided by total metallic charge, 960.0 kg ÷ 1,000.0 kg = 96.00%). Before adjusting induction furnace power profiles, slag coagulants, or charge chemistry, production and finance teams must close the physical heat balance across charge scales, pouring ladles, shakeout return totes, fettling cells, and period cutoff boundaries.

Why single-number yield reports misdirect foundry investigations

Foundries operate a closed-loop material cycle in which a substantial share of every furnace charge is intentionally solidified inside sprues, runners, gates, and feeder risers to ensure directional solidification and sound castings. Once shaken out and shot-blasted, those runners and risers return to the charge make-up bay as clean internal remelt stock.

When an ERP or shop-floor reporting configuration tracks only two transactions—raw charge issued to the melt and good castings received into finished or machined stock—three distinct operational realities collapse into one variance figure:

1. Planned gating circulation and pigged ladle heels look like process waste: In our illustrative heat HEAT-DI-2026-114, pouring 960.0 kg of liquid metal into moulds at a 75.00% gross cavity-to-poured box yield (720.0 kg rough casting weight and 240.0 kg runner-and-riser system) and pigging a 12.0 kg ladle heel yields 238.0 kg of shot-blasted gating returns (after 2.0 kg of pouring splatter) plus 12.0 kg of pigged heel—250.0 kg of circulating inventory in Return Tote RT-09, not irreversible melt loss. Mixing charge-basis net good yield (680.0 kg ÷ 1,000.0 kg = 68.00%) with mould-poured yield (720.0 kg ÷ 960.0 kg = 75.00% gross box yield; 680.0 kg ÷ 960.0 kg = 70.83% net good yield on poured weight) confuses pattern feeding geometry with furnace oxidation performance.
2. Unposted return-to-melt totes inflate apparent oxidation loss: When shakeout operators stage Return Tote RT-09 (250.0 kg) at the furnace bay at shift handover without scanning the return receipt before the daily ERP cutoff, the ledger treats the missing 250.0 kg as vanished metal.
3. Remelt double-counting distorts purchased metal consumption: If the next shift charges that 250.0 kg tote back into a subsequent heat without recording it as internal return metal—or worse, issues it under a generic raw-material code—finance either double-counts material cost or loses visibility into how much virgin pig iron, steel scrap, and ferroalloy was actually consumed.

In transformation traceability models such as the GS1 Global Traceability Standard, a transformation event must account for all input lots and all resulting output classes—primary products, recyclable by-products or internal returns, and process losses—under consistent units of measure and time boundaries. Similarly, operational performance frameworks such as OEE.com's OEE calculation methodology emphasize that quality and yield denominators must be explicitly defined rather than blended across incompatible process boundaries.

Four-bucket mass balance for foundry heat reconciliation

To prevent false alarms between metallurgy, moulding, fettling, and cost accounting, the heat reconciliation workflow must partition every metallic charge into four mutually exclusive categories at shift or heat closure:

• Bucket 1 — Accepted good castings (W_good): Net weight of castings passing visual, dimensional, and non-destructive inspection after degating and fettling, weighed on a calibrated dispatch or machining-intake scale.
• Bucket 2 — Recoverable metallic returns (W_return): Solid metallic material suitable for remelting in the same alloy grade, separated into:
  • Gating and feeder returns: sprues, runners, gates, risers, and pigged ladle heels recovered at pouring, shakeout, and degating.
  • Internal casting scrap: short-poured moulds and metallurgical or dimensional rejects (W_reject) returned to the charge yard with alloy heat identity intact.
• Bucket 3 — Measured and estimated non-recoverable process loss (W_loss): Metal permanently removed from the metallic cycle through furnace slag skimming, melt oxidation/vaporisation, pouring splatter, shot-blast scale, and fettling/grinding swarf that cannot be remelted directly.
• Bucket 4 — Unresolved cutoff or scale discrepancy (W_unresolved): The residual difference W_charge - (W_good + W_return + W_loss). Instead of forcing W_unresolved into furnace oxidation loss through an automatic plug formula, the ledger holds W_unresolved as an open investigation item until delayed return totes, pigged heel weights, or scale tares are verified.

Close the heat mass balance before escalating melt loss

Follow the workflow below, or open its complete text version.

Read the complete workflow as text

1. Record calibrated metallic charge weight by material class → Capture poured mould weight, pigged heel and slag weight
2. Capture poured mould weight, pigged heel and slag weight → Weigh degated good castings, rejected castings and gating r...
3. Weigh degated good castings, rejected castings and gating r... → Compute unresolved heat balance against shift cutoff
4. Compute unresolved heat balance against shift cutoff → Classify variance into cutoff lag, gating ratio or true mel...

