Most engineers who ask their foundry for better surface finish get the same answer: use finer sand. That advice is not wrong, but it addresses maybe 30% of the problem. The roughest surfaces I have pulled out of flasks in 15 years were not caused by coarse sand — they were caused by a combination of factors that nobody controlled together.
Sand casting typically produces Ra 12.5-50 um, roughly 10x rougher than investment casting. But that range is enormous. The difference between Ra 12.5 and Ra 50 is the difference between a surface you can seal with a gasket and one that needs full machining. Where you land depends on at least five interacting variables, not one.
What Happens at the Mold-Metal Interface
When molten metal fills a sand mold, the liquid front conforms to every grain it contacts. Each sand particle imprints its geometry onto the casting surface — peaks, valleys, and inter-grain gaps all transfer directly.
Two additional forces compound the roughness. Metallostatic pressure — the weight of the metal column — pushes liquid into gaps between grains. On tall castings, I have measured Ra differences of 2x between the cope and drag faces with no change in sand or process. Metal penetration creates a semi-fused sand layer that is extremely difficult to clean.
Thermal degradation is the second factor. Sand binders break down the moment metal contacts the mold. As binder burns away, surface grains loosen or get carried into the casting. Steel poured at 1550 C destroys binder faster than aluminum at 700 C, which is one reason steel castings are consistently rougher from the same sand system.
Before you pour, check your mold wall hardness with a scratch test. If the binder is marginal, no amount of fine sand saves your surface.
Five Variables That Control As-Cast Roughness
I have watched foundries cut grain size in half and get worse surfaces because they ignored permeability and binder gas. Here is what actually happens inside the mold when each variable changes.
Sand Grain Size and the GFN Threshold
The AFS Grain Fineness Number (GFN) measures sand grain size. Higher GFN means finer grains and, up to a point, smoother castings.
Research from the University of Northern Iowa found something most foundries do not expect: surface improvement plateaus at specific GFN thresholds. Silica sand stops improving at 67 GFN, hitting a floor of 185 RMS. Chromite plateaus at 77 GFN (250 RMS). Olivine at 84 GFN (244 RMS). Going finer wastes money with zero finish benefit.
But there is a trap. As GFN increases, permeability drops. Lower permeability demands more binder to maintain strength. More binder produces more gas during pouring. More gas creates blowholes and pinholes — defects worse than the grain roughness you were trying to eliminate. I have watched foundries chase finer sand only to create gas porosity that sent scrap rates through the roof.
Grain Shape Matters More Than Grain Material
The same UNI research revealed a hierarchy most sand specifications ignore: grain shape outperforms grain material for finish improvement. Rounded grains produce smoother surfaces at coarser sizes than angular grains achieve when ground finer. Round grains pack tighter, leaving less space for metal penetration.
A foundry running round-grain silica at 55 GFN can match the finish of angular olivine at 80 GFN — silica costs a fraction of what olivine does. If your purchase order specifies sand material but not grain shape, you are leaving the biggest finish lever uncontrolled.
Mold Compaction
Higher compaction packs grains tighter, reducing the gap size between them. Manual green sand molds produce 500-900 RMS, while automatic molding on the same sand drops that to 250-500 RMS.
Compaction is the cheapest variable to control — no material change required, just consistent process discipline.
Refractory Coatings
Mold coatings fill gaps between surface grains with a fine refractory layer. On rough or porous surfaces — loose cores, 3D-printed molds — coatings improve finish by 48-64%. On well-compacted green sand, the improvement is smaller and sometimes negligible.
The most common mistake I see in new patterns is assuming a coating will fix a poorly compacted mold. Coatings work best when the substrate is already reasonably smooth. On a rough mold, coating thickness becomes uneven, and thick spots crack and flake into the casting.
Pouring Temperature and Alloy Selection
Higher pouring temperatures increase fluidity but also drive more penetration into sand gaps and faster binder breakdown. Steel at 1500-1600 C produces rougher surfaces than iron at 1300-1400 C or aluminum at 680-750 C from identical molds.
Specifying a pouring temperature range on your casting order — not just an alloy grade — gives the foundry a target to work toward.
Typical Ra Values by Process and Alloy
These are achievable ranges under production conditions, not laboratory bests.
| Process | Ra (um) | RMS (approx.) | ISO 8062 Grade | Best For |
|---|---|---|---|---|
| Green sand (manual) | 25-50 | 500-900 | CT12-CT13 | Low-cost, large parts |
| Green sand (automatic) | 12.5-25 | 250-500 | CT10-CT12 | Medium volumes |
| Resin-bonded (no-bake) | 6.3-12.5 | 150-600 | CT9-CT11 | Ferrous, large cores |
| Resin-bonded (fine) | 3-6.3 | 75-150 | CT9-CT10 | Precision ferrous |
| Shell mold | 1.6-3.2 | 63-125 | CT8-CT9 | High-volume, small-medium |
| V-Process (vacuum) | ~6.3 | ~125 | CT9-CT10 | Smooth finish, no binders |
| Investment casting | 1.6-6.3 | — | CT5-CT7 | Complex geometry, aerospace |
| Die casting | 0.8-3.2 | — | CT5-CT7 | High-volume non-ferrous |
| Permanent mold | 3.2-12.5 | — | CT7-CT9 | Medium-volume aluminum |
The jump from green sand to resin-bonded is the single biggest Ra improvement within sand casting — roughly 2-4x smoother — because resin systems allow finer sand with better compaction and no moisture-related defects. V-Process achieves similar results by eliminating binders entirely and holding fine sand with vacuum pressure.
For ferrous castings, automatic green sand or resin-bonded no-bake covers 80% of industrial applications. Where Ra below 6.3 um is required, you need shell molding, a process switch, or post-machining of critical surfaces.
When to Improve, When to Machine, When to Switch
The decision is about total part cost, not which process produces the best finish.
Improve the sand casting process when your current finish is worse than your sand system’s capability. Switching from manual to automatic molding, specifying a GFN target, or adding refractory coating can bring Ra 40+ down to Ra 12-20 with minimal cost increase. No tooling change required.
Machine critical surfaces when only specific areas need tight finish — bearing bores, sealing faces, mating flanges. A casting at Ra 25 overall with three CNC-machined surfaces at Ra 1.6 costs far less than switching the entire part to investment casting.
Switch processes when the part requires better than Ra 6 on most surfaces or when volumes justify die casting tooling. Investment casting makes sense for parts under roughly 50 lb where 80%+ of surface area needs Ra below 3.2. Die casting wins above 5,000-10,000 pieces where tooling cost amortizes.
The trap I see most often: engineers over-specify premium sands across the entire mold when only critical surfaces need improvement. Silica at 67 GFN reaches 185 RMS — only marginally rougher than chromite at 250 RMS — at a fraction of the cost. Spend the money on targeted coatings and machining instead of blanket material upgrades.
Getting the Finish Right on Your Next Casting
The next time you send a casting drawing to your foundry, specify three things beyond the alloy grade: a GFN target, a grain shape requirement (round or sub-angular minimum), and the surfaces where you need coating applied. Those three specifications address the variables that most purchase orders leave entirely to chance. If the foundry cannot tell you their sand GFN and grain shape distribution, that tells you something about how much control they have over your finish.