Foundry operations are the activities that melt metal and pour it into molds to make castings: pattern and mold making, melting, pouring, solidification, shakeout, and finishing. The work is defined by handling molten metal safely, controlling every heat's chemistry and temperature, and squeezing usable castings out of the metal poured, because a large share of it goes to gates, risers, and scrap that must be remelted.
That last fact drives foundry economics. Unlike machining, where you cut away material you paid for, casting lets you recover the metal you do not use, but only by remelting it, which costs energy and time. So a foundry lives on two numbers at once: how sound the castings are (chemistry, defects, dimensions) and how much good metal ends up in shipped parts versus back in the furnace. Get either wrong and margin evaporates.
This guide covers what a foundry actually does step by step, the main molding methods, why casting yield and scrap matter so much, the safety picture around molten metal and silica, and where digitizing heat and quality records changes the game. For the broader shaping and joining context, see metal fabrication processes.
What are foundry operations?
A foundry is a plant that produces metal castings by melting metal and pouring it into a mold cavity shaped like the finished part. Foundry operations span everything from making the mold, through melting and pouring the metal, to knocking the casting out of the mold and cleaning it up for inspection and shipment.
Foundries are usually classified by the metal they pour, gray and ductile iron, steel, aluminum, brass and bronze, and specialty alloys, and by the molding method they use. What every foundry shares is the core problem: liquid metal is hot, heavy, and unforgiving, and the quality of the casting is largely decided in the seconds while the metal fills the mold and solidifies. Everything upstream exists to make those seconds go right, and everything downstream exists to prove they did.
What are the main molding methods?
The two most common are sand casting and investment casting, and they sit at opposite ends of a cost-versus-precision trade-off. Sand casting packs a mold from bonded sand around a pattern; it is cheap, flexible, and handles large parts, but surfaces are rougher and tolerances looser. Investment casting builds a ceramic shell around a wax pattern that is then melted out; it gives fine detail and tight tolerances at higher cost per part.
Most tonnage in the world is sand cast, especially iron and steel. Investment casting earns its premium where geometry is complex or the surface and dimensional bar is high. There are others, die casting for high-volume aluminum and zinc, permanent mold, centrifugal casting, but sand and investment cover the conceptual range. The molding method dictates the mold-making step, the achievable tolerance, and how much finishing the casting needs afterward.
| Sand casting | Investment casting | |
|---|---|---|
| Mold | Bonded sand packed around a pattern | Ceramic shell around a melted-out wax pattern |
| Tolerance & surface | Looser, rougher | Tight, smooth |
| Part size | Small to very large | Small to medium |
| Cost per part | Lower | Higher |
| Typical use | Iron, steel, large or simple parts | Complex, high-precision parts |
Why do casting yield and scrap matter so much?
Because a foundry ships castings but pays to melt metal, and the two are never equal. Casting yield is the weight of good, shipped casting divided by the total weight of metal poured. The gap is gates, runners, and risers, the metal channels and reservoirs a mold needs to fill cleanly and feed shrinkage, plus any castings scrapped for defects. That gap metal is not lost, but it must be remelted, and remelting burns energy and furnace time.
So two levers move foundry margin. First, gating and risering design: enough to fill and feed the casting soundly, but no more metal tied up than necessary. Second, scrap: a casting rejected for porosity, misrun, inclusions, or dimensional error is pure loss on the labor and finishing already spent, and it drags yield down twice, once as a bad part and again as return metal. Tracking scrap by defect type and by heat is how a foundry finds the few causes behind most of its losses; see scrap rate and first pass yield.
What is the safety picture in a foundry?
Two hazards dominate: molten metal and respirable silica dust. Molten iron and steel are poured at well over 1,300°C, and any water or moisture trapped where metal is poured can flash to steam and cause an explosion, so wet-versus-dry discipline around ladles and molds is life-critical. Lockout and controlled energy procedures around furnaces, ladle handling, and heavy shakeout equipment are non-negotiable; see lockout/tagout.
The slower hazard is silica. Foundry sand is largely crystalline silica, and cleaning castings, shakeout, and abrasive blasting generate respirable dust that causes silicosis over time. OSHA tightened the permissible exposure limit for respirable crystalline silica to 50 micrograms per cubic meter as an eight-hour average, half the previous general-industry limit, with an action level of 25. The American Foundry Society and OSHA jointly publish guidance for controlling it. Dust control, ventilation, and exposure monitoring are core parts of running a foundry, not add-ons.
How do you run a foundry well?
The goal is sound castings at high yield, proven heat by heat, without losing control of safety or cost. Here is a practical operating sequence.
- Control chemistry and temperature at melt. Verify each heat's alloy chemistry and pouring temperature before it goes to the floor. A heat poured off-spec turns into scrap castings that are only discovered later, when the labor is already spent.
- Tie every casting to its heat. Record which heat each mold was poured from so a defect found at inspection can be traced back to the melt, and so a suspect heat can be contained. This is traceability in manufacturing applied to metal.
- Design gating and risering for yield, then hold it. Standardize the metal-delivery design that fills and feeds soundly with the least return metal, and stop it drifting job to job.
- Classify scrap by defect and cause. Log rejects by defect type, porosity, misrun, inclusion, dimensional, and by heat and pattern, so the few dominant causes surface instead of being averaged away.
- Attack furnace and shakeout downtime. Melting is the plant's heartbeat; a furnace or shakeout stoppage starves everything downstream. Track machine downtime so the real losses, not the loud ones, get fixed first.
- Make heat and quality records auditable. When a customer questions a casting's chemistry or a batch's soundness, the full history of the heat should be minutes of retrieval, not a hunt through pour logs.
None of this requires replacing the furnaces or the molding line. It requires connecting them so melt chemistry, pour records, scrap coding, and downtime live in one place instead of scattered across pour sheets and spreadsheets (how Harmony connects the floor). Lean thinking still applies, cut waste, standardize the work, inside the hard constraints of metal and heat; see lean manufacturing. Castings that then get machined or heat treated carry that record forward.
What do the standards and numbers say?
- OSHA's respirable crystalline silica standard sets a permissible exposure limit of 50 µg/m³ as an 8-hour time-weighted average, with an action level of 25 µg/m³ half the prior general-industry limit (OSHA).
- The American Foundry Society and OSHA developed joint guidance, Control of Silica Exposures in Foundries through an AFS/OSHA Alliance to help foundries reduce worker exposure (AFS).
- NIOSH identifies casting-cleaning operations, chipping, grinding, and abrasive blasting of sand-mold castings, as producing excessive respirable silica concentrations that require engineering controls (NIOSH).
- OSHA's general-industry silica rule requires exposure assessment, engineering controls, and, where needed, medical surveillance for exposed foundry workers (OSHA).
Where does digitization fit in a foundry?
Right at the seam between the melt deck and the paperwork. Foundries rarely lack furnace capacity or casting skill; they lose money to scrap they cannot fully explain and yield they cannot fully see, because the heat records, pour logs, and scrap tallies live in separate books. Connecting them so every casting is tied to its heat, and every reject to its defect and cause, turns tribal knowledge about "the problem heats" into data anyone can act on.
That is the honest value: not replacing metallurgical judgment, but making heat chemistry, pour history, scrap, and downtime one connected record instead of scattered sheets. It is the same pattern behind any real-time operational platform, connect what exists, capture at the source, and make the record instantly available, as one manufacturer did when it replaced paper logging with real-time capture (the CLS case study). For the systems view, see what is a manufacturing operating system and for machining of finished castings, what is CNC machining.