Worked example: 1,000 kg ductile-iron heat reconciliation (HEAT-DI-2026-114)

Consider an illustrative medium-frequency induction furnace heat (HEAT-DI-2026-114, alloy grade SG-500/7) at an industrial valve-body foundry. All weights, yield ratios, and rupee valuations below are synthetic operational figures constructed to demonstrate mass-balance reconciliation; they are not client performance records or universal metallurgical norms.

Reported morning flash exception

At the 07:00 morning production review, the ERP flash report flags HEAT-DI-2026-114 for severe yield collapse:

• Recorded furnace charge (W_charge): 1,000.0 kg (420.0 kg pig iron, 310.0 kg steel punchings, 250.0 kg certified SG-500/7 internal foundry returns, and 20.0 kg ferroalloys/carburiser), valued at a blended metallic charge rate of Rs 84.00/kg (Rs 84,000.00 total metallic input).
• Recorded finished casting receipt at 06:00 cutoff: 680.0 kg of accepted valve bodies (34 castings × 20.0 kg net dressed weight, or 68.00% charge-basis net good yield).
• Unreconciled ERP variance: 1,000.0 kg − 680.0 kg = 320.0 kg (32.00% apparent loss on charge basis), prompting the plant manager to question whether excessive furnace superheat or wet charge scrap caused a 32.00% melt loss.

Physical floor and scale reconciliation

Before altering furnace practice, the production controller and cost accountant pull the load-cell logs from the charge crane, ladle transfer scale, shakeout return bay, and fettling cell for HEAT-DI-2026-114:

1. Furnace tap and melting loss (972.0 kg tapped): Out of the 1,000.0 kg solid charge, furnace oxidation and skimmed slag removal accounted for 28.0 kg (2.80% of charge) during melting and magnesium treatment (1,000.0 kg − 28.0 kg = 972.0 kg liquid metal tapped into pouring ladles).
2. Mould pouring vs pigged ladle heel (960.0 kg poured into moulds, 12.0 kg pigged heel): Out of the 972.0 kg tapped into the ladle, 960.0 kg was poured into 36 moulds (W_poured = 960.0 kg), consisting of 720.0 kg solidified in the casting cavities (75.00% gross cavity-to-poured box yield) and 240.0 kg solidified in the sprues, runners, and feeder risers (25.00% of poured weight). The remaining 12.0 kg of cooling metal at the end of the ladle was not poured into a mould; it was pigged into a cast-iron chill mould (12.0 kg recoverable ladle heel, kept outside the 960.0 kg mould-pour denominator). During pouring and shakeout of the 240.0 kg gating system, 2.0 kg was lost as unrecoverable pouring splatter and sprue-cup oxidation scale, leaving 238.0 kg of clean degated runners and risers.
3. Cutoff timing gap on Return Tote RT-09 (250.0 kg): Return Tote RT-09 (238.0 kg shot-blasted runners and risers plus 12.0 kg pigged ladle heel = 250.0 kg clean SG-500/7 remelt stock) was weighed at 06:14, fourteen minutes after the 06:00 automated ERP shift cutoff, so it remained in Pending_Receipt status when the morning report ran.
4. Fettling grinding loss and casting rejects (40.0 kg across rough castings): Out of the 720.0 kg rough casting weight across the 36 poured moulds, 34 sound castings yielded 680.0 kg of accepted dressed valve bodies (34 × 20.0 kg) after removing 10.0 kg of non-recoverable riser-neck and parting-line grinding swarf during fettling, while 2 castings totaling 30.0 kg (1 short-poured casting weighing 10.0 kg rejected at shakeout plus 1 full 20.0 kg casting that failed ultrasonic inspection for shrinkage porosity) were routed to the quarantine cage (NCR-FDY-409) for disposition as internal remelt scrap (680.0 kg accepted + 30.0 kg rejected + 10.0 kg grinding swarf = 720.0 kg rough casting weight; 250.0 kg tote + 30.0 kg rejected castings = 280.0 kg total recoverable metallic returns).

