Casting Yield

Raising Casting Yield Without Buying Shrink Porosity

Every kilogram of gating and riser gets melted, poured, cut off, and melted again. Here is how to measure casting yield honestly, benchmark by process, and improve it without trading porosity for points.

Casting yield decides how many times you melt the same metal. Yield equals good casting weight shipped divided by metal poured; pour 1,000 kg of iron and ship 620 kg of good castings and you sit at 62 percent, which means 380 kg of gating, risers, and scrap goes back to the furnace. That return metal is not free: melting costs 500 to 800 kWh per ton, you lose 2 to 5 percent of it to oxidation each cycle, and it consumes furnace capacity you could sell. On a 15,000 ton per year melt shop, lifting yield 5 points cuts roughly 1,100 tons of remelt and $85,000 to $130,000 in energy alone.

Get the measurement honest before improving it. Count both weight and pieces: yield by weight equals good casting weight divided by poured weight; scrap rate by count tells you whether the loss is riser design or process defects. Worked example: a mold pours 48 kg total for four castings of 6.8 kg each, so gross yield is 27.2 divided by 48, or 56.7 percent. If one casting in ten scraps for shrink, net yield falls to 51 percent. The Casting Yield calculator handles the weight and count arithmetic per pattern; run it by pattern number, because a foundry average hides the 40 percent tool paying for everyone else.

Benchmark by process and alloy, because physics sets different ceilings. Green sand gray iron runs 55 to 75 percent; ductile iron sits a bit lower for the same geometry because it demands more feeding. Steel castings with heavy risering land 40 to 60. High pressure die cast aluminum runs 50 to 65 once biscuits, runners, and overflows are counted. Investment castings reach 75 to 90 on well designed trees. Within any process, chunky compact parts beat thin sprawling ones by 10 to 15 points. Rank every pattern against its process ceiling, and put the bottom decile on the methods engineer's desk.

The rigging levers move the most metal. Riser efficiency is the headline: a natural sand riser feeds only about 14 percent of its own volume, while an insulated sleeve pushes 30 to 35 percent and an exothermic sleeve more; converting one 12 kg natural riser to a 5 kg sleeved riser on a pattern running 20,000 molds a year keeps 140 tons out of the remelt loop. Right size gating with fluid flow rules instead of copying the last job. Add castings per mold where the plate allows: going from 4 to 6 impressions spreads the same runner system over 50 percent more product. Simulate before cutting; solidification software pays for itself on the first avoided trial loop.

Remember scrap is yield's evil twin: every scrapped casting carries its gating with it. A pattern at 60 percent gross yield and 8 percent scrap nets 55.2; cutting scrap to 3 percent recovers 2.8 net points without touching the rigging. Attack the big three: shrink, gas, and sand inclusions, which together explain 60 to 80 percent of scrap in most iron foundries. Hold pouring temperature in a 15 degree Celsius window, since cold pours make misruns and hot pours make shrink and penetration. And weigh actual pour per mold weekly; ladle overpour of 2 to 4 percent is pure yield loss nobody logs.

The failure mode that matters most: chasing yield into porosity. Shrink a riser 20 percent past what the casting's modulus requires and you save 2 kg of remelt while creating internal shrink that scraps the casting at machining, at the customer, or worse, in the field. Every riser reduction needs simulation or a sectioning trial on 10 castings before release. Other classics: counting returns as free because the metal is recycled, ignoring the 650 kWh and the melt loss; yield reported only at the foundry average; and methods changes made without telling the melt deck, whose charge makeup assumed yesterday's return ratio.

Run the cadence. Daily: poured weight versus shipped weight by line on the production board, plus scrap count by defect code from the morning shakeout review. Weekly: yield by pattern for everything that ran, with the bottom five patterns getting a methods engineer's initials and a date. Monthly: reconcile melt tons, ship tons, returns, and melt loss; if the balance does not close within 1.5 percent, fix the scales before the process. Quarterly: re-simulate the top ten volume patterns against current sleeve and filter technology, because feeding aids improve every few years while old tooling keeps pouring like it is 2009.

World class foundries hold gross yield in the top quarter of their process range, scrap under 2 to 3 percent, and pour weight variation under 1.5 percent, and they know all three numbers by pattern, today, without running a report. They rig every new job with simulation, revisit legacy patterns on a schedule, and price quotes from actual yield history instead of hope. Most of all they treat returns as expensive inventory stuck in a loop: melted at 600 kWh a ton, handled twice, oxidizing as it goes. The furnace does not care whether it melts new iron or your own risers again, but the power bill does.

Published 2026-07-02.