When a new casting job lands on my bench, the first two questions I ask are always the same: how heavy is the part, and how many do you need per year? Those two numbers eliminate one process or the other about 80% of the time. Shell molding (coated sand) and no-bake (resin sand) both use resin binders, but they occupy different corners of the casting world. One process tops out around 25 kg. The other has produced single castings exceeding 120,000 pounds. Tolerance and surface finish matter, but they rarely override the weight and volume answer. Get part weight and annual volume right, and the process practically selects itself.
Weight and Size Limits
Shell molding has a hard ceiling on part weight that no amount of engineering can work around. Multiple independent sources place the limit between 14 and 30 kg, depending on foundry capability and alloy density. I use 25 kg as my working threshold — parts heavier than that simply don’t belong in a shell mold.
No-bake has no practical upper weight limit. Hodge Foundry in Pennsylvania routinely pours no-bake iron castings exceeding 120,000 pounds using hand-made styrene patterns with heavy venting. I’ve personally run no-bake molds for parts ranging from a couple of kilograms to several hundred. The process doesn’t care.
If your part weighs more than 25 kg, stop comparing — you need no-bake or green sand. Shell molding isn’t an option. If your part weighs under 10 kg and you’re running thousands per year, shell molding is the default. The gray zone between 10 and 25 kg is where the other criteria start to matter.
Tolerance and Surface Finish
Dimensional Tolerance
Shell molding delivers roughly twice the precision of no-bake across the parting line. Typical shell tolerances land at +/-0.030 inch, compared to +/-0.060 inch for no-bake molds. In ISO 8062 terms, shell molding achieves CT7-CT8 grades while no-bake falls in the CT9-CT12 range.
Draft angles tell a similar story. Shell molds need only 1 degree of draft — sometimes zero on short vertical walls. No-bake patterns typically require 2-3 degrees. Reduced draft means less machining allowance, which can offset shell molding’s higher tooling cost on precision-critical parts.
Before you pour a trial run, check whether your tolerance requirements actually demand shell-level precision. For most structural and mechanical castings, CT9-CT10 is perfectly acceptable, and specifying tighter tolerances than necessary just inflates cost.
Surface Finish
Standard no-bake castings come off the line at roughly Ra 6.3 um. Fine-grain no-bake can push that down to Ra 3-6 um. Shell molding sits at Ra 3.2 um typical, with optimized shell reaching Ra 1.6 um.
The gap between optimized no-bake (Ra 3 um) and typical shell (Ra 3.2 um) is negligible. The shell molding advantage only becomes meaningful when you compare optimized shell (Ra 1.6 um) against standard no-bake (Ra 6.3 um). For 80% of industrial castings that get machined on critical surfaces anyway, no-bake finish is good enough.
Tooling Cost and Volume Economics
The cost gap between processes lives almost entirely in the tooling. No-bake patterns can be made from wood, resin, or epoxy, with tooling costs typically running $500 to $10,000. Shell molding requires heated metal patterns — and those start at $10,000 and go up fast. That’s a 10-20x difference before you pour a single casting.
Per-mold material cost for shell runs about $0.25-$0.30 per kilogram of sand mix, and no-bake mold cost roughly doubles green sand on a per-mold basis. But the real economic question is amortization. At 50 parts per year, that $15,000 metal pattern adds $300 to every casting. At 5,000 parts per year, it adds $3. Shell molding becomes cost-efficient only at medium-to-high volumes where tooling amortizes across enough parts to offset the upfront investment.
The most common mistake I see in new patterns is specifying shell molding for low-volume work because the surface finish looks better on paper. Pattern quality determines casting quality, and a well-made no-bake resin sand mold with proper gating design will outperform a poorly designed shell mold every time. Don’t let surface finish specs distract you from the economics.
The 4-Question Process Selection Shortcut
After building tooling for both processes across hundreds of jobs, I’ve boiled the decision down to four questions. Answer them in order — each one either selects a process or moves you to the next question.
Question 1: Does the part weigh more than 25 kg? Yes — use no-bake. Shell molding cannot handle it. No further questions needed.
Question 2: Are you making fewer than 500 parts per year? Yes — use no-bake. Shell molding tooling won’t amortize at this volume. The per-part cost premium on metal patterns will eat any quality advantage.
Question 3: Do you need better than CT9 tolerance or Ra 6 um finish as-cast? Yes — use shell molding. You’ve cleared the weight and volume hurdles, and the precision justifies the tooling investment. No — use no-bake. You don’t need shell-level precision, so why pay for it?
Question 4: Is this a complex internal passage requiring a precision core inside a large mold? Consider a hybrid approach — no-bake mold body with coated sand cores. This combination lets you get shell-quality internal features inside a casting that exceeds shell molding’s weight limit. It’s common practice on foundry floors.
One final note: both processes handle the same alloy families — gray iron, ductile iron, carbon steel, stainless, aluminum, and copper alloys. Alloy type is almost never the deciding factor. Weight and volume are.
Common Specification Mistakes
The biggest error I see isn’t picking the wrong process — it’s obsessing over the process while ignoring the gating design. Quality depends more on rigging than on the molding process itself. I’ve seen beautiful shell molds produce scrap because the gating couldn’t feed the casting properly, and I’ve seen rough-looking no-bake molds turn out dimensionally perfect parts because the risering was right.
Three specification mistakes to avoid:
- Specifying shell molding for parts over 25 kg because a supplier quoted it. They’ll either reject the job or produce a casting at the edge of what the process can handle. Neither outcome is good for you.
- Ignoring tooling amortization. A $15,000 pattern makes sense at 5,000 parts per year. At 50 parts, you’ve just added more to tooling than the castings themselves cost. Always calculate your per-part tooling burden before committing.
- Over-specifying surface finish. If the casting face gets machined after shakeout, paying a premium for Ra 1.6 um as-cast finish is waste. Specify the as-cast finish you actually need, not the best one available.
Before you pour, check those two numbers: part weight and annual volume. Get those right, and the rest of the specification falls into place.