The table below reconciles the naive ERP flash report against the closed four-bucket physical mass balance:

Arithmetic verification and operational decision impact

1. Physical mass balance closure across stage boundaries:
   • Furnace to ladle tap: 1,000.0 kg (charge) − 28.0 kg (furnace oxidation/slag) = 972.0 kg (tapped liquid metal).
   • Ladle tap to moulds and pigged heel: 972.0 kg (tapped) = 960.0 kg (poured into 36 moulds: 720.0 kg rough castings + 240.0 kg gating/risers) + 12.0 kg (pigged ladle heel).
   • Four-bucket heat closure: 680.0 kg (good castings) + 250.0 kg (runner/riser/heel tote RT-09: 238.0 kg gating + 12.0 kg pigged heel) + 30.0 kg (rejected castings on NCR-FDY-409) + 28.0 kg (melt oxidation/slag) + 12.0 kg (2.0 kg pouring splatter + 10.0 kg grinding dust) = 1,000.0 kg (0.0 kg unresolved residual).
2. Distinguishing furnace metallic recovery, mould-poured box yield, and charge-basis good yield:
   • Total recoverable metallic weight (charge basis): (680.0 kg + 250.0 kg + 30.0 kg) ÷ 1,000.0 kg = 960.0 kg ÷ 1,000.0 kg = 96.00% (note that 960.0 kg total recovered metal equals 960.0 kg poured into moulds only because the 12.0 kg pigged heel outside the moulds numerically offsets the 12.0 kg combined splatter and grinding loss inside the pouring/fettling loop).
   • True irreversible process loss (W_loss, charge basis): (28.0 kg + 12.0 kg) ÷ 1,000.0 kg = 40.0 kg ÷ 1,000.0 kg = 4.00%, well within the foundry's illustrative configured engineering tolerance band of 3.50%–4.50% for ductile iron.
   • Mould-poured yield (poured basis, excluding 12.0 kg pigged heel): 720.0 kg rough casting weight ÷ 960.0 kg poured into moulds = 75.00% gross box yield (240.0 kg ÷ 960.0 kg = 25.00% gating ratio), and 680.0 kg accepted dressed castings ÷ 960.0 kg poured into moulds = 70.83% net good yield on poured weight.
   • Net good casting yield (charge basis): 680.0 kg accepted dressed castings ÷ 1,000.0 kg furnace charge = 68.00% (versus 710.0 kg ÷ 1,000.0 kg = 71.00% dressed casting weight before the 30.0 kg reject disposition).
3. Financial variance correction:
   • Naive reporting overstated metallic loss at 320.0 kg × Rs 84.00/kg = Rs 26,880.00.
   • Reconciled reporting credits 280.0 kg × Rs 84.00/kg = Rs 23,520.00 back to grade-segregated SG-500/7 remelt stock, isolating the true irreversible metal loss at 40.0 kg × Rs 84.00/kg = Rs 3,360.00 while flagging the 2 rejected castings (30.0 kg) for feeding-design and pouring-practice review by the methods engineer.

Role authorities, exceptions, and accounting boundaries

A heat reconciliation workflow only stays trustworthy when operators and planners cannot plug missing weight into an arbitrary loss bucket to force a shift report to close.

Scope and accounting guardrails

• No fabricated loss plugs: If a shift closes with an unexplained weight gap, the system must record the exact gap as Unresolved_Heat_Variance rather than automatically adjusting oxidation percentage or raw-material inventory.
• Preventing remelt double-counting: Internal returns (Bucket 2) must be received into a dedicated Internal_Foundry_Returns stock valuation class by alloy grade and issued back to subsequent heats from that class, never booked as zero-cost unrecorded additions or new supplier purchases.
• Process engineering authority: The reconciliation ledger identifies whether variance came from missing return tickets, casting rejects, or furnace loss; any change to furnace temperature, inoculant practice, or gating geometry remains the decision of qualified metallurgical and methods engineers.

Acceptance tests for foundry heat mass-balance controls

Before relying on ERP yield dashboards for operational or financial review, run these five observable acceptance checks in a sandbox environment using synthetic heat transactions:

Summary and next actions

Foundry yield improves only when the plant separates gating design ratio, casting scrap, cutoff timing gaps, and true furnace oxidation loss instead of reacting to a single aggregate percentage. Closing the four-bucket heat balance before closing the shift ensures metallurgists investigate real melt losses while planners fix delayed return-tote scans.

1. Map every scale point between charge make-up, ladle pouring, shakeout return totes, fettling, and NDT quarantine so each heat links inputs and outputs by Heat_ID and alloy grade.
2. Separate ERP heat reporting into four explicit buckets—accepted castings, recoverable internal returns, measured non-recoverable loss, and unresolved cutoff variance—and disable any formula that plugs residual gaps into furnace loss.
3. Require shakeout return totes and rejected castings to be weighed and linked to the originating heat window before closing the daily production balance.
4. Configure internal remelt stock by alloy family so circulating runners, risers, and scrap castings are neither lost from inventory nor double-counted as new purchases.